I thought the MQA video was good then I watched this one! Bloody brilliant, science based explanation of how electrical interference has nothing to do with snake oil. As a direct result of this vid bought a usb galvanic isolator (the intona 7054), 2 decent usb cables (also sorted power supply end out too) and my god my setup has been improved to a new level! I will for sure sort patreon support! Great work
Dear Goldensound, In this world where pure gold and sneak oil (EDIT : « snake ») are sometimes so difficult to tell apart, your reviews are really in a league on their own, thanks to your deep understanding, objective measurements, and explanation skills. Very helpful. Kudos ! And thank you so much.
Wow, really enjoying this channel and the clear technical explanations. I'm glad to find out that my DAC has PLL for all SPDIF inputs, I was about to go researching a Pi2 AES!
I appreciate the actual graphs that prove what you're saying. I'm new to high-end audio, and for the most part I assume that the things that people tell me, like "this is better than this because this one costs 10x as much and it makes me feel good" is mostly subjective, but you've supplied receipts which actually puts weight behind your words.
Also: NONE of these people commenting here, nor the uploader has ever heard jitter in their lives. They do not conduct proper blind testing but just conclude things. Measurements do not lie but most people don't understand the decibel scale and the number in it. If it is below -100dB: do not bother, you will never be able to detect that in any real life scenario, yo can only catch it with measurements. In most cases we should not care what happens below -70dB but it is better to be safe and aim for a bit better than Redbook, 96dB dynamic range.
6:40 for what it's worth an OCXO is still voltage-tuned (EFC) with a typical pulling range of something like 1-2e-7. The ovenized part is for short-term stability, not frequency adjustment. Essentially, quartz crystals have a characteristic curve of frequency change vs temperature change over temperature, which is very different for different "cuts" (the literal way the crystal is cut and polished). The standard is the AT cut, which is pretty much all quartz crystals you see everywhere. The AT cut crosses 0 Delta-Hz/Delta-K at around 25 °C. Which is just about perfect for normal ambient room temperatures. But the slope around that point is steep (relatively speaking, so let's say you're at 20 °C, you might be getting 5 ppm/K or so). Now the clever trick about OCXOs is that they use different cuts, often the SC cut. The SC cut has a completely different response to temperature changes; it crosses 0 at around 90 °C or so, and has a very gentle slope around this area. So OCXOs keep their crystal heated to a very _stable_ 85-95 °C, using a control loop. The attenuation of ambient temperature changes by the control loop colludes with the shallow slope of the dHz/dT to give extremely low tempcos. The _expense_ of that is: SC cut crystals are much more expensive (tiny tiny tiny fraction of the market), OCXOs are a lot of hassle to build and even fast-stabilizing types will exhibit a pretty wild swing from turning on to becoming mildly stable (at least 15+ minutes), they have very high power consumption compared to normal oscillators (couple Watts steady-state vs. a few mW for a normal oscillator and maybe 50-100 mW for something fancy. (And some OCXOs tend to fail under repeated power cycling, like you'd do in a residential setting -- they're _absolutely not_ designed for that and _will not_ hold their specs when treated like that). Most (all?) OCXOs have their own reference voltage which is typically ovenized as well. Sort of like bootstrapping, the reference has a tempco (selected for ~0 tempco at the elevated operating temperature, another reason these are expensive), so it would impact the temperature controller, so it's ovenized itself. It's of course stable, the tempcos of precision references are tiny to begin with. Now the reason people went to this trouble is that for, mostly physical experiments, extremely accurate and stable clocks are a useful tool. A good OCXO achieves short-term stability better than 1e-12, that's one part per billion, or for the typical 10 MHz reference, that's less than 0.00001 Hz variation over a few seconds. (Jitter and stability are not the same thing). These are mechanically vibrating devices: Vibration creates low-level spurs. Most physical phenomena influence oscillator output, temperature is just a big one. Gravity is another; angling or rotating a quartz changes the frequency (much more than 1e-12). So for optimal performance you'd want to dampen vibrations of all frequencies. Sounds like something you could sell to audiophiles. Big fat granite rock on a bunch of thick steel rope springs. Obviously, as you allude to at the end, the oscillator can't actually be used as a reference for anything (what's 10 MHz / 44.1 or 48 kHz?), so you're looking at a fractional-N PLL anyway, and at that point the quality of the output clock is liable to be swamped by the PLL's qualities (or lack thereof). To be frank, a normal quartz is probably just dandy for audio, if the rest of the system design holds up. Audiophility seems to be especially liable to buzzword engineering instead of system-level engineering.
OCXOs are fine and dandy, but frequency stability per se doesn't matter in audio, phase noise matters - OCXOs are for long term stability, phase noise is pulse-to-pulse stability, close but not related. And even then it doesn't matter - a lowly jelly bean quartz crystal is like 4 or more orders of magnitude better than needed, you need to clock your DAC so badly, with jitter on the order of tens of nanoseconds for the artifacts to even arise from the noise floor, let alone be audible, but hey, audiophilistines know better and don't take anything more than femtoseconds.
@6:35 an oven controlled VCO is exactly what it says - it is temperature controlled to optimise clock waveform & jitter performance - it is NOT an “temperature controlled” crystal, that is voltage is still used to set the crystal into its designed resonant frequency, but temperature controlled conditions, stabilise & optimise the oscillator performance😀🤓
This is a great video! I just purchased the iFi iPurifier SPDIF 2, and I really enjoy it. It makes the optical audio on my PC so much better in terms of quality, and music sounds amazing. I can even play my songs on Tidal at 192 KHz. I haven't noticed much on my Xbox One, but I still use it to prevent jitter.
Thanks for the amazing review. Ethernet factor is huge, and it is ignored in many digital audio reviews. I did a $200 investment in my roon Server in the basement (6 core xeon with a noisy fan) and it changed the sound all over my house not only where I had the best roon endpoint ($4000 DYI).
I thought by now any reasonable DAC would have it's own internal elastic buffer and just internally re-clock using it's own precision clocks. Paul Lesso wrote an AES paper about S/PDIF design 15 years ago called "A High Performance S/PDIF Receiver". A small buffer chip should be super cheap these days, and should be able to wipe out any jitter from a PC optical port. I get the impression audiophiles like to make simple problems hard.
@@sanjacobs6261 Delay is bad if you are watching a movie or trying to mix tunes as a DJ, cuz the audio will arrive too late. But if you only listen, it doesnt matter if the audio comes out a second later then it got send.
@@gayusschwulius8490 For reclocking you usually use atleast 2 flipflops. This means that it's atleast 2 cycles of a clock worth of delay. The clock in this case has the frequency of the bitclock which is usually 44.1k*16bit*2 = 1411200bit/s. Which corresponds to 0.7microseconds of delay PER flipflop. I'd say delay typically becomes audible in live content at about 15-25ms. You could legit use a thousand flipflops and still not hear the delay.
I recently added a Gustard U18 to convert USB to COAX between PC and DAC. The improvement is huge! I get an immersive 3D soundstage with relaxed and sweet sound. After that I tried direct USB connection once and couldn't bare with the "digital", dry, lean sound at all. Your explaination makes perfect sense, noise and jitter matter most between source and DAC. BTW even USB is not really bit-perfect, as USB audio uses isochronous protocol. Unlike bulk protocl, it doesn't have error correction.
Just a question - "...the Schiit EITR is a very affordable device..." I assume you are aware that this product was discontinued about a year ago and is only available second hand? (where it commands a huge premium over the original MSRP). FWIW, the EITR was based on Schiit's older USB implementation (called Gen5) and the EITR was discontinued when Unison was introduced. Also, I agree that the Pi2AES is outstanding... I have 3 of them.
Eitrs sometimes pop up in Schitt's b-stock page for $99 plus shipping. Schiit discontinued Wyrd as well, so I guess you have a point unless someone wants to use spdif coax connection for older DACs like me.
Dude, this is the third time in a row you make a video about the stuff that is on my mind the bigger part of the day. your now my favorite Tuber. And about MQA I got your back all the way. 2 years ago I also put a lot of time (for me) in the new sensation MQA and concluded, it's kind of scam, and your video, organized all my chaotic leftover memories about that subject, and now it is as clear as a FLAC file. Respect.
Wouldn't mind seeing a video talking about power input with regards to audio equipment: the effect clean power has on equipment, usefulness of power conditioners and UPS's, and what to actually do if mains power isn't reliable with frequent brown-outs and general inconsistencies.
It sounds like some dacs would benefit a lot from a Pi2AES, where as others (the more expensive one's presumably) would benefit less if at all. I'd love to hear your recommendations for a good bang for buck combo with the Pi2AES if you end up doing a dedicated video on that device.
Outstanding review. Man, I wish you had a Denafrips DAC using the external clock function with the DDC for this round of review. I just ordered the Terminator-Plus DAC with the Gaia DDC and Kinki Studio EX-P7 with EX-B7's.
Great presentation. In my system, the subjective sound quality is noticeably impacted by jitter. Yes, lower jitter improves the sound stage, and also bass response. The biggest impact that matters to me is just the sense of natural sound. Glare is removed. That little tiring digital edge melts away. And everything sounds real. I believe great sound comes with micro detail from which we sense more than hear stuff that we equate with reality and emotion. So for me the sound of a DAC depends on: 1. the actual converter or chip, 2. Great power supplies for every consumer, 3. The output stage, and 4. The clock signal. Lots of people don't hear it. They may listen to music that really doesn't need to be well reproduced and the rest of their system could easily mask these micro details. But if you hear it, there is no turning back.
For us noobs, might I suggest you spare a few moments to explain you graphs better and how one is better versus the other? I kind of think I understood so thanks for the great effort and video!
Soooo I'm somewhat new to audio. My understanding of those graphs is the more lines you see randomly sticking out from a pack, the worse it is. At 13:50 you can see a line sticking up on the left graph towards the left of that graph, then just before the big spike, theres a bunch of bumpy lines sticking up from most of the other lines. Neither of those are in the right graph. Those things are the noise/jitter. (I COULD BE WRONG.)
I'm not into audio at all. Just came here from my recommendations. But I have done some electrical engineering. The plot has frequency on the x axis and loudness on the y axis. He's sending a 12kHz sounds wave across the line which causes the giant spike at 12k. Since it's supposed to be a pure 12k signal, everywhere else on the plot should technically be at negative infinity (no sound at all). A lot of the reason it isn't at negative infinity is because of random electrical noise or jitter. Jitter causes other frequencies to spike slightly because changing the timing between points in the wave changes the slope of the sound wave and thus the apparent frequency of that wave.
You should check out the Cherry 130dB+ DAC DAC 3, which will be introduced this summer! The coax/optical SPDIF input is internally reclocked for jitter immunity. Several levels of power supply reconditioning. True balanced, DC coupled outputs, too (:
At first I also thought usb wouldn't matter, but I noticed an audible noise coming from the apple dongle on my laptop, specially when the fan was running louder and a cheap usb isolator from alliexpress was all it took to fix it in my case.
Very informative, thank you. I have always intuitively done my best to stay away from pc in my audio chain as ive always supected id have electrical noise and jitter issues. Having recently started mixing and recording on pc though, i'm in need of paying attention! I recently switched my hifi to a Naim Atom , so hopefully my dac is being supplied with an optimal signal!
Well, I'll find out soon. I'm waiting for my Pi2AES Pro Shield to arrive. I'll tell you this much, the Topping DX7 Pro benefits from an op-amp upgrade. I installed a full set of SparkOS units, takes the unit up a few notches.
And the verdict is... We have a winner! I use my DX7Pro as a DAC-Pre straight into my amp. So my take is the Pi2AES on Rpi4 running Ropieee via I2S elevates the sound quality by a decent margin. Overall the sound is even more coherent, soundstage width and depth is wider and deeper, the sound has a bit more body to it if that makes any sense, bass is more detailed, everything seems a bit more detailed, you hear more into the finer details of the music, transients and such, better decay and layering of it in more complex passages. I'd be saying these things with the Pi2AES regarding the previous upgrade, but this brings out even more. I wasn't expecting that at all because I thought I was hitting limit with what I already had. Not bad for $350 (I bought a FLiRC dongle as well).
Really interested to hear your thoughts on the May Holo. I ended up going with an Yggy GS for only like a 1/3 or so the price, along with a Pi2AES. I still wonder how it'd compare to the Holo. I guess one perk of the Yggy is that it's upgradable, and Schiit is already working on a new analog board for it. Looking forward to your review!
Another thought. I loved this video because "USB noise" was one of the topics I was wondering about and I was also among the ones thinking bits are bits in a digital signal and audible noise cannot be generated on a USB connection. And you almost convinced me when mentioning the audible noise from a PC. We all know that, that noise is with us since the 90's. You can hear the mouse movement, GPU working, etc. But then I realised it's a completely different scenario, that is with the PC's builtin soundcard and a direct (jack) connection to the speakers or headphones. As soon as you have an external USB sound device, noise is gone. So I'm afraid I'm still sceptic about USB noise :) (jitter is another thing I understand that)
I switched from using my gaming pc as a roon endpoint to using an allo digione signature (similar to the pi2aes from my understanding). I wasn't hearing anything obvious via the usb audio on my pc but when I switched to the RPI allo streamer the signal sounded quieter and cleaner but it definitely wasn't a huge difference. Good video though I've been really enjoying all the content on this channel
I think these problems should have been solved in the DAC. The whole reason I am considering buying an external DAC is to get rid of electrical noise. Cue the "YOU HAD ONE JOB!" meme.
Considering that analog LPs are still considered to be the ultimate experience by many high-end enthusiasts, a 60db noise floor is not a problem and a 110db one even less.
@@dingdong2103 you’re misunderstanding the problem. The issue isn’t noise floor: it’s electronic noise and jitter which are a *completely* different thing.
@@GodfreyMann What is 'electronic noise' if not the thing that rises the noise floor? And jitter is not a practical problem in reality, it's just marketing jargon: hifigo.com/blogs/guide/what-is-digital-jitter-and-how-to-avoid-jitter-dac-101-part-3
@@dingdong2103 you’re correct that ‘electronic noise’ on the analogue side rises the noise floor, but the discussion is about ‘electronic noise’ on the digital side which gives rise to jitter. It’s not a marketing jargon - the clocks are voltage sensitive components and ‘electronic noise’ can cause fluctuations in voltages in the circuit. Although tiny this can affect the quality of the square wave signal and therefore how the bits are interpreted....it can change the bits and timing.
2 years later. I am now using a pi2aes 2.0, with a 850€ linear PSU that powers both the board and the Chord Qtest, with files stored on the SD card, optical out, with real glass cable, and music sounds more liquid, it has more shape somehow. I did a test: same setup, same files, external usb ssd storage vs internal storage on sd card, and turns out, it does sound a little bit better when there is no usb in the chain. Same dac, same input, same output on pi2aes, the only usb was the usb storage and impacted the music. I don’t know why, but from now on there will be no usb in my system
Thanks for the awesome video! I wish that at some point you'll be able to make a video about AoIP with devices that support Dante or Ravenna. I'm really interested to see if they can bring any noticeable improvement to the sound. I've been eyeing the Merging Anubis for quite some time now.
I do not often see videos which are pure and utter nonsense. But this one is. The spdif and the usb interfaces have error detection integrated. They will not deliver an invalid data stream to the DAC. Noise on a data line can cause problems but usually you recognize that by drops in the audio presentation or by connectivity rejection (no sound - full stop) and not by noise on the analogue side. Since the cable length and quality is good these days, you usually have however not problems or drops.
If at all possible, you should beg, borrow or steal an EtherREGEN along with a quality external master clock, such as an AfterDark Giesemann or a Mutec. I've added both to my Pi2AES along with quality LPSs. My listening experience has been transformed for the better. I've never heard such 3-dimensionality, fluidity, realistic dynamics, sweet resonant decays, etc.
I have a pi2AES which I run into a RME ADI-2 Pro and then to my Genelec speakers. The sound is glorious. I first used SPIDF but then switched to AES which to me sounded better. I do not know why it sounds better, but it does. More bass control and overall cleaner sound. I have not experimented with i2S yet. My conclusion is that the pi2AES can be used in a chain to produce the highest quality audio. As a side note, the RME ADI-2 Pro is a match made in heaven for Genelec speakers as it can be used for AES or XLR output, thus making it possible to send both digital and analog to the speakers. For example I stream Qobuz and play vinyl from my TT through it. This is perfect for me because I love both digital and analog and can’t live without both.
a phase-locked loop is literally necessary to keep multiple signals in sync, I don't think I understand how something is supposed to properly interpret a digital signal it's receiving without one, the clocks would keep going in and out of sync.
Audiophiles: Measurements don't matter, all that matters is what you hear!! Also audiophiles: This device has a noise floor at -144 dB vs that other one that has it at -120 dB and costs 1/10, we know that it is a proven fact no human being can perceive the difference but just look at these GRAPHS!!!
In fact, any unwanted noise at -144db can still affect the sound quality when such frequencies are played, its still a delta to the "perfect" soundwave.
My thoughts exactly. There's no way in hell a human can hear a signal that is 120 dB lower than any other simultaneous sound (that is a millionth of a millionth of the energy in the main signal), even in the ideal condition when the main signal is a pure tone that does not overlap in frequency with the jitter-generated harmonics. Maybe a super low noise no-jitter-at-all DAC is useful in lab environments, but as a music consumer...
Since you are basically the only one to do this kind of things on RUclips (and thanks a lot to do so!), I think you should take more time to explain what you present. This is some really really interesting and in-depth information, it deserves longer videos, or more videos. Something I don't understand: why should we need a device that clean the signal from the source to send it to the DAC if you can have that in the DAC itself? Shouldn't it be a requirement for any DAC to basically reclock what's coming from the USB and isolate from the voltage coming from the USB?
Great video. But the giant question it leaves unanswered is whether the little no-ISB Pi device that beats the Hermes is uniquely good at destroying jitter, or whether *any* transport without USB (such as my MiniDSP SHD Studio) would yield similar benefits, simply from avoiding USB.
I'm considering buying a sub $1000 DDC and have 4 candidates, but am open to other suggestions. Candidates are 1) Gustard U18 2) Singxer SU-6 3) Denafrips Iris 4) Topping U90 In stock form, which puts out the cleanest IIS signal?
All of them sound exactly the same, but I went with a Topping d90se. I use its USB input, because it is the most convenient option and handles any sampling rate. There’s no benefit to IIS, despite the marketing mistruths perpetuated by PS Audio’s Paul McGowan-who promotes that tech because it was required under his clandestine agreement with Sony to circumvent their DRM rights for outputting an unadulterated digital signal from SACD players to his own proprietary DACs. Don’t fall for this ruse, it’s an unjustified expense that has no impact on “sound quality”. Digital audio is not sound, and no matter what transport you use, the same signal is passed to the DAC and the resultant outcome is indistinguishable to the ear.
Two questions about the Pi2aes. 1- How does one change inputs on the Pi2aes? I think they all run continuously and have to be changed by hand. 2- How is the sound from an external/remote hard-drive into the Pi2aes via USB,,, then output to a DAC of choice? Thanks ahead for your help.
Got my Pi2AES yesterday and am so impressed with the SQ compared to a regular Pi3 on USB. AES sounded much better but going to i2s was a night and day change. Thanks so much for bringing this to my attention Golden!
I got a pi2 and am using it with an Allo USBridge Signature. Still can’t it to work. Any pointers? I am using the same SMPS power supply as recommended by Mike Kelly and the USBridge is powered via the PI2AES. Thanks.
Hi @GoldenSound! Great reviews, as always. But unfortunately looks like the Pi2AES project is no longer available and they have a closed and more expensive version. So, I was wondering, with the release of the new Topping M50 which is kind of a weak streamer, can it be used as a cheap USB to IIS converter? If so, could you measure that to see how clean the signal is? Thanks!!!
Considering he already did the number crunching, I should hope they wouldn't have much to screech about - though I know that's probably wishful thinking. But so long as it just means more data, and not just ridiculing, then all is well and good.
iPurifer3 didn't do anything at all for me, but iGalvanic3.0 was a massive help for Hugo2. When I sold hugo2, I thought it was only right to sell it with the iGalvanic3.0 because its pretty much a requirement.
@@-havoc-tzu-3197 Hard to believe the Hugo2 doesn't have galvanic isolation isn't it? The Hugo TT has it. I would bet anything that the Hugo3 has galvanic isolation because the competition is stiff and the Hugo is high priced.
i used a Schiit Eitr for good while on windows, it did seem to make a small difference, but nowdays i mainly use it as a way of having a 100% Linux compatible device, allowing me to use any DAC on my Linux desktop PC. Another hidden benefit.......of a Schiit Eitr
Great Video, I agree with using I2S, I was wondering if you have heard of IAN Canada. I found these HATS for the PI and I was thinking about using these two devices with a PI 4. 1. IAN CANADA FIFOPI Q3 ULTIMATE FIFO Reclocker Module PCM 32bit 768kHz DSD1024 DoP 2. IAN CANADA HDMIpi MKII Transmitter I2S / DSD / DoP to HDMI
Golden sound I'm sure you have explained more times than you care to remember so please from an American fan, what are those card or pictures on the wall I've been wanting to ask for a year or so. Thanks for the videos!
-120db noise from jitter is NOT EXTEREMELY audible at all!!, in fact your amp background hiss more audible more snake oil audiophoolery, would like you see your "golden ears" detect jitter in a blind test where you really have no clue which is which
I'd like to, but unfortunately MQA has quite deliberately made it VERY difficult to obtain any equivalent files to test. And no MQA full-unfold device is allowed to have digital output. The only actual comparison i've seen where there was an MQA and non-MQA file that were the same was done by stereophile, and it showed all sorts of problems.
Given that the Pi is using the I2s, which 'DAC' hats give the best analogue output? Ive noticed that some 'DAC hats' don't seem to include a crystal, so presumably rely on that of the Pi, so I'd guess those could be ruled out, as the Pi crystal is not a good match for the function required. I recently bought an audioinjector Ultra, so I'm interested in how the different makers implenentations compare.
The pi2AES with I2s output is no longer available. The manufacturer now only offers I2s output with their $600 Mercury streamer. However, production has halted due to a supplier issue with the case manufacturer. According to the website, the PCB board for the Mercury may have to be fully reconfigured to fit another case. Translated, it might be quite some time before the Mercury Streamer is back in production. At this point, it looks like I'll be forced to look at other options for the foreseeable future.
Question about i^2s Guido Tent ones told me that i^2s is only for internal dac use not for longer distances. And i can understand that because clock signal should as short as possible especially digital clock signals.
I gave up explaining how jitter can be CRITICAL to your music a while back. Took me a while and a lot of experiments to learn a lot of. Spending time trying to pass that knowledge to people who are not receptive is a horrible way to waste your time. I would recommend giving Lavry Engineering DA10 and similiar DAC a try with their "crystal lock" feature enabled... See how wrong engineering can go when they confused clock drift with jitter. Their tight PLL sounds great tho. Another thing i find suspect is Analog Device's ASRC that claims to eliminate jitter by running signal through SRC, but i suspect it is just blending the good and bad part of the signal into a inseparable mixture.
Why would you have to explain something that's, according to this video, "extremely audible?" I've never heard it, and have been an avid listener for over 40 years. What DAC does one buy to hear this phenomenon? What does it sound like?
@@Grassy_Gnoll it is extremely auditable if you know what to look for. If you are facing a crowd of "if it is bit perfect, then there's no difference" horde, then you will see what i mean. To me, higher jitter often results in blurred imaging, worse soundstage. To try it on the cheap... Just get a player and DAC with optical and coax. Then compare them. Do remember to use cheap plastic optical cable and cheap coax.
Interesting vid, I recently tested 3 different DACS against each other I compared the AudioEngine D1, the DAC built into my Yamaha A-S301, and the DAC in a $110 CD player from 2001 and after volume matching I couldn't tell a difference. I've yet to try a DDC but if my ears can't tell a difference between the DACs I doubt a DDC would make much of a difference for me but I would be willing to try one out.
Yeah, I'll be honest... It's super super subtle to me. Only difference I've ever been able to distinguish is the difference between multibit or R2R and delta sigma. But, even between completely different topologies, it's very subtle to me. It becomes more noticeable with high end headphones compared to my 2 channel setup.
External DACs are more or less a waste of money. Not even talking about DDCs, thats the snake oil people talk about, even external DACs are more of snake oil than real advantages. In the days where only proper DACs were in extremely Hi End gear which was probably not even aviable for the regular customers, an external, better DAC than the one in your anno 1990 100$ JVC CD Player would make a difference. But i realized how useless "better" DACs are as i switched from a cheap 90$ DVD Player (i think 96kHz 24Bit) to a proper CD Player for 600$ (192kHz 24Bit), fitting to the 700$ Stereo Amp. There was absolutely NO DIFFERENCE, except for the removed DVD drive noise whereas the Cambridge Azur 650C was totally quiet. My AVR has even a 384kHz 32Bit DAC... and it somehow sounds the same as the cheap DVD or expensive CD player. The sound difference between my 700$ AVR and my 700$ Cambridge Audio Amp is also very negliable. Where the AVR (175W/channel advertised!) begin to struggle a bit with bass at higher volumes, the Cambridge 650A ("only" 75W @8 Ohm) doesnt sound that stressed on my IQ 90s. But at any regular volume AVR aswell as Stereo Amp (both feeded completely different, Stereo Amp analoge over CD player, AVR with a bit-identical FLAC file over USB Stick) they are more or less identical, you only hear a bit of difference when it comes to the "cambridge sound", but its not far off from the "linear" sounding AVR and is pretty easy EQed even with the 2 controls on the Stereo Amp for Treble/Bass.
@@harrison00xXx I'd generally agree with you, but with one caveat. There does appear a benefit with desktop/gaming PC setups. This is largely just due to computers being very noisy (electrically) inside and there can be a benefit of moving your DAC and amp outside of the chassis, and on their own power supplies.
Like many of us i use my PC directly to my receiver amp ( Yamaha RX-A3070 with ESS pro chips ) via an HDMI cable connect to my graphic card.... HOW IS THE JITTER with this connection ?? Thank for the answer ( i hope )🙏
Very informative review, I like the measurement addition! You keep mentioning the Pi2AES price as the full price of the streamer but adding a Raspberry Pi and a good power supply is a must, so the total price will be around 600 $. Great job anyway!
Hello! First video of yours I’ve seen. It’s amazing, I’ve watched it like 3 times already haha. I bought an eitr, and as ashamed of this as I am, i must know.... how do i connect it to my modi multibit/littledot mk2? I need to know what cable to order and what not lol. Sorry for the inconvenience. But that aside, I love your content! Keep it up!!! P.s. everything’s connected to my gaming pc.
Excellent video & best explanation yet of how noise and jitter can affect DACs. Also your hint at a possible review/analysis of the SoTM sMS-200 Ultra is most intriguing! I use this unit together with the SoTM tx-USB reclocker (which I guess is a type of DDC). I find the tx-USB definitely improves sound quality, but I would love to know how it does it!
Just some "personal" perspective from my experience as a software developer for real time embedded systems and deep interest in electronics. (not professional in electronic tho) I'll only talk about those three topic you start with, over usb, some applies to other protocol, but I don't know them in detail and I'm not going to read pages of specifications for a youtube comment. The TLDR is just to say, normally, you shouldn't have to care about those three issue in a "just enough" designed modern usb device, even more one who cares about those well known issues and tells customer they're some kind of high end devices, dac, or anything else. Data integrity Being digital doesn't mean you can't have data corruption, for instance if you communicate over serial without any specific protocol, or a protocol that doesn't integrate some form of data integrity mechanism, you can have data corruption and no fix for it. That being said, the usb protocol specs specify a mechanism for that. A CRC is sent to ensure the integrity of the received data and if the check failed, the data are re-transmited. (to make it shortà That means, for any device compliant with the usb specifications, it's not possible, on the transport, to corrupt data and not notice it and fix the situation, it's by design. So no need for extra middle man. So, being "bit perfect" and "not altering data", I don't understand the take. A usb hub does it, nothing special. But I don't understand what the point is, so I just mention, to anyone worried about this, usb, by itself, protect against data corruption on the line. If you worry about that, use usb, and you'll be fine, by design. Electrical noise You mention using optical link as solution to decouple two electronic circuit. That's way more on point than it seems and is in fact quite... "original', that any "high end" dac doesn't do it. The solution is called optocouplers, it's used everywhere, even in low end electronic, it's cheap, and it does just that perfectly. The fact you have to introduce middle man to solve that issue for expensive units is a shame. Also, optocouplers main interest, outside of the noise, is mostly security concerns, and it should be used in scenario with interconnection to any other unknown circuit. For instance, you certainly don't want to blow your expensive device by linking it to a faulty one that might send huge spikes to yours. If you design something like a dac, being proactive about those situations shouldn't even be questioned, even more for expensive equipment where you should do anything possible to protect it from the outside world at the benefit of the customer and your brand image. At the same time, solving noise propagation issue through data or ground lines. Imho, any dac, even low end, should do it. And to be honest, it's a solution found in a lot of studio equipment, and you don't have to go high end for it. So blame the dac manufacturer for not doing their job. Jitter It was a real issue for instance with spdif and the reason was... mostly a "bad" design we don't rely on anymore. Also there's different type of jitter with different sources and different solutions. The way you represent what jitter introduce as a problem doesn't match the general design of todays dac, dac aren't clocked by the input protocol clock anymore (and if they are, it's on their own and they decided to go the wrong way, maybe buy something else... who easily solve the problem) But, usb has a well known workaround. The main solution to solve almost all jitter issues is to use asynchronous mode. Jitter will be small to the point it's humanly not noticeable anymore and, if you really want to solve that picosecond problem, just introducing a buffer or related, audiophile dacs, imho, can do it, I wouldn't mind it a all as a client, but might be wrong. The fact this problem is still present really makes me wonder: are they using 20 years old design that just "get better over time" but don't do re-design that would solve the issues ? Once again, it's the dac manufacturer fault, it's a design issue, it's a known problem, with known or even integrated solutions, if problem persist, maybe go back to your R&D and fix it before requesting any substantial amount of money from your clients. Paying a middle man device to fix those three issues, at least over usb, until you are like using an old dac you really love, shouldn't even be a thing. Shouldn't even be mentioned on modern hardware, not even as a selling point. It's something people sometimes say about dac to say every dac is the same, I don't have opinion I want to share on that topic, but for those three issue, if you have a modern dac that features them, I would consider them "by design broken dac", any mention of those issue on something I plan to buy would make me totally never ever go toward that manufacturer again. To me, those issue, on new hardware, shows a total lack of dedications and I really hope I get this video wrong and it's about fixing old gear issues. At least for usb, don't know for the rest, but wouldn't be surprised there's solutions too and modern hardware shouldn't rely on middle man.
you seem to know a lot about the electronics, can you explain the concept of electrical noise? i get it on analog signals where stuff like magnetic fields and whatnot can influence the voltage on the wire, which causes the speaker to produce the wrong amount of air pressure, but for a digital signal you'd have to flip a bit, no? wouldn't this directly contradict the claim of the data being bit perfect?
@@LastExceed Hi, I'm not sure what you want to know, so let me know if it's not what you meant to understand :) Technically, digital signal is analog signal. Electricity is always analog and will always be. Digital signal is a way we interpret the analog signal. For instance, let's say you have a 5v digital signal, everything above 2v will be considered a 1 and below 2v considered a 0. So at a specific time a signal can be 3.5v or 4v, so they"re not the same electrically, but they'll be both considered a 1. (To simplify) Now if you have any effect strong enough, you can flip a 1 to 0 and vice versa. Also, it's time related, there's different clocking strategy and ways to define what a bit is and how is value is "read" from that analog signal, but depending on that, you can have a really unstable signal that might be misinterpreted. But as I said, there's integrity mechanism most of the time, especially nowadays, to detect those corruption and re-emit or fix the data. Or identify something went wrong and apply a strategy, that can sometimes be invalidate everything and consider everyhing corrupted, so basically untrustworthy garbage, sometimes it's more about being sure not unwanted change occurs, than preserving the "user experience" and trying to find a fix, so maybe shutdown and lock the machine until human intervention, I mean, lots of possible scenario depending on the context. Let's say it's a strategic infrastructure machine, do you prefer to take the risk to autofix a situation that might have been caused willingly by a bad actor that can use that as an attack vector, or realize something went wrong, lock everything up, pause everything and wait until investigation ? Bit perect digital signal is, in a way, impossible by itself. But by implementing control and error fixing mechanism, it can be done on a specific scope. Or at leat, be done consistently and reliably enough, that we consider it bit perfect. (doesn't mean it can't maybe fail once every billion year, or could be bypassed willingly by an actor, or other stuffs, depends on your context, stakes and how much you're willing to put in place to protects against that problem) For instance let's say you have a voting machine, RAM corruption in that voting machine would be a catastrophy, so we use ECC memory that protects against that for intsance. In my country, we had, almost 10 years ago, a voting machine with a strange bug during elections, long story short, it was a cosmic ray and the machine didn't had ECC memory. (might sound like a crazy conspiracy anti science stuff, but nah, cosmic ray issues with electronics are a well known and managed problem, I'll let you google for more details, you'll find research paper about that starting from decades ago by trust worthy actors like NASA or renowned universities and a lot of mitigation strategy since then) The register containing one of the candidate vote count flipped, and so his or her results. When at a scale large enough and with stakes high enough, you start to see and have to consider really really strange and "rare" effects. Also, ECC memory doesn't protects against possible corruption done before or after being written or read from ram. But yeah, there's mitigation solution. Being bit perfect is about the control and fixing mechanism you implement. Let me know if I answered your question.
@@EmmanuelIstace Oh I've heard about the voting machine thing! I'm trying to understand what is said at 4:39. Are they suggesting that the noise flips bits during transmission? Even if we ignore the USB built-in data integrity stuff, how would a special device at the end of the wire change anything about that when the data it receives is already corrupted? I think I'm horribly misunderstanding what is even being said, please show no mercy in correcting me
@@LastExceed No worries, I'll never be harsh with someone trying to genuinly understand something, no worries :) edit: before hand, I use interchangably CRC and hash in this message, consider the term the same in the context of this example. They're not, but here it doesn't matter, as our CRC is hash based. So, I listened quickly from that timestamp, not sure what they're refering to, might be relvant, or totally not. And I remember I had, let's say really mixed feeling about this video few months ago, so not going to rewatch it. But about your question, how can the receiving device detect and fix data corruption. In my case, it's about corruption on the link, so during transmission on the wire. If the Source machine send data that would qualify as "corrupted" even before sending it, there's no way to detect and fix those. (I mean, there are, but it's way too advanced for what we're talking about here and for scenarios out of context for here) But we can make sure that what was sent and what was received are "bit perfect". Data are sent by packets most of the time. (and also batch of packets, but not relevant). So small chunk of data. Let's say you want to send the data "hello". And let's imagine you use a protocol that send each letter as a packet. You'll send 5 packets, each containing a letter. For each packets, you'll have meta data, so informations about those data, for the purpose of the example, a sequence number. In those metadata you can include things like a CRC. A CRC is a hash of the data. Basically a hash is a computer generated mathematical representation of another data. It can be viewed as a "summarized" version of the data. For instance you can have a 64bit hash that correspond to a 1Gb file. Those hash are totally deterministic, that means if you put the same data in, you'll always get the same hash out. You can compute a thousand time the hash of that file, as long as the file didn't changed and you use the same hash method, you'll always get the same output hash. Also, it's not encryption, it's hashing, that means you "normally" can't compute the data back from the hash. (but you can compute the hash of every possible data and consult that table afterwards to know what data might be represented by a specific hash, that's called a rainbow table, but out of topic) So to detect data corruption, we'll use those CRC. Back to our simple improvised protocol. The source machine, for each packet, will include a CRC of the data payload. The client machine, when receiving each machine, will compute the CRC of the payload using the same algorithm and compare it to the CRC from the associated packet. If they're not the same, the client machine knows there's a problem, either the CRC was corrupted, or the data was corrupted, but something's wrong. In this case, to fix the situation, the client will notify the sender and say something went wrong with that particular packet, and request it again. In practice now : The source machine will send 5 packets, I'll write the sequence number, the data and the CRC. Each line is a packet. 0 - "h" - fs1f6s8 1 - "e" - wcx87w 2 - "l" - w354c8er 3 - "l" - w354c8er 4 - "o" - k468h38 As you see, packet #2 and #3 have the same hash, as the payload is the same. Those packets are then sent, let's take a look at what was received : 0 - "h" - fs1f6s8 1 - "e" - wcx87w 2 - "_" - w354c8er 3 - "l" - w354c8er 4 - "o" - k468h38 As a human, we see quickly a problem, the packet #2 have corrupted data. But a machine is not intelligent (not even "AI", but not the topic) and have no idea about what the sender meant to be sent. So the target machine will compute the CRC for every payload. Here's the result: 0 - "h" - fs1f6s8 1 - "e" - wcx87w 2 - "_" - yui23y1 3 - "l" - w354c8er 4 - "o" - k468h38 We see the hash of the packet #2 have now also change, as it reflects the payload which is different. That means now a machine can detect the error. It will compare the CRC received, from the CRC computed, and detect that the CRC for packet #2 don't match. So it will request again that packet unless it receives the correct information. Now, this is a simplified version of all this, also CRC can be about more than the payload, or a part of the payload, or part of the meta, there's what we call "hash collision" (that's when two different source data produce the same hash) different strategies than using CRC, etc, there's also a lot of other situations that doesn't directly imply those kind of corruptions, but I hope you now have an idea of how data corruption on the line can be dealt with. In most "usual" scenarios, strategies like the one here, or similar or derivated are used, often way more subtle, refined, etc, but the core idea and principle is often this one. If I wasn't clear on something, let me know ^^ (also, I read back the message, sorry for the grammar etc, I'm not a native speaker nor living in an english speaking country) (edit 2, haha I realized I copy/pasted wrong my list of packet and introduced a second error on the last packet, haha, that's why machine, even tho they're dumb, are more reliable than humans, a machine would have detected that packet too haha)
@@LastExceed I came back to check if you had another question and realized I maybe answered the wrong question, if the question was about noise through ground lines, then yes, datas can be bit perfect, but noise coming from ground lines can interfere. It can introduce inconsistencies in the DAC device once in the analog domain if there's electrical coupling. Long story short, for instance on a lot of amp and depending on the design (the amp inside the DAC) if there's potential difference inconsistencies with ground the output will have inconsistiencies too. That's what I adress in the point two of my original message. In that case decoupling the two device will solve that issue, and that's maybe what one of the box presented is doing. But as said, that's a huge design flaw for anything that would consider itself serious on the market and even more a security concern than a performance concern.
I appreciate that you've presented actual measurements, but this seems like a solution in search of a problem. These DDCs seem to be all about second guessing the internal design of the DAC you paid good money for. A high quality DAC should already have its own isolated power supply and isolate any electrical noise coming in from the USB. It should also already have a high quality clock which is the only place you need to worry about jitter. If you just use the USB input, there is no jitter from external sources so no need to spend more money to try and minimize it. Just buy a good DAC in the first place.
While it's true that you are essentially "secnd guessing the internal design of the DAC" and that the DAC already SHOULD have this implemented, it doesn't mean that it actually has. A good clock oscillator requires a good crystal (which is just expensive to manufacture/buy). Attenuating input jitter can only be done IF you have a proper local oscillator. You are making a compromise between how much the external clock matters and how much the local clock matters. For USB DAC's this is irrelevant because USB is asynchronous. For power supplies I completely agree though. You won't make the power supply much better than the way it's implementend internally.
@@hidjedewitje with asynchronous audio dacs you also have the option of putting in slave or master mode when using an external master clock - despite all the advantages of async usb, the search for cutting edge performance continues... Unless you're Bruno Putzeys, design a good one 10 yeara ago & its still cutting edge....
@@BogdanWeiss Bruno's stuff is indeed very fascinating. His work at Grimm is also quite fascinating! The AD1 (now discontinued) is still the best AD converter when it comes to linearity and jitter performance 20 years later! The question is, do we really need such performance?
@@hidjedewitje Performance just like intelligence is a double edged sword - once seen or heard, it's difficult to un-hear it - I didn't realiae that the grimm adc has been unchallenged for this long - truly impressive
@@BogdanWeiss While true, there's definitely diminishing returns. The difference between a laptop's on board audio to a standalone DAC with THD+n of -115dB is pretty audible. There are also DAC's with THD+n under -120dB. I doubt anyone can hear those differences, despite the performance being objectively better. There's also people who stare themselves blind on numbers regardless of looking at the weakest link in the chain. They buy DAC's of $1000, but leave the room untreated (resulting in dips of 30dB in the frequency response!!!). There sure is a place for state of the art products (like the Grimm's or the Mola Mola's), but they should also be placed in a state of the art environment to be actually usefull.
It matters if your DAC is not able to remove interface jitter and electrical noise on the incoming signal. Why not simply get a properly designed and built DAC? This is the same as buying power conditioners - you only need conditioning if you have an audio component that can’t deal with regular noisy power. You also should use an asynchronous DAC - which decouples master clock from the crystal clock in the DAC. Basically you only need extra stuff if the components you have are sub-optimal to begin with.
@@DueM I did watch. I was summarizing the implications of the video in terms of “so what should one do in practice”. There is a very old lesson here that I learned 20+ years ago. Summary: Buy well built and well designed components and connect them properly and you won’t have a bunch of issues like jitter, ground loops, hum etc. The sad thing is that well built and well designed components are extremely rare and most folks don’t get rid of poor performing boat anchors because they are attached to their toys and, as a consequence, they end up with loads of “band-aids”.
@@jeremyhughes6485 it's not the dac that's the problem it's the source, that's the point of this video. There's plenty of dacs with good internal re clockers, funnily enough even they will benefit from an external ddc in the input chain. Whether or not it's actually audible is another story all together so I leave the donglemania for the pedants.
@@DueM Sorry. I simply have to disagree. There is always some degree of jitter on every digital input (it is an interface issue) - so the better DACs are asynchronous (they ignore the input clock jitter) and in that sense you could stick anything in between the digital source and a well designed asynchronous DAC and it would not change anything as long as the signal remained bit perfect. An asynchronous DAC can be fed a very high jitter signal and remain completely unaffected. (Benchmark demonstrated this a long time ago with their DAC1)
@@jeremyhughes6485 jitter reduction is built into most dac chips, async is ok but wasapi is more stable these days and you're not using proprietary drivers which are notoriously buggy. benchmark dacs use sabre chips which have inbuilt jitter reduction anyway and are pretty much identical to a hundred other dacs using the same topology. we can agree to disagree but as long as pcs are the source theres always going to be lots of noise, hence these devices.
Respect my avatar
no
@@lwwells Yes
I thought the MQA video was good then I watched this one! Bloody brilliant, science based explanation of how electrical interference has nothing to do with snake oil. As a direct result of this vid bought a usb galvanic isolator (the intona 7054), 2 decent usb cables (also sorted power supply end out too) and my god my setup has been improved to a new level! I will for sure sort patreon support! Great work
I’m so glad I have found this channel. Looking forward to the Hollow May review.
Dear Goldensound,
In this world where pure gold and sneak oil (EDIT : « snake ») are sometimes so difficult to tell apart, your reviews are really in a league on their own, thanks to your deep understanding, objective measurements, and explanation skills.
Very helpful.
Kudos ! And thank you so much.
It's just his accent
'sneak oil' is deadly stuff. Also Snake Oil!
@@zogzog1063
Edited. Thanks !
Excellent explanation on how clocks work with USB vs SPDIF. Thanks!
I’m so glad I have found this channel. Looking forward to the Holo May review.
Wow, really enjoying this channel and the clear technical explanations. I'm glad to find out that my DAC has PLL for all SPDIF inputs, I was about to go researching a Pi2 AES!
The Pi2 is an amazing device.
I appreciate the actual graphs that prove what you're saying. I'm new to high-end audio, and for the most part I assume that the things that people tell me, like "this is better than this because this one costs 10x as much and it makes me feel good" is mostly subjective, but you've supplied receipts which actually puts weight behind your words.
Also: NONE of these people commenting here, nor the uploader has ever heard jitter in their lives. They do not conduct proper blind testing but just conclude things. Measurements do not lie but most people don't understand the decibel scale and the number in it. If it is below -100dB: do not bother, you will never be able to detect that in any real life scenario, yo can only catch it with measurements. In most cases we should not care what happens below -70dB but it is better to be safe and aim for a bit better than Redbook, 96dB dynamic range.
One of the best explained videos on the subject. Actually, probably the clear best.
This is your second video from your channel I have watched. You've earned yourself a new subscriber.
6:40 for what it's worth an OCXO is still voltage-tuned (EFC) with a typical pulling range of something like 1-2e-7. The ovenized part is for short-term stability, not frequency adjustment. Essentially, quartz crystals have a characteristic curve of frequency change vs temperature change over temperature, which is very different for different "cuts" (the literal way the crystal is cut and polished). The standard is the AT cut, which is pretty much all quartz crystals you see everywhere. The AT cut crosses 0 Delta-Hz/Delta-K at around 25 °C. Which is just about perfect for normal ambient room temperatures. But the slope around that point is steep (relatively speaking, so let's say you're at 20 °C, you might be getting 5 ppm/K or so). Now the clever trick about OCXOs is that they use different cuts, often the SC cut. The SC cut has a completely different response to temperature changes; it crosses 0 at around 90 °C or so, and has a very gentle slope around this area. So OCXOs keep their crystal heated to a very _stable_ 85-95 °C, using a control loop. The attenuation of ambient temperature changes by the control loop colludes with the shallow slope of the dHz/dT to give extremely low tempcos.
The _expense_ of that is: SC cut crystals are much more expensive (tiny tiny tiny fraction of the market), OCXOs are a lot of hassle to build and even fast-stabilizing types will exhibit a pretty wild swing from turning on to becoming mildly stable (at least 15+ minutes), they have very high power consumption compared to normal oscillators (couple Watts steady-state vs. a few mW for a normal oscillator and maybe 50-100 mW for something fancy. (And some OCXOs tend to fail under repeated power cycling, like you'd do in a residential setting -- they're _absolutely not_ designed for that and _will not_ hold their specs when treated like that).
Most (all?) OCXOs have their own reference voltage which is typically ovenized as well. Sort of like bootstrapping, the reference has a tempco (selected for ~0 tempco at the elevated operating temperature, another reason these are expensive), so it would impact the temperature controller, so it's ovenized itself. It's of course stable, the tempcos of precision references are tiny to begin with.
Now the reason people went to this trouble is that for, mostly physical experiments, extremely accurate and stable clocks are a useful tool. A good OCXO achieves short-term stability better than 1e-12, that's one part per billion, or for the typical 10 MHz reference, that's less than 0.00001 Hz variation over a few seconds. (Jitter and stability are not the same thing). These are mechanically vibrating devices: Vibration creates low-level spurs. Most physical phenomena influence oscillator output, temperature is just a big one. Gravity is another; angling or rotating a quartz changes the frequency (much more than 1e-12). So for optimal performance you'd want to dampen vibrations of all frequencies. Sounds like something you could sell to audiophiles. Big fat granite rock on a bunch of thick steel rope springs.
Obviously, as you allude to at the end, the oscillator can't actually be used as a reference for anything (what's 10 MHz / 44.1 or 48 kHz?), so you're looking at a fractional-N PLL anyway, and at that point the quality of the output clock is liable to be swamped by the PLL's qualities (or lack thereof).
To be frank, a normal quartz is probably just dandy for audio, if the rest of the system design holds up. Audiophility seems to be especially liable to buzzword engineering instead of system-level engineering.
I only listen to suspended DACs on the top levels of skyscrapers that are heavy enough to stabilise the building in at least category 4 hurricanes.
OCXOs are fine and dandy, but frequency stability per se doesn't matter in audio, phase noise matters - OCXOs are for long term stability, phase noise is pulse-to-pulse stability, close but not related.
And even then it doesn't matter - a lowly jelly bean quartz crystal is like 4 or more orders of magnitude better than needed, you need to clock your DAC so badly, with jitter on the order of tens of nanoseconds for the artifacts to even arise from the noise floor, let alone be audible, but hey, audiophilistines know better and don't take anything more than femtoseconds.
Thanks!
This is super helpful. Also, I just bought a May because of you, gosh darnit. So thanks.
Danke!
@6:35 an oven controlled VCO is exactly what it says - it is temperature controlled to optimise clock waveform & jitter performance - it is NOT an “temperature controlled” crystal, that is voltage is still used to set the crystal into its designed resonant frequency, but temperature controlled conditions, stabilise & optimise the oscillator performance😀🤓
It controls the temperature to increase long term stability. Jitter is short term deviation and thus won't be improved with OCXO's.
Great video! Very informative and i learned something new
This is a great video! I just purchased the iFi iPurifier SPDIF 2, and I really enjoy it. It makes the optical audio on my PC so much better in terms of quality, and music sounds amazing. I can even play my songs on Tidal at 192 KHz. I haven't noticed much on my Xbox One, but I still use it to prevent jitter.
Thanks for the amazing review. Ethernet factor is huge, and it is ignored in many digital audio reviews. I did a $200 investment in my roon Server in the basement (6 core xeon with a noisy fan) and it changed the sound all over my house not only where I had the best roon endpoint ($4000 DYI).
Awesome channel by the way finally someone who can measure and has an open eye for the subjective part of this hobby.
Should be good when he does the blind test with ASR. He is going to pick different dacs out. Can’t wait !
I would love to see a video just flat out on time domain vs frequency domain and what each is used to look at
I thought by now any reasonable DAC would have it's own internal elastic buffer and just internally re-clock using it's own precision clocks. Paul Lesso wrote an AES paper about S/PDIF design 15 years ago called "A High Performance S/PDIF Receiver". A small buffer chip should be super cheap these days, and should be able to wipe out any jitter from a PC optical port. I get the impression audiophiles like to make simple problems hard.
They want excuses to buy more stuff while still claiming it is "minimal chain".
Buffer = delay = bad, I guess?
@@sanjacobs6261 Delay is bad if you are watching a movie or trying to mix tunes as a DJ, cuz the audio will arrive too late.
But if you only listen, it doesnt matter if the audio comes out a second later then it got send.
@@Wassermelonenbaum and it wouldn't even be a second, a few milliseconds would be more than sufficient.
@@gayusschwulius8490 For reclocking you usually use atleast 2 flipflops. This means that it's atleast 2 cycles of a clock worth of delay. The clock in this case has the frequency of the bitclock which is usually 44.1k*16bit*2 = 1411200bit/s. Which corresponds to 0.7microseconds of delay PER flipflop.
I'd say delay typically becomes audible in live content at about 15-25ms. You could legit use a thousand flipflops and still not hear the delay.
I recently added a Gustard U18 to convert USB to COAX between PC and DAC. The improvement is huge! I get an immersive 3D soundstage with relaxed and sweet sound. After that I tried direct USB connection once and couldn't bare with the "digital", dry, lean sound at all.
Your explaination makes perfect sense, noise and jitter matter most between source and DAC.
BTW even USB is not really bit-perfect, as USB audio uses isochronous protocol. Unlike bulk protocl, it doesn't have error correction.
Just a question - "...the Schiit EITR is a very affordable device..." I assume you are aware that this product was discontinued about a year ago and is only available second hand? (where it commands a huge premium over the original MSRP). FWIW, the EITR was based on Schiit's older USB implementation (called Gen5) and the EITR was discontinued when Unison was introduced. Also, I agree that the Pi2AES is outstanding... I have 3 of them.
Pretty amusing to hear tho
Eitrs sometimes pop up in Schitt's b-stock page for $99 plus shipping. Schiit discontinued Wyrd as well, so I guess you have a point unless someone wants to use spdif coax connection for older DACs like me.
Dude, this is the third time in a row you make a video about the stuff that is on my mind the bigger part of the day. your now my favorite Tuber. And about MQA I got your back all the way. 2 years ago I also put a lot of time (for me) in the new sensation MQA and concluded, it's kind of scam, and your video, organized all my chaotic leftover memories about that subject, and now it is as clear as a FLAC file. Respect.
Wouldn't mind seeing a video talking about power input with regards to audio equipment: the effect clean power has on equipment, usefulness of power conditioners and UPS's, and what to actually do if mains power isn't reliable with frequent brown-outs and general inconsistencies.
It sounds like some dacs would benefit a lot from a Pi2AES, where as others (the more expensive one's presumably) would benefit less if at all. I'd love to hear your recommendations for a good bang for buck combo with the Pi2AES if you end up doing a dedicated video on that device.
Outstanding review. Man, I wish you had a Denafrips DAC using the external clock function with the DDC for this round of review. I just ordered the Terminator-Plus DAC with the Gaia DDC and Kinki Studio EX-P7 with EX-B7's.
Why? NONE of these things are audible.
@@squidcaps4308 if you cant here a class A amp then you must be deaf
this was extremely enlightening and informative
Great presentation. In my system, the subjective sound quality is noticeably impacted by jitter. Yes, lower jitter improves the sound stage, and also bass response. The biggest impact that matters to me is just the sense of natural sound. Glare is removed. That little tiring digital edge melts away. And everything sounds real. I believe great sound comes with micro detail from which we sense more than hear stuff that we equate with reality and emotion.
So for me the sound of a DAC depends on:
1. the actual converter or chip,
2. Great power supplies for every consumer,
3. The output stage, and
4. The clock signal.
Lots of people don't hear it. They may listen to music that really doesn't need to be well reproduced and the rest of their system could easily mask these micro details. But if you hear it, there is no turning back.
For us noobs, might I suggest you spare a few moments to explain you graphs better and how one is better versus the other? I kind of think I understood so thanks for the great effort and video!
Soooo I'm somewhat new to audio. My understanding of those graphs is the more lines you see randomly sticking out from a pack, the worse it is. At 13:50 you can see a line sticking up on the left graph towards the left of that graph, then just before the big spike, theres a bunch of bumpy lines sticking up from most of the other lines. Neither of those are in the right graph. Those things are the noise/jitter.
(I COULD BE WRONG.)
I'm not into audio at all. Just came here from my recommendations. But I have done some electrical engineering. The plot has frequency on the x axis and loudness on the y axis. He's sending a 12kHz sounds wave across the line which causes the giant spike at 12k. Since it's supposed to be a pure 12k signal, everywhere else on the plot should technically be at negative infinity (no sound at all). A lot of the reason it isn't at negative infinity is because of random electrical noise or jitter. Jitter causes other frequencies to spike slightly because changing the timing between points in the wave changes the slope of the sound wave and thus the apparent frequency of that wave.
@@keco185you should make a video about jitter not golden sound.
You should check out the Cherry 130dB+ DAC DAC 3, which will be introduced this summer! The coax/optical SPDIF input is internally reclocked for jitter immunity. Several levels of power supply reconditioning. True balanced, DC coupled outputs, too (:
So many thanks for pi2aes. This seems to be a real diamond.
Very nicely done. Thank you very much 😎👍
Fantastic informative video
Thanks for this video and for the tip my curiosity gauge is right up there.
At first I also thought usb wouldn't matter, but I noticed an audible noise coming from the apple dongle on my laptop, specially when the fan was running louder and a cheap usb isolator from alliexpress was all it took to fix it in my case.
Very informative, thank you.
I have always intuitively done my best to stay away from pc in my audio chain as ive always supected id have electrical noise and jitter issues. Having recently started mixing and recording on pc though, i'm in need of paying attention!
I recently switched my hifi to a Naim Atom , so hopefully my dac is being supplied with an optimal signal!
It'll be interesting to know if a mid-range DAC with an I2S input, like the Topping DX7 Pro, could benefit from all of this.
Well, I'll find out soon. I'm waiting for my Pi2AES Pro Shield to arrive. I'll tell you this much, the Topping DX7 Pro benefits from an op-amp upgrade. I installed a full set of SparkOS units, takes the unit up a few notches.
It helps a LOT on my Topping D70s!
And the verdict is... We have a winner! I use my DX7Pro as a DAC-Pre straight into my amp. So my take is the Pi2AES on Rpi4 running Ropieee via I2S elevates the sound quality by a decent margin. Overall the sound is even more coherent, soundstage width and depth is wider and deeper, the sound has a bit more body to it if that makes any sense, bass is more detailed, everything seems a bit more detailed, you hear more into the finer details of the music, transients and such, better decay and layering of it in more complex passages. I'd be saying these things with the Pi2AES regarding the previous upgrade, but this brings out even more. I wasn't expecting that at all because I thought I was hitting limit with what I already had. Not bad for $350 (I bought a FLiRC dongle as well).
Fun fact, a 10MHz reference clock is also used for things like phase syncing lab equipment.
Really interested to hear your thoughts on the May Holo. I ended up going with an Yggy GS for only like a 1/3 or so the price, along with a Pi2AES. I still wonder how it'd compare to the Holo. I guess one perk of the Yggy is that it's upgradable, and Schiit is already working on a new analog board for it.
Looking forward to your review!
Excellent explanation
Another thought. I loved this video because "USB noise" was one of the topics I was wondering about and I was also among the ones thinking bits are bits in a digital signal and audible noise cannot be generated on a USB connection. And you almost convinced me when mentioning the audible noise from a PC. We all know that, that noise is with us since the 90's. You can hear the mouse movement, GPU working, etc. But then I realised it's a completely different scenario, that is with the PC's builtin soundcard and a direct (jack) connection to the speakers or headphones. As soon as you have an external USB sound device, noise is gone. So I'm afraid I'm still sceptic about USB noise :) (jitter is another thing I understand that)
We need a hidden camera inside Schiit HQ when they are naming their products. They know.
I burst out laughing when he said the Shit Eater 🤣
I switched from using my gaming pc as a roon endpoint to using an allo digione signature (similar to the pi2aes from my understanding). I wasn't hearing anything obvious via the usb audio on my pc but when I switched to the RPI allo streamer the signal sounded quieter and cleaner but it definitely wasn't a huge difference. Good video though I've been really enjoying all the content on this channel
Same here. its not a huge difference but its enough for me to appreciate the digione.
I think these problems should have been solved in the DAC. The whole reason I am considering buying an external DAC is to get rid of electrical noise. Cue the "YOU HAD ONE JOB!" meme.
Considering that analog LPs are still considered to be the ultimate experience by many high-end enthusiasts, a 60db noise floor is not a problem and a 110db one even less.
@@dingdong2103 you’re misunderstanding the problem. The issue isn’t noise floor: it’s electronic noise and jitter which are a *completely* different thing.
@@GodfreyMann What is 'electronic noise' if not the thing that rises the noise floor? And jitter is not a practical problem in reality, it's just marketing jargon: hifigo.com/blogs/guide/what-is-digital-jitter-and-how-to-avoid-jitter-dac-101-part-3
@@GodfreyMann Vinyl has wow and flutter which are the analog equivalents of time domain modulation. You won't get rid of them by going analog.
@@dingdong2103 you’re correct that ‘electronic noise’ on the analogue side rises the noise floor, but the discussion is about ‘electronic noise’ on the digital side which gives rise to jitter. It’s not a marketing jargon - the clocks are voltage sensitive components and ‘electronic noise’ can cause fluctuations in voltages in the circuit. Although tiny this can affect the quality of the square wave signal and therefore how the bits are interpreted....it can change the bits and timing.
2 years later. I am now using a pi2aes 2.0, with a 850€ linear PSU that powers both the board and the Chord Qtest, with files stored on the SD card, optical out, with real glass cable, and music sounds more liquid, it has more shape somehow.
I did a test: same setup, same files, external usb ssd storage vs internal storage on sd card, and turns out, it does sound a little bit better when there is no usb in the chain. Same dac, same input, same output on pi2aes, the only usb was the usb storage and impacted the music. I don’t know why, but from now on there will be no usb in my system
Thanks for the awesome video! I wish that at some point you'll be able to make a video about AoIP with devices that support Dante or Ravenna. I'm really interested to see if they can bring any noticeable improvement to the sound. I've been eyeing the Merging Anubis for quite some time now.
T, The SMSL DO200 is a dream to own. Best, D.
I do not often see videos which are pure and utter nonsense. But this one is. The spdif and the usb interfaces have error detection integrated. They will not deliver an invalid data stream to the DAC. Noise on a data line can cause problems but usually you recognize that by drops in the audio presentation or by connectivity rejection (no sound - full stop) and not by noise on the analogue side. Since the cable length and quality is good these days, you usually have however not problems or drops.
Thanks for this review.
WHAT! A SHITT EATER! Sorry, I had to do it! You knew it was coming. Thank you for all your AMAZING REVIEWS!!!
Designing PLL is an art...and this is the -secret- to proper digital audio reproduction.
If at all possible, you should beg, borrow or steal an EtherREGEN along with a quality external master clock, such as an AfterDark Giesemann or a Mutec. I've added both to my Pi2AES along with quality LPSs. My listening experience has been transformed for the better. I've never heard such 3-dimensionality, fluidity, realistic dynamics, sweet resonant decays, etc.
Love the little fun fact about the creator of USB audio thinking it sounds like doo-doo haha
I have a pi2AES which I run into a RME ADI-2 Pro and then to my Genelec speakers. The sound is glorious. I first used SPIDF but then switched to AES which to me sounded better. I do not know why it sounds better, but it does. More bass control and overall cleaner sound. I have not experimented with i2S yet. My conclusion is that the pi2AES can be used in a chain to produce the highest quality audio. As a side note, the RME ADI-2 Pro is a match made in heaven for Genelec speakers as it can be used for AES or XLR output, thus making it possible to send both digital and analog to the speakers. For example I stream Qobuz and play vinyl from my TT through it. This is perfect for me because I love both digital and analog and can’t live without both.
Ok, nice but how do you feed the pi2AES, since it doesn`t have USB in ?
@@DJURBANBG Ethernet cable. It’s plugged directly into my LAN. Then it takes over as my Roon music server and I output that via AES to my DAC.
@@hemantishwaran5741 thanks !
a phase-locked loop is literally necessary to keep multiple signals in sync, I don't think I understand how something is supposed to properly interpret a digital signal it's receiving without one, the clocks would keep going in and out of sync.
Audiophiles: Measurements don't matter, all that matters is what you hear!!
Also audiophiles: This device has a noise floor at -144 dB vs that other one that has it at -120 dB and costs 1/10, we know that it is a proven fact no human being can perceive the difference but just look at these GRAPHS!!!
In fact, any unwanted noise at -144db can still affect the sound quality when such frequencies are played, its still a delta to the "perfect" soundwave.
@@harrison00xXx of course it will affect the sound. But you will not hear the difference because you are not Superman
Nissim Trifonov ok thats also a legit argument
That's us, yes!
My thoughts exactly. There's no way in hell a human can hear a signal that is 120 dB lower than any other simultaneous sound (that is a millionth of a millionth of the energy in the main signal), even in the ideal condition when the main signal is a pure tone that does not overlap in frequency with the jitter-generated harmonics. Maybe a super low noise no-jitter-at-all DAC is useful in lab environments, but as a music consumer...
Since you are basically the only one to do this kind of things on RUclips (and thanks a lot to do so!), I think you should take more time to explain what you present. This is some really really interesting and in-depth information, it deserves longer videos, or more videos.
Something I don't understand: why should we need a device that clean the signal from the source to send it to the DAC if you can have that in the DAC itself? Shouldn't it be a requirement for any DAC to basically reclock what's coming from the USB and isolate from the voltage coming from the USB?
Looking forward to your SOtM review. I am really interested in it. I would like to know how it compares to Pi2aes.
SoTM is very good and pretty much a connect-and-forget device.
Great video. But the giant question it leaves unanswered is whether the little no-ISB Pi device that beats the Hermes is uniquely good at destroying jitter, or whether *any* transport without USB (such as my MiniDSP SHD Studio) would yield similar benefits, simply from avoiding USB.
Would be interesting to see how a Nobsound Douk Audio U2 USB Converter compares.
So what about Pi2AES SPDIF to hermes and converted to I2s, would it be even better?
Love to see the Comments. Nice video, keep up the good work.
I'm considering buying a sub $1000 DDC and have 4 candidates, but am open to other suggestions.
Candidates are
1) Gustard U18
2) Singxer SU-6
3) Denafrips Iris
4) Topping U90
In stock form, which puts out the cleanest IIS signal?
What about including the Matrix Audio SPDIF 2?
IRIS is the best sounding in that list. I just upgraded mine to a Gaia And that's better still.
All of them sound exactly the same, but I went with a Topping d90se. I use its USB input, because it is the most convenient option and handles any sampling rate.
There’s no benefit to IIS, despite the marketing mistruths perpetuated by PS Audio’s Paul McGowan-who promotes that tech because it was required under his clandestine agreement with Sony to circumvent their DRM rights for outputting an unadulterated digital signal from SACD players to his own proprietary DACs. Don’t fall for this ruse, it’s an unjustified expense that has no impact on “sound quality”. Digital audio is not sound, and no matter what transport you use, the same signal is passed to the DAC and the resultant outcome is indistinguishable to the ear.
Two questions about the Pi2aes.
1- How does one change inputs on the Pi2aes? I think they all run continuously and have to be changed by hand.
2- How is the sound from an external/remote hard-drive into the Pi2aes via USB,,, then output to a DAC of choice? Thanks ahead for your help.
Got my Pi2AES yesterday and am so impressed with the SQ compared to a regular Pi3 on USB. AES sounded much better but going to i2s was a night and day change. Thanks so much for bringing this to my attention Golden!
I got a pi2 and am using it with an Allo USBridge Signature. Still can’t it to work. Any pointers? I am using the same SMPS power supply as recommended by Mike Kelly and the USBridge is powered via the PI2AES. Thanks.
What PSU are you using it with? Linear PSU? Thanks
@@Shoaibexpert just the standard meanwell
Hi @GoldenSound! Great reviews, as always. But unfortunately looks like the Pi2AES project is no longer available and they have a closed and more expensive version. So, I was wondering, with the release of the new Topping M50 which is kind of a weak streamer, can it be used as a cheap USB to IIS converter? If so, could you measure that to see how clean the signal is? Thanks!!!
Thank you for this excellent video! Can you please create a video that shows how to change the power input on the PI2AES to 5V? Thank you! Marcin
Your channel is a fucking goldmine holy shit!
Great video! I hope you know you just opened a can worms with the ASR number crunching crowd.
Considering he already did the number crunching, I should hope they wouldn't have much to screech about - though I know that's probably wishful thinking. But so long as it just means more data, and not just ridiculing, then all is well and good.
>'Not claiming any level of audibility'
>Makes a whole video on why ddc's make an audible difference...
Pick one.
I think he's just nerding out over the numbers. Companies like Holo are laughing all the way to the bank with people like him.
Have the iGalvanic3.0 and iPurifier3 in my chain with iDSD Black Label DAC/AMP. Makes a big difference!
iPurifer3 didn't do anything at all for me, but iGalvanic3.0 was a massive help for Hugo2. When I sold hugo2, I thought it was only right to sell it with the iGalvanic3.0 because its pretty much a requirement.
@@-havoc-tzu-3197 Hard to believe the Hugo2 doesn't have galvanic isolation isn't it? The Hugo TT has it. I would bet anything that the Hugo3 has galvanic isolation because the competition is stiff and the Hugo is high priced.
@@shawnpwatsons1 agreed, I switched to Qutest after that which has galvanic isolation, later sold that too and upgraded. Qutest is a decent dac.
i used a Schiit Eitr for good while on windows, it did seem to make a small difference, but nowdays i mainly use it as a way of having a 100% Linux compatible device, allowing me to use any DAC on my Linux desktop PC. Another hidden benefit.......of a Schiit Eitr
Great Video, I agree with using I2S, I was wondering if you have heard of IAN Canada. I found these HATS for the PI and I was thinking about using these two devices with a PI 4. 1. IAN CANADA FIFOPI Q3 ULTIMATE FIFO Reclocker Module PCM 32bit 768kHz DSD1024 DoP 2. IAN CANADA HDMIpi MKII Transmitter I2S / DSD / DoP to HDMI
Very helpful
Golden sound I'm sure you have explained more times than you care to remember so please from an American fan, what are those card or pictures on the wall I've been wanting to ask for a year or so. Thanks for the videos!
-120db noise from jitter is NOT EXTEREMELY audible at all!!, in fact your amp background hiss more audible
more snake oil audiophoolery, would like you see your "golden ears" detect jitter in a blind test where you really have no clue
which is which
You should do a vid on MQA
I'd like to, but unfortunately MQA has quite deliberately made it VERY difficult to obtain any equivalent files to test. And no MQA full-unfold device is allowed to have digital output.
The only actual comparison i've seen where there was an MQA and non-MQA file that were the same was done by stereophile, and it showed all sorts of problems.
This comment aged well 👍🏼
I hope you could review the Xduoo DT-01.. Very interesting type of device..
Given that the Pi is using the I2s, which 'DAC' hats give the best analogue output? Ive noticed that some 'DAC hats' don't seem to include a crystal, so presumably rely on that of the Pi, so I'd guess those could be ruled out, as the Pi crystal is not a good match for the function required. I recently bought an audioinjector Ultra, so I'm interested in how the different makers implenentations compare.
Also don't forget that some connections like Toslink(optical) or coax S/PDIF are bandwith limited, no DSD upsampling on these if that's your thing.
The pi2AES with I2s output is no longer available. The manufacturer now only offers I2s output with their $600 Mercury streamer. However, production has halted due to a supplier issue with the case manufacturer. According to the website, the PCB board for the Mercury may have to be fully reconfigured to fit another case. Translated, it might be quite some time before the Mercury Streamer is back in production. At this point, it looks like I'll be forced to look at other options for the foreseeable future.
Would it be possible to expand this discussion and include the Allo DigiOne Signature in the comparison?
Question about i^2s Guido Tent ones told me that i^2s is only for internal dac use not for longer distances. And i can understand that because clock signal should as short as possible especially digital clock signals.
may be its better just to use DAC with connected external master clock? for example TEAC Ud-505 and CG-10M?
I gave up explaining how jitter can be CRITICAL to your music a while back. Took me a while and a lot of experiments to learn a lot of. Spending time trying to pass that knowledge to people who are not receptive is a horrible way to waste your time.
I would recommend giving Lavry Engineering DA10 and similiar DAC a try with their "crystal lock" feature enabled... See how wrong engineering can go when they confused clock drift with jitter. Their tight PLL sounds great tho.
Another thing i find suspect is Analog Device's ASRC that claims to eliminate jitter by running signal through SRC, but i suspect it is just blending the good and bad part of the signal into a inseparable mixture.
Why would you have to explain something that's, according to this video, "extremely audible?" I've never heard it, and have been an avid listener for over 40 years. What DAC does one buy to hear this phenomenon? What does it sound like?
@@Grassy_Gnoll it is extremely auditable if you know what to look for. If you are facing a crowd of "if it is bit perfect, then there's no difference" horde, then you will see what i mean.
To me, higher jitter often results in blurred imaging, worse soundstage.
To try it on the cheap... Just get a player and DAC with optical and coax. Then compare them. Do remember to use cheap plastic optical cable and cheap coax.
@@Grassy_Gnoll all else being equal, bigger soundstage, lower jitter.
Interesting vid, I recently tested 3 different DACS against each other I compared the AudioEngine D1, the DAC built into my Yamaha A-S301, and the DAC in a $110 CD player from 2001 and after volume matching I couldn't tell a difference. I've yet to try a DDC but if my ears can't tell a difference between the DACs I doubt a DDC would make much of a difference for me but I would be willing to try one out.
Yeah, I'll be honest... It's super super subtle to me. Only difference I've ever been able to distinguish is the difference between multibit or R2R and delta sigma. But, even between completely different topologies, it's very subtle to me. It becomes more noticeable with high end headphones compared to my 2 channel setup.
External DACs are more or less a waste of money. Not even talking about DDCs, thats the snake oil people talk about, even external DACs are more of snake oil than real advantages.
In the days where only proper DACs were in extremely Hi End gear which was probably not even aviable for the regular customers, an external, better DAC than the one in your anno 1990 100$ JVC CD Player would make a difference.
But i realized how useless "better" DACs are as i switched from a cheap 90$ DVD Player (i think 96kHz 24Bit) to a proper CD Player for 600$ (192kHz 24Bit), fitting to the 700$ Stereo Amp.
There was absolutely NO DIFFERENCE, except for the removed DVD drive noise whereas the Cambridge Azur 650C was totally quiet.
My AVR has even a 384kHz 32Bit DAC... and it somehow sounds the same as the cheap DVD or expensive CD player.
The sound difference between my 700$ AVR and my 700$ Cambridge Audio Amp is also very negliable. Where the AVR (175W/channel advertised!) begin to struggle a bit with bass at higher volumes, the Cambridge 650A ("only" 75W @8 Ohm) doesnt sound that stressed on my IQ 90s.
But at any regular volume AVR aswell as Stereo Amp (both feeded completely different, Stereo Amp analoge over CD player, AVR with a bit-identical FLAC file over USB Stick) they are more or less identical, you only hear a bit of difference when it comes to the "cambridge sound", but its not far off from the "linear" sounding AVR and is pretty easy EQed even with the 2 controls on the Stereo Amp for Treble/Bass.
@@harrison00xXx I'd generally agree with you, but with one caveat. There does appear a benefit with desktop/gaming PC setups. This is largely just due to computers being very noisy (electrically) inside and there can be a benefit of moving your DAC and amp outside of the chassis, and on their own power supplies.
asplmn this isnt a thing with digital connections anyway
Like many of us i use my PC directly to my receiver amp ( Yamaha RX-A3070 with ESS pro chips ) via an HDMI cable connect to my graphic card.... HOW IS THE JITTER with this connection ?? Thank for the answer ( i hope )🙏
Can you please do a review of the blue sound Node? But review and do measurements for how well it gets rid of noise.
Thank you !
Very informative review, I like the measurement addition!
You keep mentioning the Pi2AES price as the full price of the streamer but adding a Raspberry Pi and a good power supply is a must, so the total price will be around 600 $.
Great job anyway!
Most expensive Pi 4 is $75. Pi2AES is $149. That's $224. So you're saying you need a $376 power supply to run this. Must be one hell of a PSU!
Prices have gone up over double since this video
Would you recomment using Denafrips HERMES with RME adi 2 pro FS, will it make a difference ? THANKS
Hello! First video of yours I’ve seen. It’s amazing, I’ve watched it like 3 times already haha. I bought an eitr, and as ashamed of this as I am, i must know.... how do i connect it to my modi multibit/littledot mk2? I need to know what cable to order and what not lol. Sorry for the inconvenience. But that aside, I love your content! Keep it up!!!
P.s. everything’s connected to my gaming pc.
You mentioned an upcoming SoTM SMS200 Ultra video? Any time soon?????
Excellent video & best explanation yet of how noise and jitter can affect DACs. Also your hint at a possible review/analysis of the SoTM sMS-200 Ultra is most intriguing! I use this unit together with the SoTM tx-USB reclocker (which I guess is a type of DDC). I find the tx-USB definitely improves sound quality, but I would love to know how it does it!
Just some "personal" perspective from my experience as a software developer for real time embedded systems and deep interest in electronics. (not professional in electronic tho)
I'll only talk about those three topic you start with, over usb, some applies to other protocol, but I don't know them in detail and I'm not going to read pages of specifications for a youtube comment.
The TLDR is just to say, normally, you shouldn't have to care about those three issue in a "just enough" designed modern usb device, even more one who cares about those well known issues and tells customer they're some kind of high end devices, dac, or anything else.
Data integrity
Being digital doesn't mean you can't have data corruption, for instance if you communicate over serial without any specific protocol, or a protocol that doesn't integrate some form of data integrity mechanism, you can have data corruption and no fix for it. That being said, the usb protocol specs specify a mechanism for that. A CRC is sent to ensure the integrity of the received data and if the check failed, the data are re-transmited. (to make it shortà That means, for any device compliant with the usb specifications, it's not possible, on the transport, to corrupt data and not notice it and fix the situation, it's by design. So no need for extra middle man. So, being "bit perfect" and "not altering data", I don't understand the take. A usb hub does it, nothing special. But I don't understand what the point is, so I just mention, to anyone worried about this, usb, by itself, protect against data corruption on the line. If you worry about that, use usb, and you'll be fine, by design.
Electrical noise
You mention using optical link as solution to decouple two electronic circuit. That's way more on point than it seems and is in fact quite... "original', that any "high end" dac doesn't do it. The solution is called optocouplers, it's used everywhere, even in low end electronic, it's cheap, and it does just that perfectly.
The fact you have to introduce middle man to solve that issue for expensive units is a shame.
Also, optocouplers main interest, outside of the noise, is mostly security concerns, and it should be used in scenario with interconnection to any other unknown circuit. For instance, you certainly don't want to blow your expensive device by linking it to a faulty one that might send huge spikes to yours. If you design something like a dac, being proactive about those situations shouldn't even be questioned, even more for expensive equipment where you should do anything possible to protect it from the outside world at the benefit of the customer and your brand image. At the same time, solving noise propagation issue through data or ground lines.
Imho, any dac, even low end, should do it. And to be honest, it's a solution found in a lot of studio equipment, and you don't have to go high end for it.
So blame the dac manufacturer for not doing their job.
Jitter
It was a real issue for instance with spdif and the reason was... mostly a "bad" design we don't rely on anymore.
Also there's different type of jitter with different sources and different solutions.
The way you represent what jitter introduce as a problem doesn't match the general design of todays dac, dac aren't clocked by the input protocol clock anymore (and if they are, it's on their own and they decided to go the wrong way, maybe buy something else... who easily solve the problem)
But, usb has a well known workaround.
The main solution to solve almost all jitter issues is to use asynchronous mode. Jitter will be small to the point it's humanly not noticeable anymore and, if you really want to solve that picosecond problem, just introducing a buffer or related, audiophile dacs, imho, can do it, I wouldn't mind it a all as a client, but might be wrong.
The fact this problem is still present really makes me wonder: are they using 20 years old design that just "get better over time" but don't do re-design that would solve the issues ?
Once again, it's the dac manufacturer fault, it's a design issue, it's a known problem, with known or even integrated solutions, if problem persist, maybe go back to your R&D and fix it before requesting any substantial amount of money from your clients.
Paying a middle man device to fix those three issues, at least over usb, until you are like using an old dac you really love, shouldn't even be a thing.
Shouldn't even be mentioned on modern hardware, not even as a selling point.
It's something people sometimes say about dac to say every dac is the same, I don't have opinion I want to share on that topic, but for those three issue, if you have a modern dac that features them, I would consider them "by design broken dac", any mention of those issue on something I plan to buy would make me totally never ever go toward that manufacturer again. To me, those issue, on new hardware, shows a total lack of dedications and I really hope I get this video wrong and it's about fixing old gear issues. At least for usb, don't know for the rest, but wouldn't be surprised there's solutions too and modern hardware shouldn't rely on middle man.
you seem to know a lot about the electronics, can you explain the concept of electrical noise? i get it on analog signals where stuff like magnetic fields and whatnot can influence the voltage on the wire, which causes the speaker to produce the wrong amount of air pressure, but for a digital signal you'd have to flip a bit, no? wouldn't this directly contradict the claim of the data being bit perfect?
@@LastExceed Hi, I'm not sure what you want to know, so let me know if it's not what you meant to understand :)
Technically, digital signal is analog signal. Electricity is always analog and will always be. Digital signal is a way we interpret the analog signal. For instance, let's say you have a 5v digital signal, everything above 2v will be considered a 1 and below 2v considered a 0. So at a specific time a signal can be 3.5v or 4v, so they"re not the same electrically, but they'll be both considered a 1. (To simplify) Now if you have any effect strong enough, you can flip a 1 to 0 and vice versa. Also, it's time related, there's different clocking strategy and ways to define what a bit is and how is value is "read" from that analog signal, but depending on that, you can have a really unstable signal that might be misinterpreted.
But as I said, there's integrity mechanism most of the time, especially nowadays, to detect those corruption and re-emit or fix the data. Or identify something went wrong and apply a strategy, that can sometimes be invalidate everything and consider everyhing corrupted, so basically untrustworthy garbage, sometimes it's more about being sure not unwanted change occurs, than preserving the "user experience" and trying to find a fix, so maybe shutdown and lock the machine until human intervention, I mean, lots of possible scenario depending on the context. Let's say it's a strategic infrastructure machine, do you prefer to take the risk to autofix a situation that might have been caused willingly by a bad actor that can use that as an attack vector, or realize something went wrong, lock everything up, pause everything and wait until investigation ?
Bit perect digital signal is, in a way, impossible by itself. But by implementing control and error fixing mechanism, it can be done on a specific scope. Or at leat, be done consistently and reliably enough, that we consider it bit perfect. (doesn't mean it can't maybe fail once every billion year, or could be bypassed willingly by an actor, or other stuffs, depends on your context, stakes and how much you're willing to put in place to protects against that problem)
For instance let's say you have a voting machine, RAM corruption in that voting machine would be a catastrophy, so we use ECC memory that protects against that for intsance. In my country, we had, almost 10 years ago, a voting machine with a strange bug during elections, long story short, it was a cosmic ray and the machine didn't had ECC memory. (might sound like a crazy conspiracy anti science stuff, but nah, cosmic ray issues with electronics are a well known and managed problem, I'll let you google for more details, you'll find research paper about that starting from decades ago by trust worthy actors like NASA or renowned universities and a lot of mitigation strategy since then) The register containing one of the candidate vote count flipped, and so his or her results. When at a scale large enough and with stakes high enough, you start to see and have to consider really really strange and "rare" effects. Also, ECC memory doesn't protects against possible corruption done before or after being written or read from ram. But yeah, there's mitigation solution.
Being bit perfect is about the control and fixing mechanism you implement.
Let me know if I answered your question.
@@EmmanuelIstace Oh I've heard about the voting machine thing!
I'm trying to understand what is said at 4:39. Are they suggesting that the noise flips bits during transmission? Even if we ignore the USB built-in data integrity stuff, how would a special device at the end of the wire change anything about that when the data it receives is already corrupted?
I think I'm horribly misunderstanding what is even being said, please show no mercy in correcting me
@@LastExceed No worries, I'll never be harsh with someone trying to genuinly understand something, no worries :)
edit: before hand, I use interchangably CRC and hash in this message, consider the term the same in the context of this example. They're not, but here it doesn't matter, as our CRC is hash based.
So, I listened quickly from that timestamp, not sure what they're refering to, might be relvant, or totally not. And I remember I had, let's say really mixed feeling about this video few months ago, so not going to rewatch it.
But about your question, how can the receiving device detect and fix data corruption. In my case, it's about corruption on the link, so during transmission on the wire. If the Source machine send data that would qualify as "corrupted" even before sending it, there's no way to detect and fix those. (I mean, there are, but it's way too advanced for what we're talking about here and for scenarios out of context for here)
But we can make sure that what was sent and what was received are "bit perfect".
Data are sent by packets most of the time. (and also batch of packets, but not relevant).
So small chunk of data. Let's say you want to send the data "hello". And let's imagine you use a protocol that send each letter as a packet. You'll send 5 packets, each containing a letter.
For each packets, you'll have meta data, so informations about those data, for the purpose of the example, a sequence number.
In those metadata you can include things like a CRC. A CRC is a hash of the data. Basically a hash is a computer generated mathematical representation of another data. It can be viewed as a "summarized" version of the data. For instance you can have a 64bit hash that correspond to a 1Gb file. Those hash are totally deterministic, that means if you put the same data in, you'll always get the same hash out. You can compute a thousand time the hash of that file, as long as the file didn't changed and you use the same hash method, you'll always get the same output hash.
Also, it's not encryption, it's hashing, that means you "normally" can't compute the data back from the hash. (but you can compute the hash of every possible data and consult that table afterwards to know what data might be represented by a specific hash, that's called a rainbow table, but out of topic)
So to detect data corruption, we'll use those CRC.
Back to our simple improvised protocol.
The source machine, for each packet, will include a CRC of the data payload.
The client machine, when receiving each machine, will compute the CRC of the payload using the same algorithm and compare it to the CRC from the associated packet. If they're not the same, the client machine knows there's a problem, either the CRC was corrupted, or the data was corrupted, but something's wrong.
In this case, to fix the situation, the client will notify the sender and say something went wrong with that particular packet, and request it again.
In practice now :
The source machine will send 5 packets, I'll write the sequence number, the data and the CRC. Each line is a packet.
0 - "h" - fs1f6s8
1 - "e" - wcx87w
2 - "l" - w354c8er
3 - "l" - w354c8er
4 - "o" - k468h38
As you see, packet #2 and #3 have the same hash, as the payload is the same.
Those packets are then sent, let's take a look at what was received :
0 - "h" - fs1f6s8
1 - "e" - wcx87w
2 - "_" - w354c8er
3 - "l" - w354c8er
4 - "o" - k468h38
As a human, we see quickly a problem, the packet #2 have corrupted data. But a machine is not intelligent (not even "AI", but not the topic) and have no idea about what the sender meant to be sent. So the target machine will compute the CRC for every payload. Here's the result:
0 - "h" - fs1f6s8
1 - "e" - wcx87w
2 - "_" - yui23y1
3 - "l" - w354c8er
4 - "o" - k468h38
We see the hash of the packet #2 have now also change, as it reflects the payload which is different.
That means now a machine can detect the error. It will compare the CRC received, from the CRC computed, and detect that the CRC for packet #2 don't match.
So it will request again that packet unless it receives the correct information.
Now, this is a simplified version of all this, also CRC can be about more than the payload, or a part of the payload, or part of the meta, there's what we call "hash collision" (that's when two different source data produce the same hash) different strategies than using CRC, etc, there's also a lot of other situations that doesn't directly imply those kind of corruptions, but I hope you now have an idea of how data corruption on the line can be dealt with. In most "usual" scenarios, strategies like the one here, or similar or derivated are used, often way more subtle, refined, etc, but the core idea and principle is often this one.
If I wasn't clear on something, let me know ^^
(also, I read back the message, sorry for the grammar etc, I'm not a native speaker nor living in an english speaking country)
(edit 2, haha I realized I copy/pasted wrong my list of packet and introduced a second error on the last packet, haha, that's why machine, even tho they're dumb, are more reliable than humans, a machine would have detected that packet too haha)
@@LastExceed I came back to check if you had another question and realized I maybe answered the wrong question, if the question was about noise through ground lines, then yes, datas can be bit perfect, but noise coming from ground lines can interfere. It can introduce inconsistencies in the DAC device once in the analog domain if there's electrical coupling. Long story short, for instance on a lot of amp and depending on the design (the amp inside the DAC) if there's potential difference inconsistencies with ground the output will have inconsistiencies too. That's what I adress in the point two of my original message. In that case decoupling the two device will solve that issue, and that's maybe what one of the box presented is doing. But as said, that's a huge design flaw for anything that would consider itself serious on the market and even more a security concern than a performance concern.
I hope at some point in the future you get to do a comparison video of the CTH and Vali 2+.
I appreciate that you've presented actual measurements, but this seems like a solution in search of a problem. These DDCs seem to be all about second guessing the internal design of the DAC you paid good money for. A high quality DAC should already have its own isolated power supply and isolate any electrical noise coming in from the USB. It should also already have a high quality clock which is the only place you need to worry about jitter. If you just use the USB input, there is no jitter from external sources so no need to spend more money to try and minimize it. Just buy a good DAC in the first place.
While it's true that you are essentially "secnd guessing the internal design of the DAC" and that the DAC already SHOULD have this implemented, it doesn't mean that it actually has.
A good clock oscillator requires a good crystal (which is just expensive to manufacture/buy). Attenuating input jitter can only be done IF you have a proper local oscillator. You are making a compromise between how much the external clock matters and how much the local clock matters. For USB DAC's this is irrelevant because USB is asynchronous.
For power supplies I completely agree though. You won't make the power supply much better than the way it's implementend internally.
@@hidjedewitje with asynchronous audio dacs you also have the option of putting in slave or master mode when using an external master clock - despite all the advantages of async usb, the search for cutting edge performance continues... Unless you're Bruno Putzeys, design a good one 10 yeara ago & its still cutting edge....
@@BogdanWeiss Bruno's stuff is indeed very fascinating. His work at Grimm is also quite fascinating! The AD1 (now discontinued) is still the best AD converter when it comes to linearity and jitter performance 20 years later!
The question is, do we really need such performance?
@@hidjedewitje Performance just like intelligence is a double edged sword - once seen or heard, it's difficult to un-hear it - I didn't realiae that the grimm adc has been unchallenged for this long - truly impressive
@@BogdanWeiss While true, there's definitely diminishing returns.
The difference between a laptop's on board audio to a standalone DAC with THD+n of -115dB is pretty audible. There are also DAC's with THD+n under -120dB. I doubt anyone can hear those differences, despite the performance being objectively better.
There's also people who stare themselves blind on numbers regardless of looking at the weakest link in the chain. They buy DAC's of $1000, but leave the room untreated (resulting in dips of 30dB in the frequency response!!!).
There sure is a place for state of the art products (like the Grimm's or the Mola Mola's), but they should also be placed in a state of the art environment to be actually usefull.
It matters if your DAC is not able to remove interface jitter and electrical noise on the incoming signal. Why not simply get a properly designed and built DAC? This is the same as buying power conditioners - you only need conditioning if you have an audio component that can’t deal with regular noisy power. You also should use an asynchronous DAC - which decouples master clock from the crystal clock in the DAC. Basically you only need extra stuff if the components you have are sub-optimal to begin with.
you really should watch the video before commenting.
@@DueM I did watch. I was summarizing the implications of the video in terms of “so what should one do in practice”. There is a very old lesson here that I learned 20+ years ago. Summary: Buy well built and well designed components and connect them properly and you won’t have a bunch of issues like jitter, ground loops, hum etc. The sad thing is that well built and well designed components are extremely rare and most folks don’t get rid of poor performing boat anchors because they are attached to their toys and, as a consequence, they end up with loads of “band-aids”.
@@jeremyhughes6485 it's not the dac that's the problem it's the source, that's the point of this video. There's plenty of dacs with good internal re clockers, funnily enough even they will benefit from an external ddc in the input chain. Whether or not it's actually audible is another story all together so I leave the donglemania for the pedants.
@@DueM Sorry. I simply have to disagree. There is always some degree of jitter on every digital input (it is an interface issue) - so the better DACs are asynchronous (they ignore the input clock jitter) and in that sense you could stick anything in between the digital source and a well designed asynchronous DAC and it would not change anything as long as the signal remained bit perfect. An asynchronous DAC can be fed a very high jitter signal and remain completely unaffected. (Benchmark demonstrated this a long time ago with their DAC1)
@@jeremyhughes6485 jitter reduction is built into most dac chips, async is ok but wasapi is more stable these days and you're not using proprietary drivers which are notoriously buggy. benchmark dacs use sabre chips which have inbuilt jitter reduction anyway and are pretty much identical to a hundred other dacs using the same topology. we can agree to disagree but as long as pcs are the source theres always going to be lots of noise, hence these devices.
Hi..appreciate if you can make Pi2AES assemble video and software installation needed to be used via smartphone
Yes can you do a session on the Pi2AES
Damn! Your head mounted cam really triggers my vertigo.