EEVblog
HTML-код
- Опубликовано: 27 авг 2024
- How to measure switching noise on a power supply properly, because Dave completely goofed it in the Riden RD6006 PSU review video.
s.click.aliexpr...
All about coaxial cable impedance matching and compensation.
Mysteries of x1 Oscilloscope Probes Revealed: • EEVblog #453 - Mysteri...
PSU Ripple & Noise Measurement: • EEVblog #594 - How To ...
EEVblog on LBRY: lbry.tv/@eevbl...
Forum: www.eevblog.co...
EEVblog Main Web Site: www.eevblog.com
The 2nd EEVblog Channel: / eevblog2
Support the EEVblog through Patreon!
/ eevblog
AliExpress Affiliate: s.click.aliexpr...
Buy anything through that link and Dave gets a commission at no cost to you.
Donate With Bitcoin & Other Crypto Currencies!
www.eevblog.co...
T-Shirts: teespring.com/s...
Way to turn lemons into a lemonade video, Dave.
This is why Dave channel(s) are so good. When he makes a goof or oversight or whatever you want to call it. He makes ANOTHER and sometimes even MORE INFORMATIVE video on why the goof and delves deep into the topic. Those videos are almost MORE informative than the initial. I have so much respect for this dude and learn so much from these videos. Treasure trove for sure.
Thanks Dave!
Agree. And I like Dave's description "come a gutser" for a goof.
"Second suck of the sav"
same with the supply that blew up. interesting segment of videos about ceramic caps came out of that - was incredibly informative.
I sometimes wonder if he does it deliberately to get another good... errr video out of the subject. Thanks all the same, Dave.
I'm too dumb to think of doing it deliberately.
David, I would really like to see a video on how to reduce the noise on the output of a switching power supply.
Me too! Especially on those ruideng DPS, DPH power supplies
Yeh, me too.
Same here. Lots of beginners and hobbyists who can't spend a fortune on professional gear would probably be interested in ways to improve such budget devices. Might even learn something along the way.
For a given level of output noise, there's really not all that much you can do - you'll be piling on capacitance in parallel, inductance in series, a linear regulator downstream or all of the above.
You’d like Dave to flesh out the suggestions for where to place the ferrite beads, in other words?
6:08 important to note, that the BNC feed thru adjuster has the built-in resistor soldered between the internal pin and the outer casing (in parallel), while the one at the PSU terminal is in series with the coax. So, when you say, you can add a 50 Ohm resistor to either end, it's worth mentioning, that at the PSU end it should be connected IN SERIES, while at the oscilloscope end it should be connected PARALLEL to input, that is between the internal pin and the casing. If you connect it in series, that won't work, and will just add up to the internal 1 MO resistance, hence same ringing will show up on the screen.
If someone as sharp and experienced as Dave can still make mistakes it makes me feel a little better about my own stupid mistakes 😅
I measured the grid current with my little, yellow, china export meter once. It's too damn high! ;)
Well, it makes me sad knowing how many headaches I'm still to suffer.
@@fatnindja Oh no shi-, your Harbor Freight free multimeter blew up when you plugged it in, too? That'd make three I've heard; two in person LOL
@@jimmio3727 LOL, it's not that bad. I had my fuse well wrapped in foil, so what blew up is a trace on board. It was not one of the cheapest. It contained the explosion very well and the meter worked after that but I lost the small amp ranges ... hmmmm.
This meter died from a BEMF spike, which was kind of surprising.
From my experience in the technical industry, admitting a mistake and owning it is one of the rarest qualities. Thank you for demonstrating the right way to handle such a scenario, and using it as a teaching moment for your audience. All the best.
I just want to THANK YOU for all your videos. I've learned so much of electronics just because you decided to make this kind of videos. I hope life gives you back something of what you have given to me and many others that want to learn electronics in a different way. Big thumbs up to the EEVBlog
A video explaining some noise mitigations would be nice.
Yep, I'd love to learn more about noise types and common causes.
+1 on how to mod RD6006
m.ruclips.net/video/wopmEyZKnYo/видео.html
First time ever I knew the correct answer instantly! I said "where is your 50 ohm resistor". All those years installing coax for Ethernet.
How many computer techs spent time troubleshooting a 10base2 networks, only to find that the terminating resistor was missing.
Edit: Just remembered that 10baseT was actually for Twisted Pair. 10base2 for thin coax networks, 10base5 for thick coax networks. Sorry, distant memory and late nights.
@@gorillaau I once used a 10 MHz signal generator and an oscilloscope to do a home made TDR to figure out how far away the open BNC connector was.
@@zadrik1337 They can do that with breaks in optical fibres these days. Sure different medium but same idea. Send a pulse down the transmission line and time the signals return.
Does anyone use coax ethernet any more?
Or Token Ring?
@@EEVblog Probably not, bit I wouldn't want to discount the possibility of there being an antiquated system still in use.
No one take a step, the network cable wasn't terminated properly and the token has fallen out. Its must be on the floor somewhere.
Ringing is not caused by not terminating the coax cable, the problem is the HUGE loading capacitance and inductance of the coax cable. Therefore you should use high impedance probe. Even terminating the coax at it's system impedance is not giving you the correct measurement since you are loading the circuit with a low impedance circuit. Also for high frequency measurements get rid of the long ground lead, use a ground lead as short as possible.
When measuring small noise it is a very good idea to use a 50 Ohm coax, or even a x1 scope probe, with a few ferrite beads around the coax cable, or run the cable multiple turns through a ferrite ring. That easily lowers the apparent differential noise by 10 dB. The problem is that coax cables are not ideal, they do have a ground impedance (search for "surface transfer impedance") which is especially notable below about 7 MHz and which converts common mode noise currents into differential mode voltage. Try it again, Dave...
I liked Dave's correction, a good professional knows how to do that. I have to say I would also like to see how Dave reduces the noise in this same power supply. It would be a very interesting video. Thank you for your videos, I encourage you to continue doing them.
Great review and update Dave.
You should definitely do another video on knocking that switching noise down.
I got caught there too !
Never thought about the importance of terminating the probing, because this will actually load any PSU you are testing hence heating the termination.
Thanks for the reminder 💕
Even Dave makes mistakes. The cool part is that you tell us all about the fantastic mistake you make! Good on ye, Dave. I built one of those earlier RD buck power supplies you posted about awhile back, by the way -- works fine, great for small projects, haven't even had a need to push it hot, and it holds up fine to the light duty I give it.
Please, do make the mitigantion video. I would be very interested and hopefully many others would too. Plus there's a hope, that RD could look at it and incorporate it into their next design.
Petr K , I'm sure a lot of Chinese designers would sure thank you a lot for doing their work for them!
Great point about the 20MHz bandwidth - it was the scope bandwidth that Computer Products (CPI) had when they released their first switch-mode supply in 1971. I recently got a tender document from a very devious German test and measurement company who tried to claim 10x lower noise than they really had - they specified DC-20kHz bandwidth via a tiny side note when all other suppliers had used the conventional 20MHz bandwidth.
appreciate the integrity here Dave. An easy mistake any of us can make.
I'm learning how to measure RF and other things for analog video processing, I've been using 75 ohm video BNC coax for probing - but I'm just getting back to this and had forgotten about termination to avoid distortion, so THANKS! I've always initially used AC mode on my scope, which has no termination (and I always use termination on video transmission lines). But I don't know all these circuits yet and I don't want to blow up anything or kill or freak out any agc's or oscillators with a 75 ohm load on perhaps a 10v (or whatever) DC internal signal line. Termination does kill noise contamination nicely tho! Highest frequency to look at here is 5mhz FM.
That’s the way to go Dave. I watch your channel since years. I like it very much that you tell your audience when you did something wrong. We all make mistakes and we can learn from those mistakes.
Best method for accurately measuring noise from switching power supplies is to use a differential probe. This reduces the variation in readings due to switching common mode noise being transmitted through to the scope via the probe earth connection. The effect of the common mode noise was possibly the cause of the changing display while Dave manipulated the probe earth lead.
In the absence of a differential probe the scope's sensitivity to common mode noise could possibly be reduced by adding some ferrite clamps on the scope probe lead.
A livestream(of consciousness) on your approach and methodology on cleaning up the PS output would be interesting to watch ! Thanks for the informative followup vid, and reminder of a great probe edu vid to circle back to Dave !
Dave - now's the right time for some transmission line theory...
I was about to ask same thing. Can you please help explain impedance mismatching and voltage reflections? I’ve never really understood the concept, thanks Dave
And this is why i default to using only my scope probe for most things lol. I saw that mistake coming the moment you mentioned you did something wrong. I just don't trust myself after all the times I've made this mistake.
I found when CE marking equipment that switching PSU noise is a big problem. the CE marking testing for the PSU is done at a defined power % of full output. the PSU are made to pass this test! When it comes to selecting the PSU for your product you don't want to run the chosen PSU at near to full load for reliability, but now we have a problem your product fails CE testing on PSU noise emissions. try adding filters to a sealed PSU, adding in to you product can help but it kinda the wrong end of the cable. if you ask them to fix the issue boy does the price of the PSU go up. it almost ends like "take a lot of different PSU's to the test house and do a select on test. great video gets the old grey cells working.
Much more informative than the video the noise appeared in !
This video made me buy the 6012. Which actually better on the noise as well
I would love to see a video on improving the ripple from a smps.
Years ago, I put my hands on an old Hameg HM-203 in a fleamarket. There was no probe, so I made 2 with a Ethernet 10base2 cable. And I only got garbage. So I went to the computer shop and purchased 2 standard 1x-10x probes, and made the calibration for my oscilloscope following the user manual.
Oh, that brings back flashbacks of the thincoax network I had to work with in the 90s. People tended to take the T-connector and their cables with them when moving office, and sometimes (when they were at the end) the terminator, not realizing they were part of a chain on (basically) one coax cable. In cases where the terminator was "stolen" there was no standing wave and bandwidth went into serial connection territory (if at all), while taking one in the middle cut off all that were "behind" it. You young people with your fancy UTP don't know how hard it was in "the good old days" :)
And of course all our cables were RG58 and our terminators were 50 Ohm.
Thank you for being such a great example to today's young engineers on how to handle inevitable mistakes.
In some PSU measurements there is a 0,1µF ceramic Cap and a 1µF Electrolyte specified.
Look at Intel Power Supply Design Guide, there is a part for "Measuring Ripple".
He screwed up a second time..... Let's wait, may be will notice that....
@@MrPerrey Yeah, Power Electronics is a special field that many Engineers don't seem to like much at all for some reason...
Maybe this gives dave some reason to look a bit more into power electronics and make some PSU stuff? For example how to measure 40A peaks that happens for an ms or two.
@@Stefan_Payne Yeahh, but happens not only power application, it's more basic eaven in RF engineering. To measure PSU ripple and noise there are out industrial specifications how the right measurement setup looks like. Blocking the DC content for this measurement is also good practice eaven if You follow Your own setup. Don't understand why he screwed up a second one and so badly.... He was just better.
It takes a real Aussie to admit when SHIT happens, because when we’re shit we’re really shit, but when we’re good our shit doesn’t stink, and as usual Dave comes up smelling like roses.
Thanks Dave for this video. I make mistakes every day, and it makes me angry with myself. Now I am more relieved to experience more, because this is a hobby, not a matter of live or die for a small error
Life or death you'd have procedures and others checking. When it's just you...
Univerity of Dave
I think we have to be careful when we switch a scope into the 50 ohm mode; my Rigol warns to limit the input to 5V rms when that mode is selected. I assume using an external transformation as you did here lets you exceed that.
Yes. 0.5W is a common 50ohm limit inside scopes.
15:33... yeah that would be a good video for general switching PS noise reductions.
RF transmission line shall see the same Z (matching the line) at both ends.
Exactly, it should be terminated with 50 ohms at both ends. The output of the power supply is effectively 0 ohm so a series 50 ohm resistor does it as he showed in the video.
Digital audio lines such as AES/EBU and S-PDIF terminate at BOTH ends of the line with 110R and 75R respectively. Is there any downside to terminating both ends?
It's the only correct way.
@@Audio_Simon The line TERMINATION and SOURCE are to be the same. However, at audio frequencies the reactive component of the Z formula is lesser than the ohmic resistance (cooper losses) I am not familiar with lines you mention. 600 ohms Z has been the standard for audio in most cases, as the old, now nearly defunct telephone land lines.
Terminating at either end is still much better than no termination because any reflections will only reflect once before being absorbed. With mismatched impedance at both ends, reflections will just ping pong back and forth until they dissipate in the very low loss of the line.
Did the same thing trying to deliver a signal from my signal generator to the input on an audio amplifier. Instead of an audio tone, I had an RF oscillator instead. Amp was oscillating at about 2 MHz at full power.
Try to measure it differential with two probes. That is the correct way.
That Keysight $cope "does not have" a 50ohm impedance choice? FINALLY! Something my boat anchor has that Dave's scope doesn't. Damn I (finally) feel proud for a New York minute 😎
Thanks for the follow up. I needed to be spoon fed that info. Cheers Mate! 😁
Most low end scopes don't have a 50ohm input.
Please make a video about how to reduce the noise on this unit. It appears to be otherwise very well made and the user-friendly interface, it all looks pretty good. If the noise problem can be reduced to 10 or 20 mV, it's a serious power supply. Maybe some dampened CLC filter at the output could do it at the expense of losing some accuracy.
I bought a baggie full of banana connectors at goodwill for 99 cents, in the bag was two Tektronix 50 ohm BNC connectors.
It doesn't matter. Interested buyers will only see previous video and judge the product accordingly. Thanks youtube for stripping annotation feature.
You always have good content. Happy Thanksgiving from the USA
Calibrating the probe connected the 1V p-p output of the scope until the screen depicts the known predefined square wave is a great start before making assumptions about unknown signals.
Calibration can be used in many areas of life. You start of with something you know before probing unknown territory.
Charles Wheatstone the inventor of the electric telegraph, the Wheatstone bridge and more is known for the principle of comparison, but even more interesting he may be the cause of where civilization is at right now as he, by coincidence, was also Oliver Heaviside's uncle by marriage. Heaviside rewrote and added understanding to Maxwell's equations and, guess what? - invented the coaxial cable!
In this case even a straight wire need impedance calibration not to distort the signal.
nice job Dave!!
Wiggle wiggle wiggle. Yeah!
JohnAudioTech 13:15
I would really love a video about how to reduce the ringing on the supply. Please do it if you have the time!
Kudos to Dave for owning up!
Almost 10 times difference, holy cow!
Is there any detailed explanation of what exactly happens here available somewhere online?
I've been there more times than you, Dave. LOL LOVE the remark by Steven Leibson, "Way to turn lemons into a lemonade video!" remark! SO TRUE!
Like others have said - my favourite kind of videos.. "So, I goofed and here's why.."
This is a great video, I fell into this trap (yep I'm a total newbie) only the other day, and now I know something new!
Someone may already have pointed it out (280 comments already made... I've not read all of them...) but, Dave , why don't you have used a "pig tail" probe in x1 mode ? That's the best way to get rid of most of the induced switching noise in the measurement process ? (Sorry for my globish english, l'm a frog eater !)
I would have used a BNC to banana adapter, but the connectors are not a standard distance apart to allow that.
Maybe you can email and contact the developer of that product and ask him for help and suggestions to clean up the noise in his power supply. Probably something he just left out because of size constraints or cost we all know how that goes.
Given that the PSU is isolated ground, I'dve used two probes on two channels and set the display for differential mode. Why get all the nasty common-mode ground noise involved?
@Roberto Gatti Not the greatest sign of design quality, is that?
Excellent. Well spotted.
Truly geniuses makes lessons from mistakes. Cheers Dave!!!
Hi Dave, thank you for your videos! I just wanted to let you know (if you do not know it already) that there is an advanced version of the RD6006: the RD6006P. I assume that "P" stands for precision. RD Tech: "It has 5 digit voltage resolution 1mV, current resolution 0.1mA , lower ripple" and is "more accurate." Typical output ripple is said to be "20mV Vpp". However, it costs a little bit more and it needs a little bit more "headroom" for the input voltage (+10% of output voltage + 2V). So in order to achieve 60 V output voltage it is recommened to set the input voltage to 68 V. May be this is an inspiration for independent test by you. Link to their own ripple measurement: ruclips.net/video/ZncOU4eSjvk/видео.html
Best regards from Germany!
Come a guster? Austrailian Slang Dictionary... "
Come a guster
Definition:
(Expression) Fall heavily to the ground; fail through error or misfortune.
Example:
He started a business but things came a guster pretty quickly." Straya.
Nice slang.
Sincerely, an American.
Its gutser not guster
@Jaycsee66 - came here to mention this. “Gutser” as in “fell on yer guts” - or more eloquently to fall on ones frontal region, ergo, “fall on your face”.
Strewth......ripper
Not guster...... "Gut-ser". Your hearing skills are drier than a dead dingo's donger. LOL
I noticed Ali express has this, case and power supply all on sale for black Friday if anyones thinking of buying!
Besides the quite high resistance in the x1 probe center wire, it also is designed to minimize the capacitance. That is done by minimizing the center wire diameter to the shield diameter plus also making the insulator of foam material. That way it is almost like plain air. Finally, you probably never find those probes wit more than about 1 meter length. Anyway, good story again!
All is forgiven since Dave measure is upside down. RD 6006 seems like a winner at this prince point. I hope they release 50V15A at some point. :)
I don't mind learning from my own mistakes; but I absolutely LOVE learning from OTHER people's mistakes. ;-)
Some measurements in the frequency domain with no-, half-, full load at different output voltages would be nice.
So a 50 ohm termination means having 50 ohms of resistance in line with the measurement? I always thought that meant 50 ohms between the input line and ground. A schematic diagram on the white board would be very helpful.
Good question. Normally I consider a termination as a parallel load. Even a 'in-line' terminator puts the load in parallel. Either way, the solution in this video shows the resistor in series with the center conductor of the coax.
Tbh I noticed on the original video that the coax was going straight in but just assumed you had an internal 50 ohm load selected internally on your scope!
Would love to see how to improve on this Dave, thanks very much for this.
take a shot every time dave says "come a gutsa"!
Would it be okay to have a DC blocking capacitor followed by a 50 ohm termination resistor? That would keep the DC voltage off the terminating resistor. Just a thought.
Yes, you can use an external DC blocker.
I quite releaved that its not that bad. I was quite exited when i saw this PSU, I was going to build two of these in a case for dual output. That ripple in the last video was quite a deal breaker.
Thanks for the follow up!
A quite nice video, thank you, Dave.
Hah I just got done watching the last video and read your comment on the mix up. Think I’m gonna pick this one up!
I posted on your first video on this bench supply about my suspected usage/behavior of the battery charger terminal. I'd love to see a follow-up to know if my suspicions are correct because if they are, then I'll be wanting to pick one of these power supplies up. The short version is that the battery charge terminal is expecting a BMS on the other side of it.
And that's how we learn. Thank you!
12:20 Speaking of DIY probes that reminds me of being in the lab when someone would give me back some electronics they were measuring and it would come back with Caddock resistor warts all over the boards.
Out of the bad comes alot of Good ....... knowledge. That’s how we learn ! 👍
Why not measure the noise with the excellent technique you've presented in a previous video--the 2 probe method using SUBTRACT mode on the 'scope?
Very interesting! Definetly learned soemthong today!
I’d really love to see if this and how it could be improved
videos like this are why I love you
From Manual: "Ripple and noise:
Since this is just a DC supply, why couldn’t you just put a HF filter on the output?
I'd like to see a video on this termination stuff, and why it can be done at either end.
Preferably with some diagrams and theory. (Can it be series or parallel?, Can there any divide by 2 measurement considerations?, etc?)
I have always assumed a 10x probe is so named because it is 10 times the impedance. It thus affects high frequency, high impedance sources less.
No expert on these maters, but 200mV noise on a 5V signal seems huge.
Nice follow up video!
Please investigate how to reduce ringing. Very interesting if DIY could reduce pk2pk ringing with this cheap power supply.
@EEVblog - I think you screwed up a second time - There exists specifications how to measure PSU Ripple and nois and your setup had almoste nothing to do with that. You had been just better !!!
Hello, good job, maybe test new rd6012?
Great update
Thanks for sharing👍😀
But DC(-) is connected to electric grid earth via the oscilloscope, loading it asymmetrically. At least here, it contains a lot of low frequency noise (Speaking of DC-500 kHz).
Shouldn't a differential probe be used for that?
So what's the verdict on the power supply, GO or NO GO? I'm definitely a novice with DIY electronics, but I keep plugging away. Having a benchtop power supply would be nice, as I am pretty fond of stepper motors, big and small. Is this the way to go for someone like me or is there a better choice? Having 2 output voltages would be ideal, but I guess a small 5v adapter is fine for logic...
An error that turns into a didatic lesson for probing noise in PSU. And also good to hear the Predator 2 quote :P
Would be a cool video to mod it as you suggested!
Have been considering building a portable power supply with RD6012/18 powered by 3 x Dewalt 18v slide batteries in serries, thinking that to be a good way to get a clean energy source. I’m under the impression that the slide batteries get protected from over discharge etc by electronics in the tool, so less electronics in battery to generate noise. Tell me if u think this a really bad idea, plz. Love ur work, keep it up.
Exemplary follow up. Just assumed that bikkie black box had ya covered.
Dave's SWR came a gutser
lol
It happens.
@@EEVblog when i was a kid i went down a BIG hill (it was asphalt) on my mate's shitty old dragster bike and got the death wobbles and came off. I skidded for about 20metres on elbows and knees and god knows what else. When I pushed it back to my mate's house his dad was there beaming down at me and he said "oh came a gutser did we?" Everytime I hear "gutser" it takes me back to that day.