HTML-код
- Опубликовано: 19 окт 2024
- This video shows how to use the FFT function on an oscilloscope to measure the relative amplitude of the harmonic distortion components of a signal, and then take those results and compute the THD, or Total Harmonic Distortion. Actual measurements are shown, and results are validated by a THD calculator in the scope as well as a THD measuring multimeter.
Videos mentioned in this video are:
Audio Compressor project: • #157: Circuit fun: Aut...
Demo of above: • #163: Automatic Volume...
Basics of using FFT: • #65: Basics of using F...
Basics of dB: • #136: What is a dB, dB...
Notes from this video:
www.qsl.net/w2a...
This is what the internet was created for. I search "how to measure THD" and the first result is an absolutely excellent explanation without any beating around the bush!
For audio frequencies, I use a software called Arta. It will display THD in real time using the line in on the computer's sound port. Most computers audio will be of low distortion .1% or less. Mine goes to around -120dB (except for 1/f noise). Everything works in demo mode except for saving data. I recommend making an input protection circuit for your computer if you use it for this.
topic worth a video
Which module would you recommend for measuring harmonics in electric current and power?
@Carl Jito you obviously got the wrong file of your computer is shit.
Can you elaborate further please
Clear, concise, great pace, excellant explanation and graphics. Thank you for not over/underestimating the intelligence and experience of your audience. I subscribed, looking forward to more.
BTW, LOVED the 'organic' design of your little follower!
Thanks Alan. I always learn something from these and I have been doing electronics engineering for over 20 years.
Hands down the best channel for down to earth solid content on RF electronics on RUclips.
Cheers,
Thank you!
I've tested for THD before and knew why it was bad but didn't know the mathematical component of what I was actually observing. This is fantastic....
Thanks for this Alan. I've spent my working life in the visual arts, so maths doesn't come easy to me... but I was able to follow this information quite well, thanks to your delivery style. I'll watch a few times more till it sinks in. I really value the things I have learnt through your channel.
Fantastic as always I love getting a notification w2aew has uploaded yay time to learn something new Thank you for your time effort and knowledge it really is so valuable to folks who want to learn thanks to people like you we can and all you ask in return is the gratification of knowing your helping others you truly are a top fella Allan thanks again take care Steve
Alan you are my honorary professor. Wonderful video, thanks.
Thank you Alan! I used to know this, but needed a 'refresher' course! I'm measuring the THD of portable generator outputs... See ya on POTA! 73 OM!
Practicing with a really old General Radio 1330-A sig gen and my TDS 3012B scope, I was really surprised to find less than 1% THD output! Amazing... thanks again Allen! WR2E dit dit
Allen, would you think it's reasonable that the power company is supplying 3.5% THD AC to my home? This is what I came up with measuring the line coming in. I wanted to get a baseline before I measured the generators output. I presume this would be affected by the power factor of the things running in my home? Or would this be what is actually being supplied? I suppose that other loads on the line from neighbors homes for example, could/would also affect this?
@@Wheel_Horse ONLY 3.5% ? Wow, I used to live in a place (S.E. Arizona) where the AC power line signal looked something like a triangular wave, withe a bunch of "squiggles" on it. Here (Central NM), it's a noticeably distorted sine-wave, but still recognizable as sinusoidal. Not only the P.F.s of what's in your home, but EVERY home & business, along with all those arcing insulators, and corona discharges.
I recently inherited a HP 334A Distortion Analyzer (couldn't pass it up, it was free!). It does a great job of measuring THD, although a Keithley DMM sure would be a lot smaller! TNX for the lesson Alan! 73 - Dino KL0S
All your hard work is very appreciated Sir. You make the world a more precise place.
The distortion you're seeing appears to match a exp(sin(x)) shape, which is really common for linear BJT amplifiers with large signals and no feedback. Really cool video!
Another treasure from Alan! Great 👍 tysm!
Excellent practical and concise description on measuring audio THD. Since most hobbyists will have an 8 bit scope instead of a $10K+ 10 or 12 bit A/D scope like the MDO4104C you are using here (Nice!) your point about doing accurate measurements on a low distortion amp is important. It may require using a real spectrum analyzer as described in your video #46 which these days more people are adding to their test bench. The limitation is that the THD can only be done for frequencies above 9kHz, the bottom end of most RF spectrum analyzers. Also need to consider measuring the distortion of the function generator providing the signal input, and subtracting this from the amplified output otherwise its GIGO. Interesting topic for a follow on video if enough audiophiles are still building their own amps especially to consider class D designs which are now popular in consumer gear.
THANK YOU SO MUCH!!! I have a presentation in which I'll be defending my research protocol for biomedical engineering and this was a key component to it. THANK YOU
Nice work. I quite like using the HP 3562A for THD measurement, as it automates the calculation while letting you eyeball the relative levels of the different harmonics, otherwise an analyzer like the ABO AVX-310 does a fine job if you just want a THD+N measurement. I quite like the technique of using the 3562A on the monitor output of an analyzer (such as the Leader LDM-171 or the AVX-310) as by notching out the fundamental you can look quite deeply into the harmonic content without taxing the dynamic range of your FFT instrument - a bit fiddly, but can be effective.
Alan another superb video that explains things clearly and concisely. Thank you.
Excellent video as always, you really have a great way of simplifying the more complex subjects! Keep up the good work!
Your videos have so much knowledge in them. I love your videos
Thank u very much for the course of electronic measurements in deep , I have learnt a lot of THD and its measurement using FFT in DSOs....
I’m a big fan of these videos! Your word of caution at the end has me hoping you’ll make a video that illustrates the relationship between vertical resolution, dynamic range, and noise floor on digital oscilloscopes and spectrum analyzers!
Hmm - I'll have to add that thought to my big list....
Excellent video, as usual! Thanks a lot for sharing your amazing knowledge! Cheers from Brazil!
IMHO, in sound frequency range applications 24 bit sound card and GNU Octave can be used for measuring THD. One can record a sinewave sample, then upload it into Octave and calculate FFT and THD. It's probably tricky and not a very precise way since the sound card is a device with unknown nonlinearity, but I think it should work at least for evaluation.
it might be not very linear, but i think that if you're working relativelly inside the auditory range (20 - 20k Hz) and with voltage levels not too close to the input limits it should work fine enough
Of course, best case scenario you would have a datasheet available to look for the response curve at the input
@@Abossow77 Or a good audio-frequency sweep generator with known output characteristics so you can measure the linearity of the sound card.
As always, a total joy to watch your videos
Thanks for this video and being so honest to state, that the THD-function in a standard 8bit scope can´t supersede a truly harmonic meter (bridge), which operates down to 0,0...%. That´s why i prefer my HP334A in combination with a HP652 oscillator.
It never ceases to amaze me how much math in engineering and science goes back to the Pythagorean Theorem. Compare the math for THD to the formula for a right triangle; now compare that to Einstein's Law of General Relativity. Once you know what you're looking for, you'll see it popping out everywhere.
Hello Alan,
Thanks once again for an outstanding and concise video! It is so much appreciated! Take care.
Great video Thanks for your effort and time
I wanted to do such measurements for a long time now. Alan, thanks again for such educational videos!
If I can't measure THD of my amp with the scope, I'm just gonna say THD
Fascinating discussion and calculation, thanks!
This would be fun to perform on a 5f1 Fender champ, or a Eico 147A Signal Tracer. Thanks for the awesome video!
Fantastic tutorial - thank you! Now, all I need is a decent Oscope.:)
I will surely add it to my laboratory excercise on THD. Normally we only use analyser, and these are not common. Nice!
ReMarkable .... (wink). So high on my want list, so tired of bits of paper everywhere at my workbench. Oh & thanks for the refresher.... all to easy to forget the actual 'bones' of a spec when a test unit just gives you the number every time....
Always interesting and always useful. Thank you
I learned something very useful today.
Excellent content very well explained yet again. Thanks!
Excellent video. I used the Tektronix TDS2022C to do the same thing.
Thank you for your good illustration.
Very interesting and educative, much appreciated. Thanks!!
Excellent as always! Thank you so much.
Funny, I was literally thinking about measuring THD yesterday :)
Same
@@JasonLeaman I forgot to mention that I was also thinking that I didn't know how to do it lol
Thank you very much! Always Helpful.
Thanks Alan, great video...............................................Berni
Thank you very much sir, another wisdom dump to add to the pile :D
Hi,
Can you please tell me How we measure AC voltage by oscilloscope like Hantek 6022BE ? why it shows offset AC signal and the altitude is only 102 V instead of 120 V? thanks for your reply.
Thank you for all of your great videos. In video #65 you state that one half the sample rate will be the highest frequency you will see out of the FFT; the sample rate was 25kS/s and the maximum FFT frequency was 12.5kHz, just as you state. In this video your sample rate is 500kS/s and the maximin FFT frequency is the same, 12.5kHz (1.25kHz per division, as shown on the math readout). Please explain how you got such a low FFT maximum frequency with such a high sample rate.
Because I "zoomed" the FFT result.
Very good job sir 👍 👌 🙏
Excellent explanation.
Thanks a lot! I'm not the one who asked, but I also wondered how can this be calculated for years now. I was pretty sure it's possible. It was probably mentioned in school as well, but it was too long time ago. I can forget anything that I'm not applying within a few months.
Too bad that none of the scopes I have access to has THD calculation in their FFT maths menu, but this manual method is not too much of a hassle to apply if needed. I wonder why don't every FFT capable oscilloscope has THD calculation amongst its maths functions. It can't be too complicated to implement, if it can be calculated manually within a minute.
I guess it's because an oscilloscope is one tool you can use to compute THD, but it really isn't the best because accuracy really suffers when you go below several tenths of a percent. A spectrum analyzer is better, but many don't go to audio frequencies.
@@w2aew Hi Allan how come they dont go down to DC why start 9khz or?
Most are AC coupled, often due to mixer and preamp design.
@@w2aew Thanks for reply
This video is really helpful, thank you!
Alan, I have found better resolution with a 2 channel function generator and DSO by taking a sample sine wave voltage out of the DUT and into a resistor matrix that allows me to add channel 2, 180 degrees out of phase, to cancel out the fundamental frequency. The 180 degree phase shift is just a starting point though due to phase shift in the DUT. I then adjust channel 2 phase and amplitude for maximum null and read the RMS voltage left over.
Be careful with this - adding the inverted signal to itself it will also cancel the odd harmonics! Thus, will give you an artificially good THD result.
@@w2aew Interesting. How is this different than using a notch filter to drop out the fundamental, like an analog distortion analyzer?
The results I have gotten cross check with an HP331A. (Just to be clear, I am inverting channel 2 and adding it directly to the matrix without it going through the DUT. Only channel 1 signal goes through the DUT.)
@@robt7785 Ah - it wasn't clear that you were subtracting the *input* signal, before the DUT. In this case, you should only be left with the distortion components added by your DUT.
@@w2aew Thank you. Yes, I apologize for not being very clear.
You can also use a time series captured from a digital oscilloscope if you are willing to do a little math in Excel. Square the voltage measurements from a single period and add them up. This is proportional to the total power. The power in the fundamental is proportional to the peak (half of peak-peak) squared divided by 2.
THD = 100% X (P_total - P_fundamental) / P_fundamental
I should point out that you do need to eliminate any DC offset (e.g. use AC coupling or subtract the average before squaring) in order for this to work.
THank you so much. Interesting as all your videos!
Howdy. Nice presentation. One point however about audio amplifiers.
I suspect most specs use only one base frequency. It is likely to be 1 kHz.
I believe THD should be evaluated using several base frequencies from 20 Hz to 10.000 Hz. Harmonics of 10.000 Hz are not audiable. Base frequency spacing = double:
f / Hz THD / %
20
40
80
160
320
640
1280
2560
5120
10.240
To evaluate the overall THD the mean value of the THDs may be computed. However. I believe the rms should be used to evaluate the overall THD.
Regards.
Nice! The instrument you are using has a real spectrum analyzer. Will the FFT math function of a low end scope like DS1054z, do the trick?
I wasn't using the "real" spectrum analyzer in this case, because the signal was at 1kHz and the spectrum analyzer only goes down to 9kHz. This will also work on the DS1054z, assuming that it's FFT gives you the result in dB_. Keep in mind that *any* 8-bit scope is going to be limited in how low it can measure THD because the scope channel itself will limit the performance. In the case of my scope - a proper THD measurement with my THD DMM on the clean input signal shows 0.03%, but if I measure it on the scope, it was closer to 0.4% due to the limitations of the 8-bit digitizer. If you have THD less than 1%, the scope FFT method really won't give you an accurate result.
cool explanation and cool diagrams ....awesome
Thank you Sir for sharing your knowledge.
Quality video as usual. Thanks !
Hello Alan, first, thank you for this vidéo. Very interesting for me. I will realize this BF circuit to experiment FFT function on my scope Rigol DS2072A, based on your instructions. At 3:54 of your video, you refer to videos #157 and #162. I think this is #163 instead of #162.
Again, many thanks for all the videos.
Oops, you are right! At least the link in the video description is correct.
Since the measurements are all relative, one could set the V/DIV including fine adjust to span all +/-4 div. on the screen.
Using average function to smooth out the FFT also makes for stable values ( but not more accurate, just help the operator to look instead of estimate :) )
Very good explanation. Thanks. Super. Question, to measure the distortion introduced by the DUT, how do you correct the THD with the measured THD of the input signal (measured with the 2nd channel) ? Are there scopes which can calculate this ?
You'd have to do that math yourself by measuring before and after.
Excellent.. Thank you.
Great video!
How do you measure the signal to noise ratio using an Oscilloscope? or what lab equipment do you use to measure the S/N ratio in various power supplies because different types of power supplies when being applied to different types of amplifiers circuits like FET amplifiers, transistor amplifiers, IC op amplifiers, tube amplifiers circuits will raise the noise floor so you will hear hissing white noise, hum, ripple, Switching frequency from the power supply, etc. The hissing white noise is caused by the linear power supply or switching power supply? because the amplifier circuit will raise the noise floor higher so you hear the hissing white noise from the linear power supply and switching power supply?
Thank you!!! The video is very informative 👍🏼
Thanks for your video and sharing, regards from Rome. 73s Pasquale IW0HEX
Greetings from down under. Great video, so interesting and informative just like all your videos. Thanks so much. By the way my XYL is from Hillside New Jersey. 73 Gary VK6LX
Very nice. Hillside is about 30 minutes away.
As always good job.
Thank you so much for this video. Your the best.
Thanks, will implement it in HScope!😁
Great stuff, I have been collecting you notes on all of your other RF wisdom, Now for my question where can I downlode your PDF files on this great piece of info. Again Thanks for your time for doing the youtube on these. Thanks Keith
As with most of my videos, there is a link to download a pdf of the notes in the text description of the video on the RUclips page. You usually have to hit the “Show More” link.
@@w2aew yes I did hit the show more tab , but no link to notes
It’s there. And, now here:
www.qsl.net/w2aew/youtube/THD_with_FFT.pdf
Awesome!!! I love your vids
Hi, my doubt is, why square root, your result of SUM its in terms of power not voltage. Thanks
Thanks Alan!
👍
Thank you for the video. 4:56
I really like your videos and have learned lots. I purchased a Siglent SDS1104xe scope awhile back an would like to use it to measure THD. However its FFT function doesn't have DBc units and only has DBm, DBVrms and Vrms. I was hoping you might provide a clue as to what math I would use to calculate THD based on the scales available to me? Thanks
Easy... Take note of the dBm level of your fundamental - this is your reference. Then, simple subtract the dBm of each harmonic from the dBm level of your fundamental. This will give you the dBc level of each of your harmonics.
Yes. Very interesting. But also over my head 🤓 But that's OK 👍😁. TIP: I usually find the THC by looking in the user manual 😉
A shortcut for a very rough estimation would be using only the second harmonics. Then the result is simply 10^(dbC/20) - because square root is the same as power of (1/2).
In the example this will be 10^(-20/20) = 10^-1 = 0.1 = 10%. Close enough for many cases.
That's assuming that the 2nd harmonic is the highest one. If the 3rd is higher than 2nd, then use this shortcut with the 3rd.
you are a star.
I really appreciate.
Thank you Alan :)
why there is no more harmonics to de 2nd power in final calculations? it's just the sqrt of the sum (min 7:32)
And again Howdy.
For audio amplifiers.
1. Download ARTA Software and install. Demo mode is free.
2. Build an attenuator of, say, 30dB. 10 k and 330 ohm and connect in parallel with the dummy load.
3. Connect to the Line-In jack of your computer. NOT the microphone jack.
Regards.
Good video, but I'm confused a bit. I have the same scope as you. According to the manual and the scope's user interface, it reads out in dBv, not dBc. But when I did the math using dBv I get the same expression as you.
I figured that (since dBv=20*log(Vn/Vf)):
(Vn^2/Vf^2) = (Vn/Vf)^2 = 10^(2*log(Vn/Vf)) = 10^(dBv/10)
which is the same simplification you used for dBc.
What gives?
When you use the fundamental as the reference (Vn/Vf), you're effectively calculating dBc (dB relative to the carrier level).
Great video, thanks for uploading!!! What type of pas are you using to write on? It looks digital, but also looks really good!
it is from www.remarkable.com
I want to analyze some vibrations that are around 50Hz - 10,000Hz. Can I use a spectrum analyzer for frequencies below 9 KHz? Is there a workaround to feed the analyzer a bigger frequency correlated to my original frequency? The oscilloscope that I have access to, has a FFT based on 1 KPoints. FFT is basically a joke...
There are some spectrum analyzers that are designed with a lower corner frequency, but they're not that common. You could upconvert using a mixer. I did a video on up-conversion in the context of monitoring HF radio signals on a USB-SDR receiver that didn't receive that low - this is the same concept: ruclips.net/video/XeQnDpzhcRo/видео.html
What is the bit rate of the A/D converters in various THD multimeters? Since the Oscope is an 8bit A/D converter it can't measure THD signals that are low levels, what do you mean by that?
The sample rate of the ADC used for the THD measurement is a function of the test frequency, up to a maximum of about 100kS/s. It is a 16-bit ADC. The scope's 8-bit ADC doesn't have sufficient dynamic range (smallest digitized level is 1/256 of full scale), which limits the dynamic range of the result.
What do you mean by limits the dynamic range, I'm confused about do you mean power level/ input voltage level driving the Oscope channel that if you use a millivolt input signal into the Oscope its hard for the Oscope to measure THD of millivolt and microvolt input signals?I don't understand why an 8bit ADC would cause "more" THD harmonics compared to using an 16 bit ADC?
You were pointing at the first Harmonic and calling it the second harmonic, I'm a bit confused, great video.
Cheers
A common misunderstanding. Think of the fundamental as the first element of the harmonic series. The second element (twice the frequency of the fundamental) is called the second harmonic as it's the second element of the harmonic series. Although we tend to refer to any component that is not the fundamental as 'a harmonic', in reality they are all terms in the harmonic series. A bonus of this way of numbering the harmonics is that it tells you the frequencies. The second harmonic is twice the fundamental frequency, the third is three times, and so on.
Very good video.
Unweighted vs weighted THD+n
Some interface have these term but is difference
Thx great info mate.
Nice e-paper device. What is it? Can you draw on it itself, or is it just display?
The e-paper tablet is from www.remarkable.com
I set the FFT to read voltage, then use the first formula sqrt(V1^2 + V2^2 + V3^2...)/Vf ?
That's the way I have been doing it. Am I doing it wrong? I really want to know.
That works too, no worries.
@@w2aew Thanks, Alan!
how do you get that THD harmonics application? Is possible to get on a Silgent scope? Very new to oscope use
The THD application is something that is sometimes offered as an option on a real spectrum analyzer. I don't know if I have ever seen it available for a scope/FFT, so I doubt it is available for the Siglent scope.
@@w2aew good info to know! Thank you! My application is to measure THD of portable generators, and to see specifically on inverter generators how well the output stage is filtered for EMC
excellent!!
I have a cheap FY6900 signal generator and an HP 333A. I measured the THD of the FY6900 at 5KHz @ 1VPP to be about 0.5%. Now I want to feed that signal into my audio amplifier and measure the THD of the amplifier. If I were to get, say 0.6% out of the speaker terminals of the amp, do I subtract the 0.5% that I know the signal started with, giving me a THD of 0.1% for the amp (at least on that input/output channel)? Thank you!!!
I don't think it is as simple as that. Your amplifier may product distortion components at different phases than the generator, which will cause them to add constructively or destructively, or somewhere in between.
@@w2aew thank you! After thinking for a few more minutes I ended up coming to the same conclusion. I told my dad that and he didn't understand/believe me lol. Said "you probably just subtract them" but I told him the distortion is still just unwanted waves of energy and who knows how they will interact. Altho I did read the 333A incorrectly; my FY6900 appears to be at .05%, not .5%, so that makes a significant difference. Thanks again for the reply! I have learned a lot from your videos over the last year or so.
@@kurchak For audio measurements you should check the AudioRightMark application.
keep up the great work I learn alot from your videos 73 from K4TEP
Can we use a linear vertical scale to make the math easier?
You'd have to then square the terms - using the dBc values, you avoid that step. Plus, not all scope FFTs have a linear scaling option.
Lou Eckert .... Using logarithmic scale would make scaling much easier on screen rather linear so the small values can be shown much easier with large values on same graph...
dude you are the best