Discrete audio amplifier project PT20 Measuring distortion with computer

[@johnyang799]

Hi john. You can turn on the THD (no Noise) legend in Arta. Just in the settings. Also you can use REW as well. A little bit more clumsy software written in Java, but it can show you amplitude of all the distortion harmonics which is really convenient. Also just use 48khz. It's good enough. For 20khz distortion just use 96khz.

[@Elecifun]

Hi John, would you make a short video showing how to calibrate ARTA. I've set mine up but with looming question marks. Having you show and tell would help me immensely, and I expect many others wishing to try out their sound card for distortions measurements.

[@MarcelOoms]

Hope the amp will turn out well! Btw, the Digilent Analog Discovery 2 is a nice device for audio measurements over usb as well. Blueglow electronics on youtube demoed this with other software, very nice and less expensive.

[@markanderson8066]

our ECSE students get them and thus have a "testbench" in their backpack!

[@Sevalecan]

Awwwwww, the kitty! Wait, I was here for audio stuff!

[@peteb2]

Thanks for the reminder that ARTA exists... I’ve been scratching around trying to find out how I can update my old Tek scope for FFT which appears greyed out on the screen menus

[@zodak9999b]

Those spikes looked to be 1 kHz above and below 44.100 kHz. You could try a 2khz fundamental tone and see if the two spikes change to 42.1 kHz and 46.1 kHz.

[@SianaGearz]

Yes i was thinking. It looks like it's intermodulation between 1KHz and the sample rate of the MP3 player. Maybe its internal resampler. The Rockchip processors these players are built on only have a 48KHz audio output, 44.1 audio has to be resampled by firmware.

[@tangerinq]

In the digital audio realm, you have tons of tools that could work for you. Normally these are being used in audio production. First, you could record the measured signal as audio and then selectively remove the unwanted noise with an EQ. Second, you could use some very advanced denoise plugin (such as izotope) that lets you analyze the noise pattern and then intelligently remove that from your signal. Third, you could use a service of a local recording studio where they probably have a much better audio interface.

[@MassimoTava]

I hope keysight notices you one of these days

[@joepalovick1915]

Great video! Looking forward to Part II. I would also suggest looking at the Analog Discovery2. I couldn’t find a video or very much info on the QA401 so will be interested in a review if you get your hands on one.

[@dantheman1998]

Are you going to release the amp as open source or you going to sell it on your own? Any other designs after your done with this one?

[@JohnAudioTech]

Schematic and layout is fee for all (when it is finished). I announced other projects in a video not too long ago.

[@PeterKese]

The way I do it is to use the both channels of the signal generator. Both channels have 1 Khz main signal, but only one of them has the 1% 4.5KHz guard signal. I let one one with the 1% guard signal through the AMP, while the other I subtract after the attenuation. Then all that remains after subtraction is the 1% guard signal plus harmonics and noise (so your noise levels can be compared to the 1% guard signal). In addition most of the weird noises should subtract as well. In my case part of the 2 Khz harmonic was generated by the signal generator so that went away after subtracting two signals from the same source. Plus now you're 40db down (the 1% is the largest signal) and you can measure that even on scope.

[@jimmydburrell]

Very interesting. Are you doing this within the ARTA software or some other software tools or do you have some dedicated hardware for this?

[@PeterKese]

@@jimmydburrell I have a Steinberg USB audio interface that has balanced inputs, so I just use the input circuitry to subtract both input channels (the one that goes through the Amp and the one that I use for reference). If you don't have balanced inputs, then just create a sound track with the one channel inverted - then a simple potentiometer is all you need: just match both levels until the master signal disappears. As for the oscilloscope software, any will do. I have been playing around with writing one myself - it works in web browser uses web audio interfaces for input and output. Albeit when I was playing with that, the web-audio standard was not completely supported in all browsers. I guess now (a year later) I should revisit that project. The good part was that I could write my own signal generators and mixers have them both in the same interface with the scope. Otherwise the whole point of that 'subtraction' approach is that after subtracting two signals, your 4.5 KHz 1% (or 0.1%) is the new reference signal, whereas the main 1 KHz signal is nullified. So you get much more dynamic depth - at 40 or 60 db down, even a cheap sound input will give you quite good accuracy.

[@jimmydburrell]

@@PeterKese Thanks for the tips. Inverting everything but the 1% test signal should work for me as a subtraction method, leaving nothing but the distortion. Great idea.

[@andymouse]

This is fascinating stuff...cheers.

[@Gengh13]

Those tones are 1khz from 44.1khz, maybe some mixing between a 44.1khz clock (DAC clock?) and your output signal, I bet if you generate a 2k test signal those spikes will be at 42.1 and 46.1khz.

[@slap_my_hand]

That "audio analyzer" probably just has a USB DAC / ADC with some filtering on the power rails in it. If you want to get more than 120 db signal to noise, you'll need toput a bandpass filter on the output, separate the fundamental from the harmonics on the input and measure them separately.

[@JelckedeBoer]

Interesting stuff. How did you setup/make the attinuator part of this setup?

[@JohnAudioTech]

A voltage divider using resistors.

[@x86FTW]

Those peaks in the 40+ Khz range might be due to having the sampling rate too high as it seems to be around the nyquist frequency of 96Khz sampling rate. Try setting 44.1 or 48Khz as the sampling rate.

[@clee2423]

what kind of attenuator do you use, and what the value of the attenuator for your amplifier, i have a amplifier with lm3886 chip, with + - 35v supply, what value of attenuator should i use

[@JohnAudioTech]

I would use metal film resistors at the appropriate ratio needed. I will cover this in the next video.

[@clee2423]

@@JohnAudioTech great !!!, my amplifier project almost finish, just waiting for my transformer to arrived, and hopefully i can do the measurement soon, even though its not really the best way to do it, but its good enough for me, hopefully my PC soundcard have a pretty low noise profile, and i'm excited to see your next video about the measurement technique, btw nice video, and hopefully i can sent you some of amplifier pcb that designed by my fellow countryman, i'll try to include the parts too, if i cant, i just send the PCB and the BOM along, so you can try to build it and do a measurement too

[@tc-bladeofgrass6719]

Hey John, thanks for another fantastic video, I've been following this project from the start. I don't suppose you could help me out, I'm looking for a simple circuit for an attenuator with the protection circuit that you mention in this video.

[@andymouse]

Probaly a bit late but they show you how to make one in the ARTA manual.

[@tc-bladeofgrass6719]

@@andymouse hey thanks for your reply. Yes a bit late, i solved this issue by using a DAC with a dedicated volume control with the amp which I'd built

[@micomrkaic]

Using your music player and your pre amp is not a good idea to generate the test signal. You should use ARTA's internal generator. It is much more accurate.

[@JohnAudioTech]

I tried a loop back direct from output to input but was seeing spikes at higher frequencies.

[@micomrkaic]

@@JohnAudioTech How high were those frequencies? What was the THD of the loop back signal?

[@jimmydburrell]

I have the same issue(s) with my sound card, so I am doing the same thing and using an external source.

[@SianaGearz]

@@JohnAudioTech I'm thinking on the MP3 player at least use a 48KHz file. This way you can avoid intermodulation byproducts of resampler from 44.1 KHz to 48KHz.

[@ВиталийГоловач-щ1д]

There is a free program "RightMark Audio Analyzer" with a accurate results of measuring.

[@AlienRelics]

Not free, shareware.

[@ВиталийГоловач-щ1д]

@@AlienRelics RMAA 6.4.5 is available for free, the RMAA 6.4.5 PRO is also available with a number of features not present in the freeware version. This I read from a program site.

[@marttiinnanen4911]

FYI there is an excellent free tool called Room EQ Wizard that allows you to do audio measurements. It's far more useful than ARTA.

[@lightbit553]

You could try Room EQ Wizard.

[@aksela6912]

Have you tried RMAA - RightMark Audio Analyzer? I remember when i was active on Head-Fi that whas what most of the amateur amplifier designers were using. NwAvGuy did a write-up on it back in '11: nwavguy.blogspot.com/2011/02/rightmark-audio-analyzer-rmaa.html

[@johnyang799]

Finally

[@martindz8457]

I usually use the RMAA program. audio.rightmark.org/products/rmaa.shtml

[@mansionese]

20:13 smells like dragon's breath :v :v

[@1pcfred]

That's the trouble with electronics. There's too much expensive gear involved.

[@fredfabris7187]

Paul Frederick it is a fact that new gear is expensive, but for most audio diagnostic work 50-30 year old test equipment is good enough. Don’t be fooled that you need the high price stuff

[@1pcfred]

@@fredfabris7187 that's all well and good when it comes to audio electronics I suppose. While I'm mildly interested in audio gear it is not my main focus. Combining digital and power electronics is.

[@aljenkins9748]

....@11:41 , believe 60dB would be 1/10 of one per cent for the pilot signal . . ...... ......... shalom,a.j.

[@aljenkins9748]

....my Apology - - i see ur corrected presentation @12:35 . . .......shalom,a.j.