Analog Disovery as Audio Analyzer

[jaxbird]

First off, I know a lot of people on this board got some some misconception that audio is not engineering and everything sounds the same due to many of Dave's comments on the subject. If you feel this way, feel free to skip this thread.

Anyway, there is a lot of interesting engineering associated with audio. Personally I wanted to do some measurements on a couple of amplifiers using the software supplied with my Digilent Analog Discovery unit, but I really find it somewhat limited for this purpose, sure it will do most measurements, but it's very time consuming and the supplied software is a little buggy, not always showing the truth.

I am aware using a PC soundcard is an option (even up to 24 bits/low bandwidth) but I find the sound card option being somewhat limited as you don't know what level you are testing at and often very fiddly to configure.

The Analog Discovery is "only" 14 bits, so limited on the dynamic range, but on the positive side it's got up to ~10MHz+ bandwidth. Anyway, as I got some experience with the API from other projects, I have decided to write some dedicated software for this purpose.

Primary purpose is amplifier measurements.

[jaxbird]

Here the screen to measure THD+N vs frequency at a specific power/output level:



Note, minimum THD+N is about 0.01% due to limitations in the Discovery unit. But unlikely anyone can identify distortion that low anyway, especially considering even the best speakers distort 0.1% +++ in an anechoic chamber without any reverberations. Living room conditions is unlikely to every be below 1% distortion even a low levels.

More important is the ability to analyze the harmonics causing the distortion, generally even harmonics are considered much more pleasant than odd harmonics.

[Lightages]

Who said that audio electronics is not an engineering topic? It is very much so! I think you are mistaking the disdain here for woo woo claims in the audiphool realm as a disdain for all subjects about audio. Not so.

Looks like you have a winner there with that set up. The Analog Discovery is a great bit of kit and it is very capable of doing many things for a cheap price. I could only dream of having those capabilities when I was younger.

What amplifier are you measuring in that image?

[jaxbird]

Who said that audio electronics is not an engineering topic? It is very much so! I think you are mistaking the disdain here for woo woo claims in the audiphool realm as a disdain for all subjects about audio. Not so.

Looks like you have a winner there with that set up. The Analog Discovery is a great bit of kit and it is very capable of doing many things for a cheap price. I could only dream of having those capabilities when I was younger.

What amplifier are you measuring in that image?

Thanks, I agree, unfortunately some seem to lack the insight to discern the difference e.g. I've seen using a film capacitor instead of a ceramic as coupling being called audiophoolery, instead of common sense.

Anyway, thanks for the positive comment, much appreciated.

This is a measurement of a tiny little class D amplifier based on a Tripath chip, I will post more details.

[jaxbird]

Here measuring the output level vs % THD+N, useful to know how many watts you can expect from an amplifier with a specific load:

[c4757p]

Who said that audio electronics is not an engineering topic? It is very much so! I think you are mistaking the disdain here for woo woo claims in the audiphool realm as a disdain for all subjects about audio. Not so.

Thanks, I agree, unfortunately some seem to lack the insight to discern the difference e.g. I've seen using a film capacitor instead of a ceramic as coupling being called audiophoolery, instead of common sense.

I second this. It's pretty hard to bring up any discussion of audio quality and design without the "everything is bullshit" brigade coming out. It's a cheap and easy way for people with a limited knowledge of engineering to look smart to their peers.

Subscribed, not so much for audio as for Clever Use of Analog Discovery. It's a neat little doohickey

[jaxbird]

I second this. It's pretty hard to bring up any discussion of audio quality and design without the "everything is bullshit" brigade coming out. It's a cheap and easy way for people with a limited knowledge of engineering to look smart to their peers.

Subscribed, not so much for audio as for Clever Use of Analog Discovery. It's a neat little doohickey

Thanks, appreciate your contribution, hope it will help define the border between actual engineering and foolery, it's really a difficult topic as often some foolery claims are indeed somewhat true at high frequencies, but completely negligible below 20kHz.

[jaxbird]

This is the board I'm testing here: (it's a small class D amplifier)




This is the test setup:

[jaxbird]

Short video demonstrating the functionality implemented so far, it's still work in progress:

[mtdoc]

Very cool. Any chance you'll be releasing your "Audio Analyzer Suite"?

I just recently upgraded to the Waveforms 2015 software (now available for Linux, MacOS and ARM!) .  I like the changes they've made.  Your applications seems like a very useful addition.

[jaxbird]

Very cool. Any chance you'll be releasing your "Audio Analyzer Suite"?

I just recently upgraded to the Waveforms 2015 software (now available for Linux, MacOS and ARM!) .  I like the changes they've made.  Your applications seems like a very useful addition.

Sure, as soon as I am somewhat happy with it . This is just a very early version. I do want to add a small plug in PCB design. The Discovery unit max input is +/- 25V, that is a bit limited for testing larger amplifiers, so adding a small board with a programmable divider is needed. Plus perhaps some phono jacks or just screw terminals and perhaps some added protection to make sure the unit is never damaged.

I haven't tried the Waveforms 2015, my current version is 2.7.5 and it's got some bugs, not so obvious ones, but I've noticed it doesn't always display the spectrums correctly depending on settings.

My goal is to make it really easy to do audio measurements.

[miguelvp]

Nice, Also I'm going to upgrade to Waveforms 2015 since I saw the new Analog Digital Discovery 2 and had a link to the updated software, but have not had a chance.

[Fungus]

I second this. It's pretty hard to bring up any discussion of audio quality and design without the "everything is bullshit" brigade coming out. It's a cheap and easy way for people with a limited knowledge of engineering to look smart to their peers.

OK, I'm here now. What audio bullshit are we discussing today? My $100 system sounds fine to me.

[moffy]

Really nice plots and board. Have you tried THD+N vs frequency for your D Class amp yet? That would be interesting. Also even though it is 14 bits can't you oversample to increase your resolution?

[rjeberhardt]

Can you do intermodulation distortion with that kit?  I've often thought that intermodulation is more significant than THD especially when you see how THD rises at higher frequencies.  Music is almost never just a single tone.

Russell.

[RobK_NL]

This is the board I'm testing here: (it's a small class D amplifier)

It would be interesting to see if you can actually measure any difference between the two channels, seeing as how you've got a broken filter inductor on channel 2 

[c4757p]

Can you do intermodulation distortion with that kit?  I've often thought that intermodulation is more significant than THD especially when you see how THD rises at higher frequencies.  Music is almost never just a single tone.

Russell.

That should be possible, in theory.

[jaxbird]

Can you do intermodulation distortion with that kit?  I've often thought that intermodulation is more significant than THD especially when you see how THD rises at higher frequencies.  Music is almost never just a single tone.

Russell.

That should be possible, in theory.

Indeed, should be doable, and a feature I surely will add. The Discovery unit includes a dual 100Msps DAC and the API exposes functions to produce regular waves (sine, square, triangle plus modulation) but not multitone directly, however it is possible to pass a 4k sample buffer and have it played back at a specified sample rate, so the challenge would be producing a continuous/repeating buffer with 2 tones without glitches. Should be doable with the common 19+20kHz test signal.

Then just display the spectrum from e.g. 20Hz to 30kHz and it should give a good view of the intermodulation distortion performance.

Another option would be to just combine the two waveform channel outputs with some resistors to form the dual tone test signal. (this I have tried already and it works fine)

If anyone got a good formula for creating a continuous/glitch free buffer with 2 or more frequencies, I'd be grateful as it would save some time

[jaxbird]

Nice, Also I'm going to upgrade to Waveforms 2015 since I saw the new Analog Digital Discovery 2 and had a link to the updated software, but have not had a chance.

Thanks, Just installed the Waveforms 2015 version 3.1.5, the UX is better and it's easier to access/switch between instruments. Have not done a lot of testing, but it appears they have actually removed some features  , or maybe I just can't find them.

Really I just wanted to know if they made any API changes, but everything I've written is compatible, so it seems no API changes, the API already had the features needed for a variable output PSU as now available in Discovery V2.

[diyaudio]

First off, I know a lot of people on this board got some some misconception that audio is not engineering and everything sounds the same due to many of Dave's comments on the subject.

my alias gets me into shit every time with some members around here, yet i have no association with audio only.

[rjeberhardt]

]
Indeed, should be doable, and a feature I surely will add.

Thanks. Could give some interesting results.

Russell.

[jaxbird]

Spent some hours today adding new features, first frequency response, pretty straight forward. Will show scale either referenced to input signal or referenced at 1kHz level.

[jaxbird]

Also been playing around with the intermodulation distortion feature, this should give a good view of the IMD performance:




0 dB is the level of the two test tones.

[fivefish]

jaxbird, that is awesome. If you keep this up, and release this software I think I will be buying an Analog Discovery and then send some $$$ your way.

Please make this a "poor man's AP setup"

[fpliuzzi]

Awesome is a good word for your audio analyzer suite. I was perfectly happy with the capabilities of my Analog Discovery, but now I'm really glad that I own one (although my trusty HP334A is looking a little nervous that it could be departing my main test bench soon to free up some much needed space).

Also, a quick thought about your IMD feature. If you see some value in this idea, could the amplitude of the two test tones be made independently adjustable so that you could, for example, test a vintage piece of audio gear using the older SMPTE standard (frequencies of 60Hz and 7KHz, mixed in a 4:1 ratio) if that was desired?

That could make an interesting way to compare the test results of the various IMD standards by supplying the appropriate test frequencies and amplitude ratios.

I tip my hat to you... keep up the excellent work.