Beta Test: Multitone Loopback Analyzer software

[Rantapossu]

This is my CPU load (of 4 measurements in a row) with the previous version, swithing between the Spectrum view and the Waveform view nonstop. No effect on CPU load:

This is with the new prioritizated version. I started with the Spectrum view and switched to the Waveform view after 2 measurements (The red arrow). A huge improvement on CPU load:

Great job! Thanks a lot!

 

[Sokel]

Out of curiosity, can you please try this version? I added the prioritization for what's on the screen, so hopefully the waveform/oscilloscope view will be less jerky:

https://app.box.com/s/ue7ll9xmvwogst817x2l1xg09opvgy47

MAJOR improvement for my quantum,nuclear,ancient machine.
I mean it feels so!

The only minor flaw is that this text appears on the chart (same height where the Average counter is) during the measurement and then disappears when it's done so it's clean.

 

[BKDad]

Is there a function for observing the distortion waveform of a single tone test signal? Something like this, as shown in a recent Stereophile amplifier measurement:

Linn Climax Distortion Waveform

This would require a software generated notch to completely eliminate the fundamental from view. This is obviously distinct and separate from the need for a hardware notch prior to even a very good ADC (like a Cosmos ADCiso, for one) to minimize distortion generated within the ADC itself.

I think AP can do such a measurement, but they have a really good analog notch built into the hardware and other fancy hardware tricks.

A measurement like this would be helpful in identifying problems crossover artifacts in amplifiers.

I searched for this, but didn't find any information on the subject. That might just be revealing my lack of skill in searching...

 

[pkane]

Is there a function for observing the distortion waveform of a single tone test signal? Something like this, as shown in a recent Stereophile amplifier measurement:

Linn Climax Distortion Waveform

This would require a software generated notch to completely eliminate the fundamental from view. This is obviously distinct and separate from the need for a hardware notch prior to even a very good ADC (like a Cosmos ADCiso, for one) to minimize distortion generated within the ADC itself.

I think AP can do such a measurement, but they have a really good analog notch built into the hardware and other fancy hardware tricks.

A measurement like this would be helpful in identifying problems crossover artifacts in amplifiers.

I searched for this, but didn't find any information on the subject. That might just be revealing my lack of skill in searching...

It's not available yet Funny thing, that was the exact next thing I had planned to implement when I got side-tracked on some UI changes. I'm almost done with these, and so should get back on track, hopefully very soon! There are some limitations of how this will work, though. For example, using two devices with different/unsynced clocks will not work well. Using different sample rates for in/out will also not work well. But I'll see what I can do.

 

[BKDad]

It's not available yet Funny thing, that was the exact next thing I had planned to implement when I got side-tracked on some UI changes. I'm almost done with these, and so should get back on track, hopefully very soon! There are some limitations of how this will work, though. For example, using two devices with different/unsynced clocks will not work well. Using different sample rates for in/out will also not work well. But I'll see what I can do.

Well, that would certainly present limits for people who separate ADCs like the COSMOS ADCiso or want to use a ultra low distortion oscillator for the stimulus signal.

(Here is where I get to publicly get to display my ignorance - again)

The way I've seen this kind of thing handled in other applications is in one of two ways.

One is to use an LMS notch filter algorithm to selectively kill the fundamental. Figuring out what the fundamental frequency should be fairly easy.

The other is to use a digital phase locked loop to match an internal NCO to the test fundamental in both amplitude and (anti-)phase, so that when it is summed with the test signal, the fundamental gets cancelled. Of course, this takes very careful matching of the phase and amplitude to get a complete notch, which is what would be needed for an amplifier with -120 dBc distortion products.

I guess the two approaches are pretty similar. Just a matter of what is correlating to what.

I have no idea how difficult/impossible/desirable either approach would be, because I am not a software guy. It's just lines of code, right?

 

[pkane]

Well, that would certainly present limits for people who separate ADCs like the COSMOS ADCiso or want to use a ultra low distortion oscillator for the stimulus signal.

(Here is where I get to publicly get to display my ignorance - again)

The way I've seen this kind of thing handled in other applications is in one of two ways.

One is to use an LMS notch filter algorithm to selectively kill the fundamental. Figuring out what the fundamental frequency should be fairly easy.

The other is to use a digital phase locked loop to match an internal NCO to the test fundamental in both amplitude and (anti-)phase, so that when it is summed with the test signal, the fundamental gets cancelled. Of course, this takes very careful matching of the phase and amplitude to get a complete notch, which is what would be needed for an amplifier with -120 dBc distortion products.

I guess the two approaches are pretty similar. Just a matter of what is correlating to what.

I have no idea how difficult/impossible/desirable either approach would be, because I am not a software guy. It's just lines of code, right?

Actually, you're correct, a notch would be simple enough. It's just that it would only work for single and multi-tone test signals and not music, white noise, chirps, etc. But, that may not be such a big limitation for most measurements. I'll think some more before I do anything

 

[Music1969]

If I measure left and right channels of a DAC at the same time, how do I get get measurements for each channel separately?

Seems everything showing in the results box (like THD , IMD etc) is for one of the channels?

I know I can measure each channel separately but there is also the option for both channels at same time

 

[Sokel]

If I measure left and right channels of a DAC at the same time, how do I get get measurements for each channel separately?

Seems everything showing in the results box (like THD , IMD etc) is for one of the channels?

I know I can measure each channel separately but there is also the option for both channels at same time

You can press the results box up right and see them all.

 

[BKDad]

Actually, you're correct, a notch would be simple enough. It's just that it would only work for single and multi-tone test signals and not music, white noise, chirps, etc. But, that may not be such a big limitation for most measurements. I'll think some more before I do anything

So, a real time version of DeltaWave?

 

[pkane]

So, a real time version of DeltaWave?

That was the idea

 

[BKDad]

That was the idea

Personally, I have a hard time interpreting what I see with distortion analysis of somewhat random waveforms, like music and noise, in the time domain. But, that capability could really be useful for looking at memory effect type distortions, which don't necessarily show up in "normal" distortion testing. Whether that needs to be real time analysis or not, is hard to say. But, if you can do it, great! If nothing else, it'd rattle the hornets' nest...

If that functionality is too hard, perhaps a specific test waveform for memory effects could be added to MTA to simulate what music might be like. I'd guess that a fixed waveform might be easier to design a cancellation mechanism for in software. Might be as simple as an impulse. Dunno.