How to - Distortion Measurements with REW

[wtnh]

It is amazing how fast they seem to get snapped up on eBay. I can't find any either.

 

[kozard]

You can setup an alert on Ebay to let you know when one is listed. (See attached.)

Use the advanced options to filter out false alerts such as the power supplies.

I know I use such alerts for equipment or parts that I really need. If you just check the listing once per week for such equipment you will likely miss it. If someone really needs gear or backup gear for daily use and sets up an alert it can be gone in 5 minutes.

 

[Bonsai]

I’ve put some slides together here about getting accurate low distortion/ noise measurements using the QA401. The basic principles apply to any measurement setup

Using the QuantAsylum QA401 for accurate Audio Measurements

 

[wtnh]

Very nice! Here is an additional reference you may want to investigate. Bill Whitlock (Jensen Transformers) gave a seminar on grounding and signal noise at an AES meeting a few years ago. While not specific to instrumentation, he addresses the concepts very clearly. His slide deck (all 215 slides) is here. Worth a look.

 

[kozard]

Having more than one USB 0404 certainly changes the risk assessment - I keep looking but no luck on that aspect for me.

Unfortunately capacitance and maybe ESR can change the design performance of a dc/dc switchmode. ESR changes a lot with temp, so to have a substantial change in stability due to say a 10:1 change in ESR seems a bit concerning. It would get arduous to assess why there was a high level stability or control change when a cap is changed, but at least you can swap it back or change it again to do some basic fact finding.

The photos (links) are dead in the "E-MU 0404 USB mod wiki" so I posted over here since there are E-MU 0404 USB knowledgeable people following that thread:

E-MU 0404 usb 2.0 and some "artifacts peaks" on left channel

(I am trying to identify the purpose of two of the electrolytics. My "new-old" E-MU 0404 DAC THD improved by 2 dB on the worse channel with new capacitors. It still is a couple dB worse than my "old-old" E-MU 0404 which still has the original capacitors. The "new-old" one arrived looking pretty new physically. It might not have been used for maybe up to 10 years. The THD improve by a few dB just being powered on for a few hours and then a couple more dB when I put four new 47uF 63V Panasonic capacitors next to the DAC.)

 

[bohrok2610]

REW with Linux

What is the appropriate setup for a linux desktop if used for measurements with REW? Seems that Alsa, Pulseaudio, Jack etc. have somewhat overlapping functionalities and some don't even seem to work with REW.

 

[phofman]

Please can you specify your question more?

REW is programmed in java which supports only alsa in linux (no PA nor jackd), and enumerates cards only (i.e. devices listed by aplay -l). The alsa javasound layer does not list PCM devices (listed by aplay -L), e.g. those defined in .asoundrc. I promised to make a patch for openjdk to support listing PCM devices but honestly that plan is put on hold for now. One reason is I need to make a separate java WASAPI-exclusive connector similar to jasiohost so that REW can talk to UAC2 windows driver directly which is needed by my project. Another reason is pushing a patch to Oracle-controlled openjdk is no fun. And third reason is Oracle would not backport my patch to the legacy Java8 which REW is still using.

 

[bohrok2610]

I have only recently started to use REW in Linux audio mostly for the same reasons you mentioned (only 16-bit audio supported in Windows Java and Asio4all does not seem to work properly with higher sampling rates). My experience with Linux dates from pre-Alsa so I'm slightly confused with the current state of Linux audio. Your response was very helpful as it confirms my findings (i.e. no PA or Jackd with REW). It also explains another annoying thing with Linux audio & REW which is the lack of support for older soundcards.

Please keep me posted with the development of java WASAPI-exclusive connector. Have you looked into the existing java-wasapi libraries (XT-audio or libjitsi)?

 

[NealJ]

Sanity check

I was hoping for some input on this measurement.

I had a play around with REW a while ago after XRK originally posted but never went any further as I don't think I really grasped a lot of it. I've recently decided to have a look again as I got a Focusrite 2i2 3rd gen quite cheap. I've spent a few days playing with different cables, using the different inputs, internal and external signal generation and so on.

I've read this thread (and others) many times to see if I'm missing anything or doing anything wrong, but would appreciate some input.

I've included a measurement of a headphone amp I recently built. The testing set up is this:

REW ASIO 48kHz, 2i2 connected to desktop PC on mains supply.

Akitika oscillator -->RCA out to HPA --> TRS headphone cable to 25 ohm dummy load --> RCA to XLR--> 2i2
Voltage at dummy load 562mV RMS (checked against Siglent SDS 1202X-E scope).

Is this a trustworthy measurement (relative to the equipment used), it seems to be very good and that makes me suspicious I suppose. I want to make sure I'm doing things right before I measure my other HPAs.

 

[jan.didden]

I don't see anything wrong with it. The 2nd is about -108 below the fundamental, which is quite good, but not so low to distrust it.
If you want some more confidence, take out the hp amp and connect the source directly to the measurement inputs, so you measure the oscillator and the 2i2. That would be your baseline.

I'd say well done, both the measurement and the headphone amp.

Jan

 

[NealJ]

Thanks Jan, good to know, and thanks for the tip.

The headphone amp isn't my design, so I can't take credit for that.

 

[audio_tony]

THD looks far too low although THD+N does match Focusrite specs.

But 0.00008% THD seems a little on the low side to me.

Have you tried measuring the Focusrite in loopback mode only?

 

[NealJ]

That was the figure that made me suspicious, I haven't seen one that low before.

I've just done a loopback focusing on trying to get the lowest setting.

The first uses the internal generator with phono to XLR in.
The second is with the Akitika RCA to XLR in.

 

[audio_tony]

That's more in line with what I would expect.

Although the 2nd harmonic still seems a bit lower than I'd expect.

I would test it with Arta Steps - that will provide a plot of THD vs. frequency.

 

[jan.didden]

Why go to yet another software, with its own intricacies? REW can easily do THD vs F.

Jan

 

[audio_tony]

I'm not saying I don't trust REW but I have a lot of confidence in Arta / Steps.

Just in the same way I have a lot of confidence in WaveSpectra.

And both of the above apps will give a very quick indication as to accuracy.

 

[NealJ]

Had a quick look at ARTA, it seems quite complex.I had a go at producing THD vs Freq., never done this before, is this what you're after?

I'd like to carry on with REW first, as I've started with it. If the same needs doing in ARTA, I think I would need some help with that.

 

[jan.didden]

What I mean is, here we have a friend trying to grips with REW and he's advised to take a sidetrip to ARTA.
You may be very familiar with ARTA but that irrelevant to his adventure.

Jan

 

[jan.didden]

Had a quick look at ARTA, it seems quite complex.I had a go at producing THD vs Freq., never done this before, is this what you're after?

I'd like to carry on with REW first, as I've started with it. If the same needs doing in ARTA, I think I would need some help with that.

REW does everything ARTA does, no need to wreck your brain on two different apps.

Jan

 

[chede]

Somehow the distortion panel figures from your Akitika measurement in post #1013 don't fit together. The graph shows 3rd harmonic clearly higher than 2nd, though the distortion figures show them as almost equal, with 3rd even slightly lower than 2nd.

But I have no idea what would cause this phenomenon ...


Regards, Claas