REW Beta Release REW API beta releases

[John Mulcahy]

Builds updated today (beta 54) with the following change:

• Fixed: Null pointer exception could occur when saving RTA data with harmonic phase display selected

 

[yys310]

Is there a way to measure driver current distortion by REW?

 

[John Mulcahy]

Is there a way to measure driver current distortion by REW?

I guess you could make a sweep or stepped sine measurement across a small resistor in series with the driver, on the ground side if your interface input is single-ended.

 

[sm52]

...if you have a means of generating an inverted copy of the input signal and applying it to a second input channel the cross-correlation processing with the Invert correlation reference option may allow common mode noise on the inputs to be cancelled. There is a thread on cross-correlation processing here.

I tried this option. When one input is used, the harmonics are no higher than 3%, when the second input is fed an inverted version of the first input, the harmonic levels are almost the same as the main signal. Maybe I should do something else?

 

[John Mulcahy]

Sounds like the way the inverted signal is being generated is causing some issues. Probably worth asking on the thread I linked.

 

[sm52]

Sounds like the way the inverted signal is being generated is causing some issues. Probably worth asking on the thread I linked.

Thank you. I will try. Another question. Is there a way to make a full copy of a measurement after uploading it to REW? I know of several ways to get a copy, but they all lose the left impuls window, so the distortion is no longer shown.

 

[sm52]

John Mulcahy,​

in that DAC-ADC testing thread are we talking about a loop connection? And that's all? From output loop to input1. And if you add inversion, then from output to input1 and inversion to input2. So?

 

[sm52]

If REW can handle two inputs, can it use calibration files separate for each input at the same time?

 

[serko70]

Alignment tool API very rarely throws an irrelevant error message:

The above error was caused during an impulse alignment @ 20Hz. Details below:

If of any relevance, the same error was also here:

Beta 54

 

[John Mulcahy]

If REW can handle two inputs, can it use calibration files separate for each input at the same time?

Yes.

 

[John Mulcahy]

Is there a way to make a full copy of a measurement after uploading it to REW? I know of several ways to get a copy, but they all lose the left impuls window, so the distortion is no longer shown.

Windows are not affected, but response copy does not include distortion data. If you want an identical copy of a measurement just save it and reload it.

 

[John Mulcahy]

in that DAC-ADC testing thread are we talking about a loop connection? And that's all? From output loop to input1. And if you add inversion, then from output to input1 and inversion to input2. So?

I don't really understand the question, but if you have questions about that ASR thread then ask them there.

 

[John Mulcahy]

Alignment tool API very rarely throws an irrelevant error message:

The error buffer might be full with previous errors. I'll add some code to only keep the most recent errors.

 

[sm52]

John Mulcahy,​

thanks for the answers.

 

[John Mulcahy]

Builds updated today (beta 55) with these changes:

• Changed: API error and warning message buffers now keep the 100 most recent errors and warnings, previously they stopped logging after 100
• Fixed: Impedance cal measurements did not properly override measurement sweep frequency range
• Fixed: Impedance measurement noise filters were not being applied

 

[sam_adams]

@John Mulcahy, I don't know if this is an issue or not; when opened on a Mac (13.6.7) running beta 55, the .mdat attached to this post made with beta 55 does not display the noise floor plot in the distortion graph, while the .mdat in this post made with release 5.31.3, does. Both of those .mdats were from Mac users on 12.7.6 and 10.15.7 respectively.

 

[John Mulcahy]

Noise floor capture is optional, depending on choice made on the Measure dialog.

 

[JoinUs]

when i export to impulse wav. (Left window 1.1ms Ref time 1.1ms)
If i bring it back to REW or Audacity, there's about almost ~70ms of silence in front of it.
I've been stuck on version 5.30.9 for a long time, and this export is a bit weird for both version 5.31 and beta.

left one is original.
right one is export and import .

I downloaded each version one by one, and it's exactly like this after version 5.31. (Beta too)
5.30.9 is fine.
Or is there another hidden option for exporting from version 5.31 onwards?

I'm continuing to explore options but it doesn't seem to have changed particularly (from 5.31). Of course, I trimmed it

So I saw if there was an update to this in version 5.31, and I found a part related to the ir export, but it's mono ir, and I think it's not related to this.


+++
okay.
I've read everything from 1 to 29 of the beta thread in a long time, and I've also read the help page once again.
But there doesn't seem to be any changes or quirks about what I'm reporting.
I've repeated it many times just in case I made a mistake, but it happens the same.
The official version is after version 5.31 and up to beta (5.30.9 is fine)

 

[John Mulcahy]

when i export to impulse wav. (Left window 1.1ms Ref time 1.1ms)
If i bring it back to REW or Audacity, there's about almost ~70ms of silence in front of it.

It's difficult to comment precisely since you haven't said what export options you used and you haven't shown the time axis in your clipped image. There are 3 options for the placement of t=0 for imported impulse responses, which is appropriate would depend on where t=0 is on the measurement you exported.

 

[JoinUs]

It's difficult to comment precisely since you haven't said what export options you used and you haven't shown the time axis in your clipped image. There are 3 options for the placement of t=0 for imported impulse responses, which is appropriate would depend on where t=0 is on the measurement you exported.

View attachment 73921

this setting same.
and i use set= t 0 first sample.With the same settings, it works well in 5.30.9, and all versions since then have that kind of silence added. Of course, all versions earlier than 5.30.9 work well

 

[John Mulcahy]

Unless your measurement has t=0 at the very beginning of the impulse response (unusual, though not impossible) then putting t=0 at the first sample of the import will mean the zero times for the two will be different.

 

[JoinUs]

Unless your measurement has t=0 at the very beginning of the impulse response (unusual, though not impossible) then putting t=0 at the first sample of the import will mean the zero times for the two will be different.

interesting. I found the cause. Weird...

Left and Ref time is 1.1ms (my impulse start time)

Right Window - 500ms set to fine
Right Window - 600ms set to fine.
Right Window - 700ms set to fail
Right Window - 1000ms set to fail


I don't understand why the right windowing affects the starting IR timing (because all versions prior to 5.31 always worked)

It's almost dummy data, but my impulse is about 1000ms, so I always limit my right windowing to 1000ms, and it's always been working in previous versions.
Curiously, however, the beta version from 5.31 seems to have a problem with windowing trimming after approximately 600ms.

 

[John Mulcahy]

Perhaps you could attach the mdat, since your cropped images are not informative.

 

[JoinUs]

I also tested the impulse I had, which had a shorter data. attached the file.
Check the overlay. (It should be viewed in First Sample View)
92ms is no problem.
In 729ms, we found that about 620ms of silence was added this time. There was no reason for the start timing to change, no matter how much the right windowing increased. (At least before version 5.31)

And I also tested it with regular DiracPulse.
But this is different again. There's no problem applying right windowing up to 2000ms and exporting it.
maybe guess such a bug occurs only in IRs with a little information

 

[John Mulcahy]

From the mdat I see you are using Trim IR to windows and then exporting the trimmed IR. Trim IR to windows ensures the resulting IR length is a power of 2. If that power of 2 length went past the end of the available data zeroes were prepended to the IR. I have changed that for the next V5.40 beta build. For your exports it would make more sense to use the "Apply IR windows before export" option instead of trimming the IR.