REW Beta Release REW macOS DMG (includes private Java 8 JRE) V5.20.3

[John Mulcahy]

Thank you so much, please find the attached project, and the .plist file.

Change your IR windows so that the left window is shorter then the right, and/or update to the current REW version. Why the very large left hand window?

 

[Chuck Zwicky]

Can you attach the mdat files for the responses that were cleaner after changing the timing offset, as posted in your "Before and after (offset for clarity)" image? Why aren't you running the current REW version?

Sorry, I didn't save that session, but I've attached another with both good and bad results:



I downloaded what was supposed to be the current release..?

 

[John Mulcahy]

That is the current release, but the measurements you attached appear to have been made with one of the V5.20 beta releases, they show "V5.20" on the info dialog, not V5.20.8

 

[Chuck Zwicky]

Change your IR windows so that the left window is shorter then the right, and/or update to the current REW version. Why the very large left hand window?

I've never touched those settings.. is it here?

 

[Chuck Zwicky]

That is the current release, but the measurements you attached appear to have been made with one of the V5.20 beta releases, they show "V5.20" on the info dialog, not V5.20.8

Strange. I downloaded the latest release, and it's the only one on my system..

 

[John Mulcahy]

Uncheck the "Set window widths automatically" box and see what happens then.

 

[John Mulcahy]

When you use the MiniFuse make sure the mix knob is turned all the way to USB.

 

[Chuck Zwicky]

When you use the MiniFuse make sure the mix knob is turned all the way to USB.

Yes, I'm well aware..! (I'm actually a professional Recording Engineer, and helped Arturia with the design of that one - including re-working the behavior of that mix knob!)

 

[Chuck Zwicky]

Uncheck the "Set window widths automatically" box and see what happens then.

The defaults of 125ms and 500ms appear, should I adjust these?

 

[John Mulcahy]

Just leave them as they are. From the impulse graph something is feeding back in quite a major way when you use the MiniFuse (although it isn't certain that was being used since the input and output device selections were both at default). The latest measurements you attached (from 16th June) also indicate they were not made with V5.20.8, so I'm not sure what to make of that.

 

[John Mulcahy]

Can you attach a zipped copy of the log files folder? It is /Users/zmix/Library/logs/REW

 

[Chuck Zwicky]

Here is the Logs folder..

It's possible that the DUT was experiencing some oscillation issues, I've never had these problems before..

 

[John Mulcahy]

Thanks for that. 5.20.8 was used from about 6:30 pm on the 17th, so after the mdats you attached. For those mdats, my best guess is the 4 kHz start frequency for the sweep caused the auto window assignment to set a large left window (which I've fixed for the next build) which in turn triggered a bug with IR delay estimation and alignment that was fixed in V5.20.5, so I don't think the same issues seen in those measurements would occur in 5.20.8.

For the noisy/oscillatory-at-HF response that cleaned up after adding some timing offset I'd need to see mdats for both cases to comment further.

I think I've found what caused the null pointer exception you saw and fixed that for the next build.

 

[Chuck Zwicky]

..my best guess is the 4 kHz start frequency for the sweep caused the auto window assignment to set a large left window which in turn triggered a bug with IR delay estimation and alignment..

I'm glad you brought that up, while I was performing my sweeps, I decided that 4kHz was the lowest frequency of interest so I began them there. I felt like I had found a bug because the sweeps took as long as a full frequency sweep did, and the progress bar would often reach 50% or higher before the 4kHz signal was output.

For the noisy/oscillatory-at-HF response that cleaned up after adding some timing offset I'd need to see mdats for both cases to comment further.

Here you go.. I just made 2 sweeps, one initial sweep with no timing offset, then one with a timing offset (see attached .mdat file)

NOTE: By shifting the IR you can clean up the "noisy" sweep after the fact, for some reason in the previous version this wasn't working here...

 

[John Mulcahy]

Measurement sweeps span an octave either side of the measurement range. The time the sweep takes depends on the length setting, regardless of the frequency span.

There is a lot of odd harmonic distortion in those measurements, looks like mild clipping somewhere.

Both the sweeps actually give the same result, the second looks different because it is being converted to a 96 PPO log spaced format (per the Analysis preference) which requires 1/48 octave smoothing. That doesn't happen on the first sweep because the peak is far enough from t=0 that the conversion would cause aliasing in the phase response at high frequencies.

 

[Chuck Zwicky]

Measurement sweeps span an octave either side of the measurement range

In my case the delay was progressively getting worse and was resulting in "completed" measurements happening mid-sweep with the resultant "ringing cliff" seen in some of the above bad examples.


Thanks again for all the informative replies, John.


PS: Yes there was clipping, I discovered too late... when I lowered the signal it went away and so did the fuzzy treble in the sweep - without the 1/48 smoothing. Thanks again..!