REW Beta Release Changes in REW V5.20 beta 9

[John Mulcahy]

V5.20 beta 9 has a few minor changes and some bug fixes, heading towards the V5.20 release. I'm not planning any further major changes until V5.20 is released.

The warning threshold for the detected timing reference level has been reduced to -45 dBFS (from -30dBFS) and is only shown once in a session. The previous threshold was too high for mics that are set up for measuring higher SPL.

A window selection has been added to the stepped sine dialog. Rectangular (equivalent to no window) is recommended when the generator and input are on the same device as that provides the best frequency resolution. If the input and output devices are different their clock rates may also be slightly different, in that case a window may be required to prevent spectral leakage - for high dynamic range measurements Blackman-Harris 7 or the Dolph-Chebyshev windows are recommended.

There is now an option in the EQ window's filter adjust graph controls to show or hide the filter numbers along the top of the graph.

I've added a warning in the help about using long sweeps if the input and output are not on the same device due to potential problems from sample clock differences, which causes spreading of the impulse response.

There is now a soundcard input impedance setting on the Measure dialog for impedance measurements, entering the value from the card's specification sheet improves accuracy slightly although the effect is small.

View preferences have been added to choose whether the SPL meter, VU meters and signal generator stay on top of other windows (they do by default).

Here are the bug fixes in this release:

• Output level control didn't increment or decrement properly on the volts scale for small voltages
• If a measurement was opened while on the waterfall or spectrogram graph the waterfall or spectrogram would be generated but not displayed
• Timing signal detection could fail despite good signal level
• Waterfall and spectrogram generation could be incomplete on some devices
• Stepped sine did not operate correctly with 0% overlap for FFT lengths above 64k
• UMM-6 volume compensation was incorrect on macOS
• SPL limit on Measure dialog couldn't be adjusted

 

[Henry Seiden]

Thanks.

 

[Ultrasonic]

Working well so far, and although this was presumably unintentional my miniDSP window now opens in front of the REW measurement window too .

 

[keantoken]

I can't set the rest interval in stepped sine mode to more than 5 seconds. I need it to be at least 20 seconds. And who would intentionally use a rest interval of 1ms? Did you see my other comments in the spectral leakage thread?

Furthermore, during rest periods REW seems to output a very low level heavily distorted sine wave. I don't think it interferes with the measurement but it's disturbing.

At any rate, thanks a lot. This has been a great help.

 

[John Mulcahy]

I've changed the silence timer to be in seconds for the next build, allowing up to 60 seconds. The signal during the silence period was the next tone at -96 dBFS, but I've changed it to -150 dBFS.

 

[Breeman]

Is there a way to export or save the "Predicted" graph as text in the EQ section of the program? I think it would be a great addition to the "Save filter coefficients to file" and "Export filter settings as text" options that are already there.

 

[John Mulcahy]

No, though it can be obtained in a roundabout way: export the filter impulse response as WAV, import it, use trace arithmetic to produce A*B of the measurement and the filter response.

 

[Breeman]

No, though it can be obtained in a roundabout way: export the filter impulse response as WAV, import it, use trace arithmetic to produce A*B of the measurement and the filter response.

Thanks John.

 

[Stephen Turrisi]

Loving the new updates! I am new to the group here, so I apologize if this has been covered, but is it possible to tweak the screenshots to allow for a light background with thicker traces?

 

[John Mulcahy]

is it possible to tweak the screenshots to allow for a light background with thicker traces?

Can you explain further what you would like to see?

 

[Stephen Turrisi]

John, Thanks for asking. I enjoy using your software and we often recommend people to use it to tune systems. I often will take a screen shot and use it in a presentation. Initially I had a dark background and that can be a bit problematic when presenting. Additionally, the thickness of the measurement traces can be tricky to see as well.

Since my initial comment, I have found how to turn off the "dark background' and I see we can adjust the font size. I was hoping for more flexibility in the thickness in the traces.

I did notice a weird glitch in the software. I had noticed it in the previous two beta releases and it is still there, so maybe I should mention. Occasionally, the menu bar options do not always show up. The "REW" menu is there all the time, but the "file", "tools", "preferences", etc... are not always visible. I can mouse over them and the menu does drop down, but the words along the menu bar do not show up.

Thanks again for great software and support!


Steve

 

[Chuck Zwicky]

I've been having a peculiar issue for the last two betas. I cannot get the Timing Reference Output to produce any signal.
I've tested this with several interfaces and drivers here, including the Mac Pros built in outputs (digital and analog, even the internal speaker). The main "Output" channel can be changed and the signal from the REW Generator follows the assignment, but the Timing Reference Output does not appear at any output.

As a workaround I have used a "Y" cable at the normal output to feed both the Device Under Test and the loopback input

OSX 10.13.6

REW 5.20 Beta 9

 

[John Mulcahy]

Which OS are you using, Steve? Can you grab a screenshot when you next see that odd menu behaviour?

 

[John Mulcahy]

Are you using loopback or acoustic timing reference as the timing mode, Chuck?

There was a bug fix in beta 7 relating to that:
Bug fix: Sig gen output was on meas and timing ref outputs for all signal types when timing ref mode was loopback, should only be for meas sweeps

 

[Chuck Zwicky]

Hi John,

Thanks for the reply. I've only ever used Loopback as a timing reference.
I use the generator to calibrate and set up my levels on the hardware I'm testing.

Did the "bugfix" prevent the generator from routing to the timing reference output?

Also when I am in the measurement window and I select "check levels" I get a short burst of white noise on the meter, but no signal I can use to set the levels on the gear I'm testing.

 

[John Mulcahy]

I use the generator to calibrate and set up my levels on the hardware I'm testing.

Did the "bugfix" prevent the generator from routing to the timing reference output?

Also when I am in the measurement window and I select "check levels" I get a short burst of white noise on the meter, but no signal I can use to set the levels on the gear I'm testing.

Yes, the fix was to only send the measurement sweep signal to the timing reference output, since that is the purpose of that output. Can you describe your usual connections? The timing reference output and loopback input would usually only carry the timing signals for a measurement sweep and wouldn't pass through other equipment. Everything else goes out on the selected measurement output and is captured on the measurement input, including the level check signal.

 

[John Mulcahy]

I was hoping for more flexibility in the thickness in the traces.

I've added a trace thickness setting on the dialog for the next build.

 

[Stephen Turrisi]

Which OS are you using, Steve? Can you grab a screenshot when you next see that odd menu behaviour?

macOS 10.14.5.

Three versions of the image: First is with a 'floating window'. I then closed the window and you can see the second image has no menu options. But, in the third image, you can see that I can 'show' the options, but not see the menu header.

 

[John Mulcahy]

macOS is a little bit special, the menu bar belongs to the main window, when you are on other application windows it disappears. I've never seen it stay missing when the main window regains focus though, that would appear to be a Java runtime bug. Having said that, I thought the odd menu behaviour was just how things were on macOS, but more fool me, it turns out there is a way to keep the menu visible for all the application windows, so I've added that for the next build. Hopefully that will prevent the issue.

 

[Chuck Zwicky]

Yes, the fix was to only send the measurement sweep signal to the timing reference output, since that is the purpose of that output. Can you describe your usual connections? The timing reference output and loopback input would usually only carry the timing signals for a measurement sweep and wouldn't pass through other equipment. Everything else goes out on the selected measurement output and is captured on the measurement input, including the level check signal.

Yes, my usual test setup is to route the assigned timing reference output channel to the loopback input using a short patch cable. The main output goes to the device under test and back into the main input from that device's output.

Usually when I launch REW I open the meter and the generator to set the operating levels for the interface and the device.

Over the last few betas I have been unable to do this at all.

 

[John Mulcahy]

Odd. Is it possible the loopback and measurement connections or channel selections got swapped over? The signals are there.

 

[John Mulcahy]

Chuck, when you say "set the operating levels for the interface and the device" are you relying on the loopback connection to set the interface levels?

 

[Chuck Zwicky]

Chuck, when you say "set the operating levels for the interface and the device" are you relying on the loopback connection to set the interface levels?

I will typically send a signal out both outputs using the generator. The generator level set to the same level as the measurement signal. I will then adjust the input level on both the loopback and the measurement input so that they match.

 

[Chuck Zwicky]

Odd. Is it possible the loopback and measurement connections or channel selections got swapped over? The signals are there.

I've tested every possible routing and patching. Initially I thought there was no signal at the odd number outputs, but this wasn't the case.

 

[John Mulcahy]

I will typically send a signal out both outputs using the generator. The generator level set to the same level as the measurement signal. I will then adjust the input level on both the loopback and the measurement input so that they match.

You can select a secondary output on the generator if using ASIO drivers (image below) or use L+R as the output channels if using Java drivers. If that doesn't work let me know and I can restore the previous behaviour.