REW Hangs When Calibrating The Sound Card

[Mike Rivers]

I've found REW to be kind of flaky when running a sound card calibration. Level check is OK, the sweep runs, and when the sweep is complete, it hangs up in the Calculating FFTs process with the progress bar stuck at 95 to 96%. I can close the Calibration window but nothing else works. All I can do is close the program and restart. Sometimes I can go through this three or four times before the calibration procedure runs to completion. I've attached the log file from a crashed calibration session.

With or without a calibration file, the program otherwise works well.

I can't pin this down to a specific computer and interface ("sound card") since I've had this same hang-up during calibration problem with three different computers and three different interfaces, and have swapped interfaces among the computers. One computer is running Windows XP, one is Windows 7 32-bit, and one is Windows 7 64-bit. I'm using the manufacturer-supplied ASIO driver with two of the interfaces and the Java driver on the other one that doesn't have an official ASIO driver.

Any hints? Or do they all do that?

 

[John Mulcahy]

Odd, though quite an old version. Try the latest 5.19 beta 6 build.

 

[Mike Rivers]

I updated the Win7x64 version (the one I sent the log file from) with the latest build and it might have been a little better. But when it hangs up when processing the calibration measurement, I have to restart Windows in order to get it to send data to the interface again.

I'd been running version 5.17beta14 on the WinXP computer and updated that to 5.18 which hasn't crashed after several calibration runs. It's for things like this that I keep a few XP systems running.

One thing that I noticed on the Win7 system was that the interface seems to not want to run at 96 kHz sample rate in REW, though it's find at 96 kHz in other programs. I'm wondering if that Java ASIO host has something to do with it. Does that actually use the hardware's ASIO driver, or does it replace it similar to what ASIO4ALL does? It probably does use the real ASIO driver, or at least learns about the device from its driver. I can choose from any of the six inputs and outputs with ASIO selected, but only 1-2 with Java. The interface I'm using on that computer is an Alva Nanoface (nobody else has ever heard of it either). When using Java with the Nanoface, it calibrated successfully (44.1 kHz only) about 3 out of 5 tries.

On the XP system, I'm using a Mackie Onyx and other than the occasonal crash when calibrating it works fine at 96 kHz. I'll look around in the graveyard to see if I can find a Firewire card to fit in the Win7x64 box and see if the Mackie interface is any more consistent with that system than the Nanoface.

The reason why I'm fussing with this more than your usual user is that I'm writing a magazine article about DIY testing of audio (mostly recording) hardware - for the purpose of verifying performance at least close to the manufacturer's specs, filling in for incomplete specs (like "Frequency response 20 Hz to 25 kHz" without any tolerances) and learning what "HP Filter: 80 Hz" really means. REW is a great tool for this kind of probing, but I don't want to write around a program that's going to give the readers a hard time. 44.1 kHz sample rate is OK, but somebody is going to be interested in knowing what his dog is hearing even if he can't.

I don't know if anyone else ever tried this, but I can measure software plug-ins, hooking them up virtually with Virtual Audio Cable. I'm trying to find someone with a Mac who can try that trick using Sunflower or Loopback that seems to be its replacement.

 

[John Mulcahy]

The ASIO host uses the hardware's driver. If the ASIO buffer is small (the default is often quite low to keep latency down) the JVM may not be able to keep up as Java processes can't run at as high a priority as system drivers, worth trying a larger ASIO buffer to see if that helps.

On the machine you have updated to the latest build what is in the log file after a calibration failure? And what is the ASIO buffer size set to?

 

[Mike Rivers]

Good thought about the ASIO buffer size. That reminded me that the last time I used this interface was when I was reviewing a virtual bass plug-in, and I wanted to see how low I could get the latency. This driver doesn't have the usual "how many samples?" buffer settings, it has five presets. It was set for the smallest buffer size "High Speed" which is 6 ms output and 12 ms input.

This wasn't a very productive evening, however, because this time around, with that buffer setting, I couldn't get it to hang in the calibration routine after about a dozen consecutive tries. But I thought that perhaps with a larger buffer, I might be able to make it happy running at 96 kHz. the "Relaxed" buffer setting is 55 ms out and 24 ms in. But my conclusion at this point is that it just isn't going to be happy running at 96 kHz with REW. If I try to change the sample rate from 44.1 kHz in the Preferences window, most of the time it won't even show me any choices. Sometimes, probably first thing after a fresh restart, it will show 44.1, 48, 88.2 and 96 kHz, but when I select 96, the interface appears to disconnect from the program. When I check levels or start calibration, the output meter in REW reads, but I don't see an input on the interface's meters, and of course there's no output.

After coaxing it back into operation, it works with the higher latency buffer setting, but only at 44.1 kHz.

If I knew I was going to be troubleshooting, I'd have been more diligent with the log files, opening the file and adding a note as to the conditions when it hung. I should start doing that. Here are two logs from the latest version of the program with a lot of information in them. I'm pretty sure one was just a routine hang-up at the end of the calibration cycle and the other was when I was trying to run it at 96 kHz.

 

[John Mulcahy]

There isn't an inherent issue at higher sample rates, I can run a UR22 at 192 kHz without any problems. That's using the latest Yamaha Steinberg USB driver. I find that it is best to reload the driver after changing sample rate (there's a reload button in the REW soundcard preferences), I can make that automatic in the next build.

In one of those logs REW didn't get data from the soundcard, in the other I suspect the data was all zero-valued. I'd certainly like to track down any issues I can deal with inside REW though, so if you get a chance to capture logs from other misbehaviour that would be much appreciated.

 

[Mike Rivers]

There isn't an inherent issue at higher sample rates, I can run a UR22 at 192 kHz without any problems. That's using the latest Yamaha Steinberg USB driver. I find that it is best to reload the driver after changing sample rate (there's a reload button in the REW soundcard preferences), I can make that automatic in the next build.

If I had a more modern interface, I probably never would have brought this up. I have the latest firmware and ASIO driver for it, but still, it's at least two years old. Everything I have around here is old (including me) but still works. I nearly bought a UR44 a couple of months ago to put together a portable recording setup but decided against it when I recognized that I'd have to also get a new laptop computer to go with it (Yamaha drivers won't run on WinXP, and the old laptop doesn't have enough horsepower to run newer Windows) and new laptops aren't as portable as the one I'm using now.

I noticed the Reload button and tried it, but the Alva interface wasn't happy with what it does and didn't re-connect at the USB level unless I disconnected and re-connected the USB cable. And when it came back to life in REW, it was at 44.1 kHz. I think I probably should just accept this interface as "odd" and move on. The only other USB interface I have is a Behringer UC200, and that only runs at 44.1 or 48 kHz.

In one of those logs REW didn't get data from the soundcard, in the other I suspect the data was all zero-valued. I'd certainly like to track down any issues I can deal with inside REW though, so if you get a chance to capture logs from other misbehaviour that would be much appreciated.

If I run into anything else interesting, I'll pass it along. Now that I have a pretty good handle on what breaks it, I can avoid doing that, but if there's anything specific you'd like me to try, let me know and I'll do it.

If you're curious about the interface:
http://www.alva-audio.de/nanoface/en_index.php

 

[Ed Zeppeli]

Same issue here. Soundcard cal freezes at 96% calculating FFTs. Tried multiple times. Updated to latest beta etc...

 

[John Mulcahy]

Please attach some log files after a failure occurs, Ed. Their location is shown in the About REW dialog.

 

[Ed Zeppeli]

These should be them. Thanks

 

[John Mulcahy]

When it freezes does it say Calculating FFTs...Input or Calculating FFTs...Stimulus?

Edit: and are you using Java or ASIO drivers?

 

[Ed Zeppeli]

FFTs Input. Always freezes at 96%. I'm using an outboard Steinberg soundcard with Java drivers.

 

[John Mulcahy]

I think I may have tracked this down. Before doing soundcard calibration please click on the measure button and check the sweep length setting, if it is more than 256k set it to 256k. Then cancel the measurement dialog and try the soundcard calibration again.

 

[Ed Zeppeli]

It's at 512 so I'll try again in a few days. Heading away for a bit. I've been debating going to ASIO drivers. Are there any benefits when to ASIO using just 16 bit 48kHz?

Thanks much.

 

[John Mulcahy]

Not really. Only real advantage of ASIO is when using a multichannel output as individual output channels can then be selected.

 

[Ed Zeppeli]

Tried the souncard cal again just now. (related to your other comment on my system set-up thread).

It froze again even at 256k setting.