Can REW measure a pre-recorded sweep?

[Jim the Oldbie]

Howdy Folks,

First of all, a huge thank-you to everyone involved for some of the most useful audio software available.

Second: I've done a fair amount of searching on this, so I hope I didn't miss the answer somewhere. I've discovered that REW works great not only for microphone-based measurements, but also for directly measuring frequency response etc. of various pieces of gear like amplifiers. My question: Is there some way to do this "asynchronously," that is, to test something like a recording device (mp3 player, etc.) by recording an REW-generated sweep to a file, then playing it back? The recording part is simple enough, but I don't see a way to initiate a measurement separately from a sweep. Is this possible, and if so, what is the procedure for initiating a sweep measurement when the sweep is generated independently?

Thanks very much for any assistance.

-- Jim

 

[Wayne A. Pflughaupt]

I’m not sure exactly what you’re trying to measure: The file itself, or the recording device? If the latter, all you have to do is run the loopback routine with the recorded added to the signal chain.

Regards,
Wayne

 

[Jim the Oldbie]

Hi Wayne, sorry for not explaining better. Here's what got me started on this:

I have a Sansa Clip Zip mp3 player, running the excellent Rockbox alternative firmware. One of Rockbox's cool features is a multi-band parametric EQ. I've been curious about the accuracy of the EQ filters in this feature, and whether they're constant- or variable-Q, things like that.

I've used REW in loopback mode to measure equalizers & other gear before, but they were all capable of simultaneous input and output. To correctly measure the Clip Zip, it would be necessary to make an accurate (i.e. bit-perfect) soundfile recording of REW's cal sweep output, then see if REW could measure the soundfile playback from the DUT (instead of its own real-time generated sweep) during the loopback cal test.

Anyway - after messing around with this some more, I've figured it out. Here's a shot of my Clip Zip playing back a .wav file of REW's cal sweep, with the Rockbox EQ bands set for illustration. Turns out the EQ in this tiny little thing behaves quite nicely!

I realize that I may be in a small minority of people who had trouble figuring out how to do this. (Wouldn't be the first time, heh.) If that's not the case, I could post a more detailed description of the process if anyone's interested.

-- Jim

 

[Leonard Caillouet]

Please do! It never hurts to have a different perspective on documenting a process.

 

[Jim the Oldbie]

Hmm... Okay, so before typing this up, I thought I'd better upgrade to the latest REW version on this machine (I was still at 5.01), and unfortunately now my little trick has stopped working. I'll get back to you as soon as I figure out what's changed. :o/

-- Jim

 

[Jim the Oldbie]

It turns out my motherboard audio is having problems with the 48K sample rate selection in the latest version of REW. But everything seems OK at 44.1K. I'll try a different (better) USB soundcard when I have a bit more time. Meanwhile, here's the procedure I came up with for measuring a playback-only device with REW:

First step was to get a clean soundfile of a full-bandwidth REW measurement sweep, including the acoustic timing reference "chirp." This can be done easily, right from the tone generator in REW. Just select the "Meas. Sweep" output, set your desired output level, then click the "WAV" button. Select Left or Right channel output (or both), and be sure to tick the box marked "Add timing reference signal." The next box is where you set the bit depth, and last is the file save dialog. Done and done!

Next, I loaded the file into my Clip Zip for playback. If your device under test won't play WAV files, you'll have to convert them before uploading to the device of course.

Once your DUT is connected to your sound card's inputs you can then use REW's Measure procedure to run the test. Note that since we're dealing with a playback-only setup here, there are a couple of things to note.

The first is the "Check Levels" button. Unfortunately, without the simultaneous input & output of a regular test setup, the sound card input doesn't hear the test noise it's producing at its output when this button is clicked, so this feature doesn't really work here. You'll just have to click the "Start Measuring" button a time or two until you have the levels where you need them. Not a huge deal, just takes a little more time.

Another thing to remember is that the sweep frequency range and FFT length must match the settings used for the recorded test sweep, or the results will be unpredictable at best.

Also of note: The soundcard and microphone cal settings need to be turned off for this type of test. The microphone cal is obvious, but the soundcard perhaps less so. The issue here is that REW produces a sound card cal file via its loopback test that is the sum of its D/A and A/D characteristics. Normally, this is just what we need. But with this test we are only using the A/D portion of the soundcard, so the cal file is inaccurate. I haven't figured out a way around this, except to just use a quality soundcard that performs well enough without a cal file. These days, this can be achieved at reasonable cost, provided we're not attempting some sort of lab-grade measurements.

Once your level is set, make sure the "Wait for timing reference" box is checked. Then click the "Start Measuring" button, then start playback of your recorded sweep file. REW will wait for the sync chirp, then take the measurement. From here on, it's pretty much the same as a conventional test.

Note that if you've set the "Listen to this device" box (in Windows) for your soundcard's input, you can hear the output of your playback device while running this test. And if REW is set to output its test sweeps (which are not actually being used here) to your speakers, you can hear both sweeps at once.

Any suggestions, corrections, etc. are welcome.

-- Jim

 

[John Mulcahy]

You can save REW's measurement sweep to a WAV file from the signal generator, including a timing reference if desired.

 

[Jim the Oldbie]

Yikes! That'll teach me to keep up with the latest version!

Thanks for this John. I'll edit my writeup above to reflect this.

Regards,
-- Jim