REW Loopback test, I'm confused. Am I doing it right?

[dunkuk]

After finding no test data fro my Lexicon Alpha USB interface I thought I could try a loopback test.

I'm on a Mac and REW came up as some software that would work. I spent maybe 20 minutes trying it out and got these results.
What I'm confused about is how to set the levels. I've got the signal generator at -10db, but on the chart my signal is at +75db. Should I measure the levels with a voltmeter to get 2V RMS or something like that?

Also I'm not ure about the "Timing" and Loopback settings in the preferences, when I change the outputs it doesnt seem to make much difference.

I also noticed the Right input has an input on it at bass frequencies that gradually goes down at higher frequencies, that seems odd unless its really bad channel seperation.

The distortion chart is hard for me to understand, it seems quite high THD?

I've got the Line In/Mic input at 9 o'clock, and the output level at 12 o'clock. Turning the input up or the output up above this turned the clipping light on. I tried to adjust it for a low distortion level. I've got a short TRS cable going from Output 1 to L Input.

I was hoping for one of those charts that has a 1khz peak and then shows the harmonics to the side, but I'm doing a sweep which I'm not sure is correct.

Any help with calibration and settings would be appreciated, thanks!

 

[Foxenfurter]

I followed the advice here when testing some of my equipment a few years ago. Using Right Mark Audio Analyser.

https://nwavguy.blogspot.com/2011/02/rightmark-audio-analyzer-rmaa.html

There is quite a lot of fine tuning to get the trade off between distortion and noise about right, and you are limited by the quality of your A/D converter. I found that the old usb interface that I was using was worse than my pc soundcard.

 

[dunkuk]

Thanks RMAA looks better in that the charts are a bit clearer, its PC only though I think, I've got a Windows Laptop I could try it on though.

I managed to get the chart I was after, still not sure about the levels though.

I will read that guide. Here is the latest result.

 

[dunkuk]

OK I can see I was using the wrong window, this is much better. I've turned averaging on as well as the chart was moving up and down a lot.

 

[dunkuk]

Hmm, I measured the voltage across the output for that last chart and it's .42VRMS across the tip and ring, so .21V ground to either of the balanced outputs. The generator is set to -10db, and the volume knob is turned fairly low now, only at about 10 o'clock.

edit.
Just turned the output to full and it measures 6.4VRMS, pretty loud! So I will try it at 4v, which I guess is similar to a CD player output unbalanced at 2V.

 

[Blumlein 88]

Doing the test as a loopback on your device should be fine with that as the timing reference. Using sound for the timing reference is when you use a loudspeaker which means there will be a delay. So it puts out a high frequency tone just before the sweep to properly set delay. In a loopback on electronics you won't see much if any difference. If you were doing a speaker you would.

With the electronic loopback you could get away with using -1 dbFS. It may improve your distortion numbers a little or not. I didn't see the THD number perhaps it was cut off your picture, but otherwise you are getting distortion that would be in the mid -80 db range. Possibly/probably about right for this device. Each of the graphs for each harmonic are relative to the SPL that REW thinks you are giving it.

As for the level the software is oriented toward speakers so it ends up giving some SPL level. It really isn't important otherwise though you could calibrate it to show something like 100 db or some other even amount. Your response looks flat except slight droops at each end which is believable.

 

[Blumlein 88]

Thanks RMAA looks better in that the charts are a bit clearer, its PC only though I think, I've got a Windows Laptop I could try it on though.

I managed to get the chart I was after, still not sure about the levels though.

I will read that guide. Here is the latest result.

I notice you have it set to 16 bits with dither. The Lexicon can do 24 bits, so you would want to change that to 24 bits.

 

[Blumlein 88]

OK I can see I was using the wrong window, this is much better. I've turned averaging on as well as the chart was moving up and down a lot.

Yeah, you can use either chart. The 1st one is a sweep across the whole band. With the other window you can get results more like you want at a single frequency or even do two tone IMD tests as well.

Looks like you are already getting the hang of using it.

 

[dunkuk]

Thanks Blumein, yes its not that difficult after all. You are right about 24bit I hadnt thought about that, I will try it now.

4v overloaded the inputs a lot! Was no good at all.

Best THD seems to be about 0.84v (across tip and ring, 0.42v unbalanced) which is about the setting on the volume knob I have the speakers at.

 

[dunkuk]

Ok the last 2 results for tonight, I did try moving the test signal level up to -3db but the interface distorted a lot, it couldnt handle it.

So here are plots made with -10db and 24bit signal.

I need to compare to other sound interfaces but I think this is telling me this device is not that good, which is good information as I'm looking to upgrade. It cost about £30 (35euro / Dollars) about 5 years ago so it makes sense why it is so cheap.

I'm looking at a Topping D30 or newer interface like the Focurite Scarlett, I think they might be a lot better than this. This seems to distort really easily with large input signals. However I suppose the ADC in could be bad on this and the DAC ok, but I dont know how I can check that.

 

[RayDunzl]

This seems to distort really easily with large input signals.

You may find it common for consumer ADC to add harmonic distortions at some point above say -6dBfs (varies among units)

As for "large signal" - 24bit = 144dB

So you back off to 138dB or 132dB and get a harmonic-free capture.

Ignoring the noise floor, of course.

 

[JohnPM]

Looks like you are using a rectangular RTA window (i.e. no window) but you have not set up the generator to be locked to the FFT length. As a result your spectrum plot is dominated by windowing artefacts (hence the movement you mentioned). Either use a different window (Blackman-Harris 7, for example) or select the box to "Lock frequency to FFT" on the generator.