Getting more that 1Vrms out of my sound card

Hello all! New user, first thread, and I apologize if this has been answered elsewhere but I can't find it!

Update: Major edit based on additional research and testing.

I am using REW with a Focusrite Scarlett Solo Gen 3, for distortion analysis of high-impedance circuits. The Scarlett is spec'd to output up to +15.5 dBu, which would be 4.6Vrms into a balanced load, or 2.3Vrms single-ended. I used Audacity to generate a 0 dBFS sine wave, which got the Scarlett to output 2.4Vrms single-ended, with the output volume at max. This actually slightly exceeds Focusrite's spec (I would expect +9.5 dBu single-ended, which is 2.3Vrms).

However, if I use the Generator in REW (also set to 0dBFS), I only get 1.2Vrms single-ended - 6dB down. It's like the MSB isn't turning on somehow.

I have the interface set to 48kHz, and according to the main screen in REW (and AudioMidi Setup - I'm on a Mac) the bit depth is 24. Same settings in Audacity.

As an additional test, I used both REW and Audacity to generate .wav files of sine tones at 0dBFS, and compared the signal maxes and mins by reading the arrays into a Jupyter notebook in Python, for both 24 bit PCM and 32 bit float encoding. The peak maximums are the same. Audacity's peak minimums are a bit lower but the difference is on the order of 10^-7.

Any help troubleshooting this would be greatly appreciated! I could really use the extra voltage! Thank you

windchimes said:

if I use the Generator in REW (also set to 0dBFS), I only get 1.2Vrms single-ended

Click to expand...

With what driver selection? If you are using ASIO you should get the same result, with Java drivers the Windows volume control may be in the path. For Java choose device names starting with EXCL to use WASAPI Exclusive drivers.

Hi John, thanks for REW, and thanks for your reply!

I am on a Mac (14.6.1). I believe everything goes through CoreAudio? I don't seem to have a lot of controls/options.

Here are the relevant preference screens. I didn't include the System Settings/Sound one - if I select the Scarlett for either input or output, the system volume level fader is grayed out.

I did another experiment where I calibrated the Scope in REW, then set the Scarlett as input to REW and output of Audacity, and verified my previous result with the REW Scope.

Please let me know if there is any other troubleshooting to try. Thank you!

REW doesn't try to control the input or output volume on Mac. It generates data at the requested dBFS level and sends that data to the CoreAudio driver via JavaSound. If you save to file from the generator you get exactly the same data it sends to the audio driver, which from your WAV test is the expected 0 dBFS, so any subsequent change in level must be happening later in the path. If there is a mixer function of any sort active in the replay path that may be a factor.

Thanks John. I don't think there is any additional mixer in the path, so this remains mysterious. I have a few more tests I plan to try. I have a second USB audio interface, so I can repeat my tests using that interface. I also have a newer Mac (with a more recent OS) that I could try. A friend suggested I could use a virtual audio cable (e.g. blackhole) to pipe audio directly from one program to another, and verify the stream that way. I'll report back after I try some more things.

windchimes said:

Thanks John. I don't think there is any additional mixer in the path, so this remains mysterious. I have a few more tests I plan to try. I have a second USB audio interface, so I can repeat my tests using that interface. I also have a newer Mac (with a more recent OS) that I could try. A friend suggested I could use a virtual audio cable (e.g. blackhole) to pipe audio directly from one program to another, and verify the stream that way. I'll report back after I try some more things.

Click to expand...

have you tried calibrating the generator output?

@windchimes, You say "have the interface set to 48kHz" and yet your Mac Audio Setting appear as 44100 Hz... And it appears that you are using the Internal Mic? I don't know if this would be the cause of your issue...

oliver said:

have you tried calibrating the generator output?

Click to expand...

Hi Oliver, this is a good point, and if I could start this thread over I would change the title. By default, REW assumes that 0 dBFS = 0dBV = 1Vrms. This may have been "generally" true at some point in the past, but is not true anymore (especially with external sound cards aka USB audio interfaces, many of which work at pro line levels.) There are actually two calibrations, one for the Generator, and one for the Scope. The generator is easy. Turn on the generator at 59Hz, 0dBFS, and measure the output with an RMS-reading DVM (today I'm getting 1.235Vrms for my system). Then enter the measured value in the Generator Calibration box. Again - easy!

Next, I have to calibrate the Scope. Unfortunately, this is not as easy. The problem is that the the input and output disagree on the meaning of 0dBFS by the factor of the channel gain. If you know the gain is at unity (which you can do manually via Preferences/Calibrate Soundcard, connecting input and output, and adjusting the input gain) then it's easy - enter the same rms voltage you read from the generator output into the channel calibration box. However, my interface has specs for a maximum output (with the output volume pot all the way up), and a maximum input at minimum gain. In my application, I need to max out the output level. Therefore, I can't operate the input at unity gain, because that will clip the input. I set the Generator to 0dBFS, set my gain where it needs to be, and temporarily enter 1Vrms in the scope cal. Then I write down the Vp value I'm actually seeing on the scope (in my case, I get 166mVp.) I set my generator to read volts, and it shows me that 0dBFS is 1.75Vp (in my system). Then I can calculate a scale factor for the scope calibration: Vp (generator)/vp (measured). In my case, I get 10.54V. I enter that as the scope calibration, and then I can verify that the scope is reporting a signal that I know is 1.75Vp as 1.75Vp.

Alternatively, I could keep my sound card at unity gain, and use an attenuator with a known attenuation factor. This will definitely be important for testing amplifiers, but right now I'm just working on high impedance, low voltage circuits, so I can postpone the attenuator.

ddude003 said:

@windchimes, You say "have the interface set to 48kHz" and yet your Mac Audio Setting appear as 44100 Hz... And it appears that you are using the Internal Mic? I don't know if this would be the cause of your issue...

Click to expand...

Sorry for the confusion. When I made the initial post, I had the interface set to 48 kHz. I subsequently switched to 44.1 kHz, which didn't change the "missing 6 dB" problem. The screen shots are from testing at 44.1kHz. The previous screen shots also show a special test where I had Audacity and REW running at the same time. I assigned the Scarlett output to Audacity and the Scarlett input to REW, (and Macbook Mic to Audacity in, Macbook speakers to REW out) and verified that I can get a bigger output from Audacity than from REW. Here are some additional screen shots of that test. First, Audacity driving the (calibrated) REW scope to 7Vpp, then REW driving the (calibrated) REW scope to 3.5Vpp. I didn't touch input gain or the cable form the output to input. Again, the cause is still unknown.