Absolute level cal with REW generator and SLM

[craigctodd]

I'm having a tough time level calibrating my new Anthem AVR MRX 1140. I've run ARC Genesis, and I've used the internal noise generator and I've measured level with the REW sound level meter (SLM) with (batch) calibrated UMIK-1 mic, and with Radio Shack SLM. I think the Anthem is setting things 10 dB hot vs commercial cinema where -20 dBFS pink noise comes out at 85 dBc at reference fader setting of '7'. It sure sounds HOT. Using REW generator and SLM (double checking the Radio Shack SLM) for calibration gives levels that sound right to my ears.

I'm very familiar with how modest errors can creep in when trying to calibrate so that -20 dBFS pink noise comes out at 85 dBc. Some analog meters are averaging vs true RMS. I have a Fluke DVM and an HP 400EL and both are averaging. The actual bandwidth of the pink noise can exceed what a meter would measure. Some SLMs are RMS and some (Radio Shack) are averaging. The bandwidth of the noise sent into the system can exceed the bandwidth reproduced by the speakers. (E.g. the pink noise bandwidth could be 1Hz to 100 kHz and set to -20 dBFS but what would come out of the speakers and into the room would be missing several octaves and so a cal to 85 dBc would be wrong. I expect REW gets it right but would like confirmation.

I'd like to better understand how REW gets its output noise level to match the level setting (e.g. -20 dBFS) and whether anything could get in the way of that level getting into the AVR.
I'm using ASIO4ALL to send digital audio from REW into the AVR. I can't measure the actual signal level on HDMI (no test point to put a meter on!) but I can use the AVR as a DAC. With AVR volume and trims set to 0 dB, and a 0 dBFS sine wave sent to HDMI 01 (left channel) the analog L ch output level was 9.2V p-p. In order to determine if that actually represented full scale digital (REW won't let me request >0 dBFS to look for clipping level) I had REW put out high level noise that did clip; the AVR L ch output clipped at 9.2V so I'm now trusting that REW's 0 dB FS for sine waves is actually that and the computer is actually putting that level onto the HDMI output and its making it all the way to the AVR DAC.
Next I'm wondering how the generated noise level (dBFS) is calculated . It should be a true RMS measurement, easy to do in a computer. I like that the pink noise bandwidth can be limited to what the speakers can actually deliver to the measurement mic, e.g. 32Hz to 10kHz, or the "speaker cal" 500-2000 Hz band can be selected.

Next question is how does the REW SLM measure? True RMS?

Can someone confirm REW is using true RMS for generator level and SLM measurement?

 

[John Mulcahy]

Yes, REW uses RMS for output and input levels. OS volume controls along the input and output path can affect levels outside REW, however. ASIO drivers normally bypass such controls, WASAPI exclusive usually does as well, though not always.

 

[craigctodd]

Yes, REW uses RMS for output and input levels. OS volume controls along the input and output path can affect levels outside REW, however. ASIO drivers normally bypass such controls, WASAPI exclusive usually does as well, though not always.

A bit more info and another question for John.

Anthem let me know that ARC reference level is at Volume setting of -10 dB. So if that's 85 dBc SPL for -20 dBFS pink noise it explains what I'm seeing; I've been targeting 85 dBc at Volume setting of 0 dB. ARC is apparently calibrating so that Volume=0 gives 95 dBc, or 10 dB hotter than calibration in a cinema. Seems an odd choice.

If I set level with "speaker cal" pink noise, 500-2000 Hz, and then check with full bandwidth pink noise I see a drop in level of a couple of dB. So some broadband energy is going missing. The RTA shows response pretty flat 15Hz to 15 kHz.
Question: What is the bandwidth of the full band pink noise? I presume at the top end it stops about 20K, can't go above 24k unless there's no anti-aliasing filter. The lower the pink noise goes in freq the more octaves there are below my 15Hz cutoff and those will contribute to the -20 dBFS true RMS noise level (source signal measurement) but won't show up at the mic input. I presume the periodic pink noise is band limited on the low end to a couple of Hz due to the period of the cycle.

 

[John Mulcahy]

The help covers those questions in detail.