Mod RME Quadmic I?

[Khron]

What sort of numbers do you get at 96kHz, or 48kHz? Cause that's a LOT of ultrasonic noise curving up from 50kHz; 30-40dB's nothing to sneeze at...

(Presumably this is all at 192kHz sampling rate..?)

 

[opnordahl]

Yes, all is done at 192kHz. I will test all channels when the mod is finished.

SMD soldering in tight areas with lots of plastics is a pain.

 

[opnordahl]

I got a much better SNR at 96kHz sampling rate:


I must have done something wrong on channel 2. I cannot see anything obvious.

All measurments with HiGain Swith, 20DB amplification, -5dB output:
Channel 1:

Channel 2:

Something strange is going on channel 2. This channel is unstable. Maybe a bad connector or a bad soldering. (Edit: Seems to be a bad output connector) Because suddenly it measures like this:

Channel 3:

Channel 4:

 

[opnordahl]

It was both a bad output jack connector and a bad soldering joint on channel 2.
After fix 20dB gain -4.75dBFS output. 96kHz sample rate:


The last test I will do is to see if there is any reason for upgrading to a linear power supply.

Unmodded channel 4:

Modded channel 4 (after channel 2 repair):

 

[opnordahl]

I tested the preamp with a low noise power supply. The results where exactly the same. The amp uses a LM2585S -12V voltage regulator. The circuit provide about +/-14V for the rails and a 48V phanthom power. It is possible to upgrade this part of the circuit to a linear power to get rid of some ripples but that is for next time.

Here is the finished modification with some hot glue to stabilize the capacitors. I used Wima Mkp 0.1uF bupass capacitors in the signal path. In my experience the treble sounds a bit more pleasing with these. I have also changed all 10 opamps to OPA1612. When soldering these some extra care was needed not to melt any of the plastic parts.

 

[tinn]

Redarding P48 Noise:

I would do a test:
Connect XLR input pin 3 to GND with 100uF
Connect Input Pin 2 to Ground with serial 200 Ohm and 100uF to Ground
Listen or measure how the P48 sounds.
In normal use the P48 noise will be much lower, if your source is symmetrical.
You might do a symmetrical test as well. P48 noise might not be detectable.

Best, Tinn

 

[opnordahl]

I am no REW expert but SNR measurement but the SNR number looks strange. I would expect it to be about 132dB SNR or is the calculation only done on the 2nd fundamental distortion?

 

[Khron]

I am no REW expert but SNR measurement but the SNR number looks strange. I would expect it to be about 132dB SNR or is the calculation only done on the 2nd fundamental distortion?
View attachment 146149

https://www.roomeqwizard.com/help/help_en-GB/html/spectrum.html

A signal-to-noise ratio figure is also displayed if a rectangular window is being used and the FFT is two or more times the signal length, in which case the tones will only occupy bins which are multiples of (FFT length/signal length) and anything in the other bins will be noise. The noise figure is obtained by rms summing the noise bins and scaling the result to account for the proportion of excluded bins. SNR is the ratio of the total rms input level to that unweighted noise level.

 

[Voyager10]

Also from that page, in the 'Spectrum/RTA controls' section: For example, with a 64k FFT length and 48 kHz sample rate the bins are 0.732 Hz wide. The plot shows the energy in each of those bins.

So, sadly that '-150dB' number is a bit of an illusion, the actual noise level is a weighted sum of thousands of those bins. (If you change the bin width/FFT length or use a different display mode, e.g. 1/3rd-octave RTA, the number will change).

On the plus side, that plot is showing the noise of the signal source, as well as the preamp itself, and the signal source may well account for most of that -97dB noise. To get the preamp noise on its own, calibrate REW with the signal source connected, then measure with the input pins 2 and 3 shorted, or connected via some representative source resistor e.g. 150 ohms.