virtins.com

Hi
RME ADI-2 Pro comes with a set of files to check "bit perfect"ness of the signal path.

Looks like with Multi Instrument, there is always a dithering applied.
This is all good, since in 99% of cases, that's what we need.
And I know we can create "bit perfect" files, that MI Pro will run without dithering.
But I tried to convert and import the RME files into some bit perfect file in MI Pro format and didn't succeed.
MI Pro always refuses the format.

What exactly should be the format for bit perfect?

RME bit-test wav files are ordinary WAV files containing a sequence of predefined sample values. The bit-perfectness of the transmission path can be checked at the receiver end as it knows what the original sequence should be.

The BitPerfect library file used by the signal generator of MI is a specially designed CSV text file. It contains the sample values to be output by the signal generator directly without being resampled, rescaled, and dithered, or being modified in any way.

For example, to generate a 16-bit positive full-scale DC signal using a BitPerfect library requires only one line:

0,32767,32767

where the first value is a sequential number, followed by two integer values (always positive) for left and right channels.

Similarly, to generate a 16-bit alternating positive/negative full-scale DC signal using a BitPerfect library requires only two lines:

0,32767,32767
1,32769,32769

Note that: a 16-bit WAV file uses signed value based on two’s complement.

The signal generator of MI does not add dithering implicitly. If required, dithering can be added using the MultiTones function of the signal generator. That is, add a certain amount of white noise.

OK, Thanks

So what I was missing is how the "Always positive" was actually working.

I thought 0 was -FS and 65535 was +FS.
Now I understand 32768 is -FS and 32767 is +FS.
Correct ?

Which, for 24 bits, would translate in -FS=8388608 and +FS=8388607
Still correct ?

Yes. Following the WAV file format:

For 8 bits, unsigned value is used, -FS = 0, +FS = 255
For 16 bits, signed value is used, +FS = 32767, -FS = 32768
For 24 bits, signed value is used, +FS = 8388607, -FS = 8388608

The software manual will be updated to make this clearer.

Thanks !
Indeed, and adding some examples may be very useful