[LAU] bit-transparent playback/recoding over AES using an RME 9632 card

[Markus Seeber]

On 11/16/2016 04:17 PM, Tino Mettler wrote:
> Hi,
>
> I try to get bit-transparent playback/recording of a stereo stream with
> a RME 9632 card.
>
> My test setup the card clocked at 48 kHz and the AES input and output
> looped.

Which means you connected the digital output to the digital input I guess?

> When I tried to use pure alsa, I recorded 10 channels to get channels 9
> and 10  which is the AES input (I wasn't able to find a way to record
> only channels 9 and 10). However, I'm not able to play back anything
> during the recording. I get the error message "set_params:1361: Unable
> to install hw params".

Do you count "channel 9 and 10" one or zero indexed? :)

Here is the channel mapping from the source for reference:

static char channel_map_H9632_ss[HDSP_MAX_CHANNELS] = {
    /* ADAT channels */
    0, 1, 2, 3, 4, 5, 6, 7,
    /* SPDIF */
    8, 9,
    /* Analog */
    10, 11,
    /* AO4S-192 and AI4S-192 extension boards */
    12, 13, 14, 15,
    /* others don't exist */
    -1, -1, -1, -1, -1, -1, -1, -1,
    -1, -1
};


>
> I also tried jack and started jackd this way:
>
> $ jackd -R -d alsa -d hw:DSP -z none -M -H
>
> I was able to play back a square sample and record channels 9 and 10
> using jack_rec. However, the result looks like it went though bad AD/DA
> conversion: jumpy edges and the linear parts unclean, too.

Depending on what you "mean" with "the result" this looks perfectly fine
to me. To be expected of a sampled square wave signal.

How did you generate the signal?
How did you measure/visualize the output?

Keep in mind, that with a limited sample rate, you will always have an
approximated square wave, looking exactly as you described: rippled
linear parts and jumpy edges. This has nothing to do with a loss of
quality in the digital domain.

If you want to check bit transparency, you need to compare the samples
of your digital input and output.

>
>
> I'll try to find out if the playback or recording part is the culprit.
>
> However, I'd like to get some comments if bit-transparent
> playback/recording can work at all with my setup. If it does, I'd
> appreciate to get some hints how to achieve this.

I don't see any reason why it would not work if you connect the digital
output to the digital input.

>
>
> Regards,
> Tino
>

Greetings
Markus