0

DAC+ADC pro annoying crack on output when starting to record from input

Nicolas Hifiberry

December 29, 2021 13:59

None

Follow

Hi,

I am using a bunch of raspis with various Hifiberry Boards on them. My architecture is to use plain Raspbian and some Ansible Roles to configure the audio setup. We just use alsa and everything is set to use a sample rate 44100 (CD format).

My main raspi is connected twice to my main amp:
* The DAC output goes to the "Tuner" Cinch Input of the amp, this is selected as output to the loudspeakers.
* The ADC Input is connected to the "Rec Out" of my amp. The amp has a separate REC selector (this is important to make my use case happen).

In order to listen to my vinyl records in all rooms, the audio path on my main raspi is as follows:

* The record player plays a record.
* The rec out selector on the amp is on phono
* The vinyl sound exits the amp and enters the ADC
* I have a permanent `acable` ( a `arec | aplay` pipe) process copying the dsnoop input from the ADC to the dmix input of a virtual alsa audio card (snd_aloop).
* The snapserver takes the audio from the snd_aloop output dsnoop device and feeds it (FLAC/CD-format) to all snapclients on the network, including the local one.
* The local snapclient plays the audio to the ADC dmix output
* The audio signal goes to the "tuner" input cinch connectors
* and is played on the local loudspeakers (with a local delay of about 1 sec, but in sync with all other rooms).

When I stop the `arec` on the ADC-dsnoop device, nothing happens (except that I have no analog sound any more).

But when I start the `arec`, a very loud crack is heard on my loudspeakers. The crack has neither delay nor can it be heard in the other rooms. So I am sure that the crack is NOT transported via snapcast, but there must be hardware (or driver?) issue.

I own a DAC+ADC (without pro) too, and I believe it has exactly the same issue.

• Facebook
• Twitter
• LinkedIn

5 comments

• 0
  
  HiFiBerry team December 30, 2021 05:53

  The problem comes from the fact that you start recording while playback is active. As this requires an new initialisation of the clock circuit, you see this behavior. As the record and playback part aren't independent, this is just normal.

• 0
  
  Nicolas Hifiberry December 30, 2021 23:04

  Thanks for that quick answer. So you are telling me that I would have that effect with any soundcard and that it is not possible to start rec without reinit of the clock (which in turn affects any running play)?

  If yes, if I would like to avoid that crack I have only those options:

  * start rec at boot time and never stop it again.
  * when I really want to (re-) start rec later: stop play, then start rec, then start play again?

   

• 0
  
  HiFiBerry team December 31, 2021 05:59

  > So you are telling me that I would have that effect with any soundcard and that it is not possible to start rec without reinit of the clock (which in turn affects any running play)?
  
  I can't say anything about "ANY", but you will have this with any of our DAC+ ADC boards

  > If yes, if I would like to avoid that crack I have only those options:
  
  Yes

• 0
  
  Nicolas Hifiberry December 31, 2021 08:35

  Thanks for your answer, that makes it clear. Now (I hope) my last question to this issue:
  
  Would it theoretically be possible to build a soundcard without that limitation (I guess that would mean independent "clock circuits" for play/rec) or to stack two soundcards with individual clocks? Or would this raise other issues? I'd be willing to pay a bit more for that feature - but of course that won't be enough motivation for you if I am just one of those 0.1% users that ask for a feature that nobody else is using ;-)

• 1
  
  HiFiBerry team December 31, 2021 08:57

  Theoretically: Maybe, practically it's complicated. As the Pi doesn't feature independent interfaces for input and output, it not trivial. That's definitely not something we have on our roadmap.