[LAU] MotU Ultralite AVB observations.

Mario Lang mlang at delysid.org
Wed Mar 11 10:41:25 CET 2020


Hi.

I have a MOTU 8A in UAC mode running under Linux/JACK.
Firmware is from 2017, I don't remember the exact version off-hand.
Time from switching it on to Linux seeing the USB device is
roughly as you describe.
I haven't heard the bit-crush bug yet (lucky me).
The card seems to work pretty fine, except for one problem:

Sometimes, when I freshly power on the card and start up JACK (via a
udev rule[1]), JACK doesn't complain, but there is no audible output.
Restarting JACK (sometimes two or thre times) will fix the issue, and
once JACK has managed to init with audio being audible, the card
will work for hours without any problem until I power down.

What I discovered recently is, that the default analog output level
is pretty high.  I use the following script to trim it down to something
that (hopefully) resembles line level:

#!/bin/sh -e
motu=http://8A.local/datastore
analog_out=ext/obank/1
# trimRange -24:0, we're guessing that -16 is +4dB
for ch in $(seq 0 7)
do curl --data 'json={"value":-16}' $motu/$analog_out/ch/$ch/trim
done

[1] /etc/udev/rules.d/10-MOTU-D8A.rules
ACTION=="add", SUBSYSTEMS=="sound", ATTRS{id}=="D8A", TAG+="systemd", ENV{SYSTEMD_USER_WANTS}="motu at D8A.service"

~/.config/systemd/user/motu at .service:
[Unit]
Description=%I JACK Server

[Service]
ExecStart=/usr/bin/jackd -d alsa -d hw:%I -r 48000 -p 128 -n 3

John Murphy <rosegardener at freeode.co.uk> writes:

> Getting my Ultralite AVB working on Linux Mint 19.3 has been quite
> a struggle and I've learnt a few things along the way, which may
> save others some wondering.
>
> Ignore any thoughts of USB power management causing any issue.
> The device never uses any USB power, as confirmed with an inline
> USB V/A meter.
>
> Firmware more recent than 1.3.2+520 suffers from the 'bitcrusher'
> effect, where audio can be heard for about 30 seconds, until it
> fades to junk while appearing on higher number channels, until
> it does the same again. That firmware is pre Touch Console™.
>
> Firmware and release notes in the Ultralite AVB section here:
> https://motu.com/techsupport/technotes/firmwarechangelog
>
> A great deal of good information over at the Linux Musicians forum:
> https://linuxmusicians.com/viewtopic.php?f=6&t=18046
> (Thanks to Eric Schoster for pointing me to that one.)
>
> One gem is that it can be set to class compliant mode with:
> curl --data 'json={"value":"UAC"}' IP address/datastore/host/mode
> curl --data 'json={"value":"24"}' IP address/datastore/host/maxUSBToHost
>
> (Replace "IP address" with its address on your network)
>
> Change to its proprietary mode with:
> curl --data 'json={"value":"USB2"}' IP address/datastore/host/mode
> curl --data 'json={"value":"64"}' IP address/datastore/host/maxUSBToHost
>
>
> On my current hardware (i7-8700K) the Ultralite takes 27 seconds
> from first light in the display to showing detail in the display
> and there's a further 20 second delay before any activity shows
> up in a terminal window running 'tail -f /var/log/syslog'.
>
> There are very many messages like:
>
> 'clock source 1 is not valid, cannot use'
>
> Yesterday I was seeing 240 of those in blocks of 40, with each
> block followed by:
> '[pulseaudio] module-alsa-card.c: Failed to find a working profile.'
> and etc.
>
> Today, after trying the latest firmware, seeing the bitcrusher
> effect and reverting to 1.3.2+520 again, I only see two complete
> blocks and the reporting stops about half way through the third
> block. Qjackctl will then start the server, on either hw:AVB or
> hw:AVB,0. It always seems to need changing to the one of those
> it isn't set to. :)
>
> I can only think that my dual booting and using it on Windows 10
> caused the trouble I was seeing when using it on Linux afterwards.
>
> I still can't use anything which isn't 'jack aware', but I can
> live with that, for now. VLC offers 13 'Ultralite AVB, USB Audio'
> devices to chose from, including "Direct sample snooping device".
>
> Linux, generally, doesn't seem to maintain an optical output if
> a player app stops playing...

-- 
CYa,
  ⡍⠁⠗⠊⠕


More information about the Linux-audio-user mailing list