Subject: Re: [linux-audio-dev] Homebrew control surfaces was (ardour,LDSPA... A marrage)
From: Benno Senoner (sbenno_AT_gardena.net)
Date: Tue Nov 21 2000 - 21:19:00 EET
On Tue, 21 Nov 2000, you wrote:
> On Fri, 17 Nov 2000 20:11:35 +0100, Benno Senoner wrote:
>
> >Of course USB is better than serial/MIDI because of the up to 12Mbit , vs
> >31/100kbit of the MIDI/serial port.
> >
> >Internally you should scan the controllers as fast as you can,
> >and if the controller's new_value != old_value then send out new_value.
>
> Well thats just it... I need to know how fast "as fast as you can" needs to be as that decides
> the minimum ability of the microcontroller needed to scan the controls and thus the system
> complexity.
Hmm.. no numbers here, but I think if you can ensure a latency
(trigger-to-midi-out) of a few msec then most people will be happy.
(preferably in the low single digit range, aka 1msec :-) )
That means if you have 20 controllers and must scan them one by one , plus if
you want to achieve 2msec latency max, then you need to scan
each controller once every 100usec.
I could be wrong with the numbers, but using MIDI fader boxes, you won't achieve
better than 1msec latencies. I'm not sure what the pro-audio guidelines suggest
in that area. (anyone knows USB's minimum packet latencies (just for
curiousity))
>
> What and how many controls would most people consider the minimum necessary on a control
> surface to be usefull assuming it was talking to something like Ardour.
Just clone Steinberg's Houston and many will be happy with you controller :-)
Benno.
This archive was generated by hypermail 2b28 : Tue Nov 21 2000 - 20:06:53 EET