ANSWERED: Running stereo audio out thru hardware one CH at a time without phase probs

[nms]

As someone who prefers to buy two different expensive analog outboard pieces rather than doubling my investment to get two channels of the same unit, this is an important concern.
I just route the stereo content out from my DAW printing one channel at a time and don't do anything silly in this unlinked stereo environment to cause undesirable effects on my stereo image. The bonus here as well is you don't have to worry about matching settings or the sound of two units that aren't calibrated the same.

Gregory had me worried for a moment when we were talking about running audio through Distressors one channel at a time, since I don't take it lightly when he weighs in on something:

Originally Posted by u b k

The thing is that the time alignment of two phase-coherent tracks that pass thru converters in different passes will constantly drift in and out over the course of the song, it's part of the phenomenon known as sub-sample latency and there is absolutely no way to correct for it.

It's one of the dirty little secrets of the digital recording industry.

Gregory Scott - ubk

Well, I'm very happy to report that I just ran a mono piece of audio through my Motu 828mk2 twice, inverted the phase of one take, and nulled it against the other to be left with the sound of pure silence at -96db RMS! Now I can't say this will be the case for every interface.. but it's as easy as that if you want to check if yours is as safe and not going to cause any phase misalignment. The 828mk2 is an underdog allstar performer where line level conversion is concerned. Still hard to believe I went from a Steinberg MR816 to an 828mk3 to a Mytek 96 to this thing! Can't argue with the results though.. no matter what I throw at it it holds it's ground.

[u b k]

Well, thanks for respecting the things I say, and thanks for doubting me enough to test this on your own. I love being wrong, especially when it means things can be done that I thought weren't possible.

That said, I'd be more convinced if you took a stereo source with good phase coherence, e.g., a well recorded drum submix, and printed each channel one at a time, then posted the resulting split stereo file. Any chance of that happening? I can provide the stereo source if you need it.

Thanks!


Gregory Scott - ubk

[nms]

For you, I shall do just that. PM me the link.

And yes.. being right or wrong is never as valuable as knowing for sure! It's especially nice when it reopens possibilities. Sometimes myths get passed on without anyone testing them properly, other times an experience unique to a particular unit throws you. In the end, proper testing beats theory and hearsay any day!

[ltjohnrambo]

Out of curiosity, were you running a session with a single piece of test audio or was it a full mix with plugins and such? I'm thinking that plugins and latency are giving me problems when I try to do the same thing.

[nms]

There will be latency of course, but that latency will be exactly the same for both takes, so just run each channel through then open both files in a wave editor and put the two mono files into one stereo file before you do anything else.

I use the track delay on the sending channel in Ableton to adjust in samples a negative amount so that my audio prints without the lag though. To figure out the delay I run a click through first with an easy to spot transient and then measure in samples how much it lags.

[Muser]

I was going to mention this to UBK before but didn't get round to it.

It's concerning PSP84 delay plug-in and subsample delays.

If you have an instance of PSP84 try this.

set mode to (Time)

Ranges
0.1 = 100ms
1 = 1000ms
5 = 5000ms

Note:
not until you float your mouse over the up and down arrows do you see the DLY readout indicated.

method 1. in Logic.
switch range of the channels to 5 and hit the down arrow till one or both, reads 0.000
switch range to 0.1
now when you hit the up arrow you'll see the highest resolution of that increment takes it to 0.100

However...
switch the plugin from (Editor) mode to (Controls) mode. then you will see the basic representation controls as sliders with numericals.

hold the mouse over the numerical DlyL and push up one.
That readout will be 0.0001
switch the plugin back to (Editor) mode
and hover your mouse over the arrow on the far left again.
you should see that that increment has actually made the front face LED value now read 0.010
which I believe is 0.010 of a millisecond.

If you hit the arrow now, you'll see that the GUI resolution designed by PSP takes the values up in
0.100 > 0.200 etc increments. This I believe is 0.1 of a millisecond steps.

Logics control mode gets down to a tenth more resolution.

Method 2 - Logic and Automap with a Novation Remote SL data entry pot assignment.
what I discovered was that if I set Novation Automap up with my Remote zero SL to control that DelayTime
parameter with a points resolution of 4500001 (which means that Automap creates 4500001 discrete control points on the PSP 84 DelayTime parameter) I can get delays which are down to and even beyond what the display is capable of displaying. so that's a possible delay resolution of down to 0.0001 of a millisecond. which would have to be subsample delay accuracy.

I'm guessing PSP 84 has this resolution for very smooth modulated delay resolution.
but they maybe just thought this level of delay was not practical to control for the user.
so they decided to make the simplest and most intelligible choice for user resolution.
or maybe it something to do with a prime number approach to delay times.
so you have a more versatile musical delay. aka lexicon prime time

but It looks like the resolution is there in any case.

I did test this in the past and was pretty satisfied that this resolution is obtainable in PSP84.
above is the outline of 2 ways I found I seemed able to get at it.

if you combine this with an Airwindows AU sample delay you can use that plug-in to delay
a signal earlier in time and the PSP to offset that subsample accurately a little later.
Airwindows also do a phase nudger.

anyhow, be interested if someone can confirm it.

[2pulse]

As long as the DA and AD are synced to the same clock, you shouldn't get any drift whatsoever from separate round-trips. However, if you're capturing the output with a separate, internally-clocked AD converter, you may encounter some bad times.