Timing difference between converters?

[Alistair]

Hey all-

I've changed my setup recently to include 2 different AD/DA units- nothing fantastic, just an old RME 8 channel ADAT, and a Motu 828.
I'm having a little bit of an issue with using them both, as they seem to process audio at slightly different speeds. For example, if I've got a track running through the RME, then bus it out to a channel on the Motu for some parallel processing, I get huge phase issues.

I've done a little test to see the differences, doing a round trip (sending a track out the D/A, and back in the A/D), I'm getting a delay of 45 samples on the Motu and 47 on the RME (pretty high!). However, I'm not sure how to measure what the difference in just the D/A speed is? After the D/A stage, it all hits a summing mixer so that's the important difference to get. I'm hoping once I know the magic number I can just delay or anticipate the tracks by that number in my DAW to fix the issue?

Any ideas or experience with this problem??

[audioio]

For example, if I've got a track running through the RME, then bus it out to a channel on the Motu for some parallel processing, I get huge phase issues.

When you say you have a track running through the RME, does this mean the sound originates in the analog domain?

Or, you are playing out of the RME and the MOTU at the same time for parallel processing?

[Alistair]

Hi David,

This is just mixing and printing of stems from stuff recorded previously (and on different systems).
I actually did a few tests and some maths today, and I think I've got a workaround. Using an oscillator I was able to find out exactly how much latency each converter created in a D/A out and back into A/D loop. Pretty interesting, and I've now got a bunch of numbers for sample delay settings to remedy this stuff. The only annoying this is I still can't use a parallel compression track on different converters- I've got no way to negative delay/anticipate bus tracks, so I've gotta print first and then move around. Not perfect.
If anyone needs help doing this sorta thing, PM me and I'll pass on the method I used. It's good to know if you're using different converters.

[justin]

Hey Al,

probably a silly question, but are you running them from the same clock?

I remember having a 002 and another unit not playing so nicely together til I sorted that..

[Alistair]

Hi Justin,

Yep, all clocked to a BLA Micro Clock so it wasn't that- it's literally just that the processing of digital to analog takes longer in one set of converters than the other. Never would have guessed, but I suppose it makes sense when you think about it!

[The Tasmanian]

Ive had this problem before - you just have to plan your tracks and fit the drums on 1 set of convertors - a few times I relegated the far drum room mics to a separate box of converter's, so I could get 8 channels of the close mics on one 8ch box of A to D's.
No way around it bar getting the same brand of A to D's

[Alistair]

Well, I'm not so sure that there's "no way". Through doing some tests sending a test tone round a full round of D/A and then A/D, and then flipping the phase and trying to null it, I found that one set of converters take 58ms, the other 91ms, and if I go out one and in the other it's 73ms.

So the difference between converters is 33ms. To be pretty unscientific, I halved all the amounts to get one conversion (either A/D or D/A- this is assuming they take the same amount of time to do either)- so 29ms for one converter, and 45 (roughly) for the other. Add these together, and that's 74ms, which is close enough to 73- the delay induced by using one of each to do a round trip.

So, using that, I just have to delay the faster set of converters by 16ms so they both take 45ms to spit audio out when I'm doing a mixdown with outboard/analog summing etc. If I'm printing stems, I use a sample delay of -58, -73 or -91 depending on converter chain. It's a bit of a pain, but once I got the numbers down it's easy, and it does fix some phase problems.

I still can't do this with aux tracks send to outboard though- Logic needs a plugin that works on busses/aux tracks that lets you sample delay in the negative to achieve this. The only workaround would be to delay EVERY other track if you use an aux track which is a bit ridiculous most of the time.

Does this make sense?

[The Tasmanian]

Sorry I meant no physical way of changing the hardware.
Still it must be a PITA to be mucking around with delays - but whatever it takes to do the work.

[Alistair]

It's not great, but it's more of a pain hearing the phase between drum tracks!

[Kurt]

The problem has to be somewhere else.. Dodgy 4am maths tells me that moving a microphone 1mm is the same as a delay of 50 samples (using 30cm=1msec), so 1/25 of a mm difference is two samples? That hardly seems like it would cause a huge phase issue. Perhaps the output of one of the converters is 180 out of phase, that would cause issues.

[Alistair]

No, there's an inherent error rate in my maths of a couple of samples. It's more a problem when it's 90ish samples out. Either way, regardless of maths, I can hear it.

[Kurt]

As you were then! Smpro audio sell a little phase rotation box, but if you had to buy 8 of them it might be cheaper to get a matching converter..