Drift compensation?

[bobkatz]

Hi, John! Long time no talk. Because REW just works so well and you keep on keeping on!

I will be aligning my new bedroom 7.1 system soon. I'll be needing to use a multi-output or "Aggregate" device in MacOS. There is an option for "drift compensation" and I don't know if I need it for this measurement application or not.

I will use my Prism Lyra 2 Interface at 48 kHz on internal sync for the measurement microphone and for timing reference output and timing reference input.
I will use an HDMI output (through a USB to HDMI adapter) to feed an Onkyo AVR PCM multichannel.

So I make an aggregate device between the Prism and the Onkyo feed so that REW will see all the inputs and outputs of both devices available.

Note that I don't care that the latency of the Onkyo Receiver and that of the Prism Lyra are different, so the timing reference will not be absolutely accurate. All I care about is that I am able to get an absolute relative reference between loudspeakers so I can time align all the channels to each other.

So do I need to enable drift compensation in the apple aggregate device menu?

 

[John Mulcahy]

Drift correction is needed for different physical devices which don't share a clock source, so it should be selected. REW can also perform clock rate correction (the options are in the Analysis preferences), but for that to work all your outputs (including timing ref) would need to come from the Onkyo.

 

[bobkatz]

Drift correction is needed for different physical devices which don't share a clock source, so it should be selected. REW can also perform clock rate correction (the options are in the Analysis preferences), but for that to work all your outputs (including timing ref) would need to come from the Onkyo.

Thanks. I was just worried that because the USB-based HDMI audio output is on a different clock than the Prism interface, whether the Drift correction might make things worse, or better, in a random sort of way. I'm only concerned about the precision of the timing reference for setting speaker delay.

I bet, In truth, typical master crystals these days are so stable and frequency accurate, that it's likely that the speaker location error, with or without drift correction, will be in small numbers of microseconds.

For extreme exactitude, I could lock the Prism somehow to the HDMI generator. I could do that by using a box that receives HDMI input and outputs SPDIF (these are fairly common) because the Prism can lock to an SPDIF source.