it’s my first Motu interface so I didn’t have any previous generation devices, but so far had zero issues with the driver or their CueMix app which is actually very nice, so maybe the older generations had issues but the mk5 is solid
off topic there is a click in my left headphone even when I pan all the way right but it’s only above a certain volume level and it’s only when the power is on (noise cancelling) when power is off there is no click. did I fuck up one of the drivers? Oh and it is only audible (for whatever reason) when I’m listening to unmastered music, if I listen to youtube or real music it’s inaudible, only when I’m making beats right out of the headphone jack on a box. I can only think that I blew up one side or something. sony noise cancelling, not super great headphones but not super cheap either.
did you try with different headphones or maybe speakers?
if it’s driver or software thing it should be audible on every device, if it’s not then it’s probably the headphones…
I meant the driver like the little thing that moves to make sound inside the headphone. it does it on elektron boxes and also plugged into the computer if I’m listening to something not mastered. But the conditions are different for each, it’s a bit strange. I think it’s odd also that it clicks if there’s no noise going to that side, and only when the battery is powered on. You’re probably right that it’s the headphones which is what I was afraid of.
Since people are throwing a lot of measurements around, I thought I’d post the theoretical limits.
buffer size / sampling rate in kHz = latency in ms
so:
16 words / 48kHz = 0.333…ms
16 words / 44.1kHz = ~0.363ms
(where word size = sample depth)
As you can no doubt tell from this, the vast majority of the latency is in software. Drivers, undocumented buffers, DAW routing, interface post-processing/routing, and general shenanigans. A tiny bit will come from the converters, but a random TI reference has ΔΣ-ADCs roughly in the 0.05-0.25ms range. Nothing to cry about, but make sure you get the really good converters if you’re building digital effect processors I guess.
Anyway, here’s a big fat table:
44.1 kHz/16: 0.36281179138322ms
48 kHz/16: 0.3333333333333333ms
88 kHz/16: 0.1818181818181818ms
96 kHz/16: 0.1666666666666667ms
192 kHz/16: 0.0833333333333333ms
44.1 kHz/32: 0.7256235827664399ms
48 kHz/32: 0.6666666666666666ms
88 kHz/32: 0.3636363636363637ms
96 kHz/32: 0.3333333333333333ms
192 kHz/32: 0.1666666666666667ms
44.1 kHz/64: 1.4512471655328798ms
48 kHz/64: 1.3333333333333333ms
88 kHz/64: 0.7272727272727273ms
96 kHz/64: 0.6666666666666666ms
192 kHz/64: 0.3333333333333333ms
44.1 kHz/128: 2.9024943310657596ms
48 kHz/128: 2.6666666666666665ms
88 kHz/128: 1.4545454545454546ms
96 kHz/128: 1.3333333333333333ms
192 kHz/128: 0.6666666666666666ms
44.1 kHz/256: 5.8049886621315192ms
48 kHz/256: 5.333333333333333ms
88 kHz/256: 2.9090909090909092ms
96 kHz/256: 2.6666666666666665ms
192 kHz/256: 1.3333333333333333ms
44.1 kHz/512: 11.609977324263038ms
48 kHz/512: 10.666666666666666ms
88 kHz/512: 5.8181818181818183ms
96 kHz/512: 5.333333333333333ms
192 kHz/512: 2.6666666666666665ms
If you have some oddball rate/size (is 88kHz even a thing?) that isn’t listed it’s probably half or double one of these. Or maybe half and double. Or double double or half half. You get the idea. Latency halves as you double the sampling rate, and doubles as you double the buffer size.
isn’t it 88200 which is double of 44100? that’s what’s on my ULmk5 alongside with 176400 which is 44100*4
Ah yeah, that’s right. I might update it later.
On Windows it was very unstable, on Mac it has been decent for a long time. It has been improved on Windows though the last years.