I don’t want you to go out of your way but if you have any other stats like inputs->mixer dir or inputs-thru machine would love to know…
My testing showed using the “Dir” option in the mixer does not add any latency, but does cause about +30º phase shift at 12kHz… Which is odd because if it’s not adding latency it must be before the converter… possibly some sort of aliasing filter on the input?
Also even going via “Direct”, if you turn on Trk 8 as “Master” latency increases by about 0.75ms.
EDIT: Sorry, I just went back and checked my measurements. My mistake. latency on the “DIR” mixer was +1.29ms compared to the cable loop, so yes DIR must be on the other side of the converter. (Phase shift and additional latency from Trk 8 Master still appears to be correct though)
Thanks Avantronica for making me second guess that.
Certainly something to keep in mind if you’re using the OT for parallel processing and the like.
It has been officially confirmed however that even DIR is post ADC, everything happening is post conversion afaicr
Interesting… I wonder why there isn’t a change in latency then? Maybe I need to remeasure that part…
Can you link me to where it was “officially confirmed” so I can get some more context?
EDIT: I just went back and checked my measurements. My mistake. latency on the “DIR” mixer was +1.29ms compared to the cable loop, so yes DIR must be on the other side of the converter.
Nice work! thanks for keeping up with it.
I made my tests with OT internal recording and there were no added latency. That would mean the latency is added globally?
Something also to consider while making these tests is the latency compensation. Did you both have it off? …On?
I might need you to clarify what you mean there.
I did no internal recording when I made these tests, it’s just using smaart to measure round trip latency.
I guess I can think of one way I could check if the delay increase was “global” I could measure an input on tracks 1 thru 4 with neighbours and another input on say track 5 going straight to the master out and see if there is a difference in the time it takes to make the trip.
Is that what you’re getting at?
Isn’t delay compensation on the OT just a switchable delay of a fixed value on the input?
If I understand correctly (and I may not, I’m certainly not an OT power user) that wouldn’t make any difference to the latency increase from adding a neighbour machine if that’s the case…
Thread is about neighbor latency but since tests were done I was curious about mixer dir, and thru latencies which I inquired about earlier in the thread and you were helpful with the dir stat. The delay comp only affects mixer direct, which a stat was posted, seems you probably had it off though as it’s quite low. Still curious about thru if anyone tested.
I’m gonna say I probably had it off at the time.
I can probably set up the test rig again sometime this week though…
Or maybe Elektron will chime in and just tell us lol.
That’s quite a lot. (For comparison, my Soundcraft digital mixer has > 0.8ms from mic in to line out, all processing egaged (and that’s a lot of processing)). I’ll fire up Smaart to double check if I have a break from work
Please do, I’d like to see if someone else can replicate the results.
I agree that’s a lot, but it is an older device these days.
I verified against a cable loop before putting the OT in the chain so I’m fairly confident… but we’re only human after all. Avantronica already helped bump up the standard getting me to double check.
Which actually turned out to be an error in my note taking rather than the measurement but just goes to show right?
Yes I have some doubts!
My computer is still in storage and my car is messed up…
When I can get a ride to get it I’ll jump in on the analyses…
Meanwhile I’m stranded and living in a music studio, could be a lot worse…
Audio latency measurements (input to output), delay compensation=off
Thru: 3.10 ms
+NEI : 4.54 ms
each NEI adds 1.46ms
With master track:
DIR: 1.98 ms
Thru: 3.81 ms
+NEI: 5.27 ms
each NEI adds 1.46ms
-master track adds 0.71 ms, to the chain
-delai compensations adds 1.83ms to DIR so that it aligns with a thru track, but doesn’t compensate for additional NEI’s
-buffers for NEI are instanciated when the track is first played
-buffers for NEI remain allocated even when the track is removed, until the source track is stopped
-playing back a recording buffer through it’s flex track doesn’t add latency (maybe at the sample level, but that’s beyond what I can measure using IR)
-Midi to audio latency might be different, I didn’t measure that and I don’t know how midi buffers are clocked inside (this is not ASIO ) Probably less.
That’s consistent with your test results.
Intersting results. Thanks both.
Neighbor latency is noticeable by ear, 2 identical tracks, 1 with Nei…
I have to make my internal tests again…
I should have known my neighbours machine could increase latency…
Latency is relatively high*. AD/DA converters seem to be on the slower side. Response curve is not entirely flat and there is some weirdess at the upper end of the spectrum, must be some capacitIve effect on the outputs acting like a 1st order low-pass filter, maybe in an effort to warm up the sound a bit. @psynthetic also measured the symptoms. When I got my OT the first thing I did was a I/O comparison and I noticed this immediately even if it’s very subtle. It doesn’t make much difference when playing x0x samples but when sampling acoustic instruments or even electric guitars and basses it’s audible. Some of the harmonic content is altered ever so slightly resulting in a somewhat duller rendering than my other samplers/loopers.
(please note that this is not a rant, I really like this machine extraordinaire, I’m just confirming my real-world experience with measurements by interpreting the results ).
*EDIT: the latency being somewhat on the high side shouldn’t be much of a problem in a smaller rig or a analog environment. But in a digital environment, all latencies add up (mixer, outboard processors in insert loops , inline processors such as master compressors, amplified monitors… all add a microsecond here, one or two there…not forgetting sources such as digital guitar processors)