extra latency using SPDIF

  • I checked this again a few days ago with my RME Fireface 400. I did a reamping loop and recorded spdif and main out simultaneously. At 44.1 kHz, spdif arrives around 1 ms later than analog. Interestingly (and that is why I was measuring), at 96 kHz, spdif is half a ms later still that at 44.1 kHz (1.5 ms later than analog in total). When I measured a few months ago, I measured higher additional latencies with spdif at 44.1 kHz, around 2-3 ms. I did not investigate wether this has to do anything with the specific patch (I used the same patch in the latest comparison, but a different one in the first).


    It would be interesting if tech support has anything to offer on this subject!


    It is defintely related to the audio interface, not the Kemper. On my RME HDSP 9632 , SPDIF arrives a bit earlier than analog.

  • Hm, the latency caused by the analog input of the audio interface should only be a few samples (below 100 µs) and any compensation should be in this range.

    I just tested this (I like testing ;) ) : The analog latency of the Fireface is negligible and the same spdif reamp loop recorded through a behringer ada8200 for some reason arrived 0.6 ms before the output signal. I just sent a sinus through an analog ramp box and through the Kemper (or now without Kemper) and record the signal. Everything is done in Logic.


    With the Kemper I get up to 3 ms additional latency over spdif compared to analog. Can you (tylerhb) maybe confirm this with your setup? I have the latest profiler OS btw.

  • No news from my side. I just ran the test again with the newest profiler OS. See the screenshot, the uppermost track is the sinus signal that is sent out to the Kemper, below is the analog input of the RME Fireface 400, the lowest track is the spdif input.screenshot:

  • I just noticed something weird: When I ran the test again to day I had the same additional latency of spdif compared no main out of 120 samples at 44100 Hz. After I went into the Kemper's output menu and dialled the sample rate to 96k and immediately back to 41k, the delay was reduced to about 20 samples. This seems to be reproducible behaviour, I measured the same variations earlier. Strange, right?

  • Hello everyone,

    I bought a new Kemper Profiling AMP and I love it!

    Unfortunately I have the same issue with extra latency using SPDIF!


    With my new Kemper and latest KOS connected to my RME HDSPe AIO the S/PDIF signal arrives 1 ms later than the converted(!) analog signal.

    Is anyone at Kemper looking for this issue?


    Regards

    Johann


  • I'm glad I'm not the only one anymore, who cares about this :)


    I did a few SPDIF latency tests in August 2018 (with the newest OS at the time, I think it was 5.5.2) , which looked very much like the one in post #27 from Refuge Studio. I got latencies between 7.8 ms and around 10 ms, all with clean profiles. Constant latency was around 10.6 ms, but strangely enough, I never got it to be constant. It changed all the time.

    So I called up the Kemper support hotline. They didn't know about this issue ("10 ms? That's worse than the old Line 6 POD!"), but the man on the phone said it must be part of the new update, where they added the 96 khz sample rate, which doesn't give you a better sound, since it's just upscaled, but people requested it. And he said something like it's technically not possible to have 44.1 khz AND 96 khz sample rates AND low latency. He was kind enough to send me the last OS (5.3.1) where SPDIF has low latency (around 3 ms; constant latency around 4.9 ms) , and he said he will put it on their Bug Fixes list.


    In August 2019 I did new tests with Kemper OS 7.0: A clean profile was 7.5 ms late, with "Constant Latency" on it was 9.2 ms later than the original signal. So apparently they didn't fix it. I called up the support hotline again, asking if I was the only one who uses SPDIF for recording, since it's unusable in this form. From the way it sounded, I apparently was the only one. He said that most studios he knows use the Main Output and a cable (that has it's own sound, and can be faulty, and needs space) that plugs into their interface preamps (which have their own sound), where they have an additional AD/DA conversion (which has it's own sound, and adds latency). But he said he will put it on the Bug Fixes list.


    So here I am today, not knowing what I should do. I can not upgrade Rig Manager, since version 2.1.61 is the last version that works with my old low latency 5.3.1 KemperOS. But I am not sure the old Rig Manager will run on a new Mac with OS Catalina, which I will need to upgrade to in the near future.


    I really love my old fashioned low latency SPDIF recording, I can quickly choose between stereo recording, or DI on one track and Master on the other, which is something I do all the time. But in the current form, I think SPDIF on the Kemper is dead. Is there anybody who cares for groove that uses it?

  • I just did some testing with the newest Kemper OS 7.3.2 at 44.1 kHz

    • Source is a 440 Hz Sine Wave (Track 1) from the DAW -> Headphone Out of Interface -> Kemper Front Input
    • The loaded profile is TJ-JTM 45 Max-2 (Distortion all the way up, no FX)
    • Track 2: SPDIF Output Git
    • Track 3: SPDIF Output Master Mono
    • Track 4: SPDIF Stereo
    • Track 5: Kemper Main Out (Master Mono) into Line Input of my RME Fireface 800 Interface

    Test 1: Constant Latency OFF

    • SPDIF Latency: 6.1 ms
    • Main Out Latency: 3.4 ms



    Test 2: Constant Latency ON - Take 1

    • SPDIF Latency: 7.8 ms
    • Main Out Latency: 5.1 ms



    Test 3: Constant Latency ON - Take 2

    • SPDIF Latency: 6.8 ms
    • Main Out Latency: 4.0 ms


    Results:

    • Constant Latency is in fact Random Latency
    • SPDIF is slower than Main Out since OS 5.4, even though there is one less AD/DA conversion

    What shall we do?

  • RME UFX+ @ 44.1kHz, 32bit buffer

    Total system roundtrip latency (Interface & Cubase Pro): 3.287ms

    Kemper Stage (S/PDIF in slave mode)

    Main Out & S/PDIF output set to Master Stereo

    All slots are switched off (Stomp, Stack, Effects)


    1. KPA analog (Constant Latency off):

    Total latency: 172 samples = 3.900ms

    Net KPA latency: 0.613ms


    2. KPA digital (Constant Latency off):

    Total latency: 226 samples = 5.125ms

    Net KPA latency: 1.838ms


    The latency diff between analog and digital is 1.225ms


    3. KPA analog (Constant Latency on):

    Total latency: 277 samples = 6.281ms

    Net KPA latency: 2.994ms


    4. KPA digital (Constant Latency on):

    Total latency: 332 samples = 7.528ms

    Net KPA latency: 4.241ms


    The latency diff between analog and digital is 1.247ms

  • That's interesting. I did my tests with the PowerHead. Maybe the Stage has different SPDIF hardware (as somebody suggested here), since it has the slave option, which I don't have.


    Lightbox, is your Constant Latency really constant though, is it 7.528ms every time?

  • I figured that my test can only be complete when I compare my findings above with how it could be, how it has been for the longest time. So I made a rollback to the fastest Kemper OS in the history of the world, version 5.3.1 ?

    • Kemper PowerHead at 44.1 kHz, Constant Latency off
    • Main Output Latency: 4.0 ms
    • SPDIF Latency: 2.5 ms ⚡️

    Constant Latency is indeed constant at 4.1 ms for SPDIF and 5.6 ms for the Main Output.

  • Maybe the Stage has different SPDIF hardware

    Maybe this ... or they just forgot to "wire" it accordingly in the board layout cause they thought it's not necessary.



    is your Constant Latency really constant though, is it 7.528ms every time?

    Didn't spend enough time to tell for sure. What I can tell in addition to the above data:

    I did one run with all slots loaded (but switched off) and another run with all slots cleared (and off).

    It was exactly the same result, so I didn't mention it in my prior post.