1 Topic by masiusima13 (edited by masiusima13 2024-09-19 06:37:24)

  • Registered: 2024-09-19
  • Posts: 1

Topic: Rme Digiface Usb/RayDat ADAT Latency Question

Hello, i recently bought an external Adat ad/da converter (Black Lion revolution exp) planning to make it a center of my studio with all the inputs and outputs being routed trough it.

The plan is sell my current interface and to soon also buy a Digiface Usb or Raydat to use it with the converter since Rme support and drivers are rock solid.
However i have a few important questions because of the issue that im currently having with my current main interface.

Im running the external adat converter with two adat cables( in+out) to my main interface which is currently an audient ID44.

The problem im having is that using the adat protocol on the id44 introduces a lot of additional latency which becomes very noticeable when playing instruments even on the lowest possible buffersize so i dont have any way to lower it.

Using RTL utility and measuring the audient's own analogue input and output loopback with buffer set at 8 samples (the lowest possible) and 48khz i was having a 4.1 milisecond of latency, which for playing guitar was fine and was almost unnoticeable.

Now, adding my external converter via adat and running the loopback test trough the external adat converter's output to input i get a latency reading of 5.6ms. This is now really noticeable when playing trough software (i am using the external converters input and its output connected to my monitor speakers) and since i cant decrease the buffer size anymore i am stuck.

I thought at first that the adat converter might have added that latency but its not the cause since i have figured it out today:

Running an ADAT cable from the audient id44's adat out back to itself's adat in and testing the pure digital adat roundtrip latency i am getting 5.1 miliseconds!!

This is a whole milisecond slower than the loopback trough the audient's ad/da converters at the same buffersize. And this measurement is purely digital and doesnt even include its own ad/da conversion which audient states at about 0.68 miliseconds!
So that means the driver's 8 sample buffer size latency is about 3.5 miliseconds + with the ad/da ends about right at the 4.1ms i was getting before. Having all that figured out means that the adat protocol is introducing 1.6 miliseconds of latency!?! Where is that coming from? And adding on top of that the 0.5/0.6 ms from the external converters ad/da we end up exactly right at the 5.6ms im getting currently which is way too much for me and impossible to use since i cant lower it any more. Seems there is no fix for this, i have contacted audient too to see if this is even normal.

So, we come to what the main question is, does the digiface usb or raydat have as much of a adat latency or are they almost zero? Ideally i dont want any added latency on top of my converters own conversion latency and the drivers buffersize, I really dont want to buy one of those interfaces and then find out im having the same issues again that i cant get the latency under 4.5 miliseconds.

  • vinark
  • DAW Mastermind
  • Offline
  • Registered: 2009-06-05
  • Posts: 2,779

Re: Rme Digiface Usb/RayDat ADAT Latency Question

No extra latency on adat with RME, zero.
But honestly, I can detect a difference between 5ms and 10ms latency, but not between 4 and 5ms. Here sub 5 a 6ms feels as no latency.

Vincent, Amsterdam
https://soundcloud.com/thesecretworld
BFpro fs, 2X HDSP9652 ADI-8AE, 2X HDSP9632
  • ramses
  • DAW Mastermind
  • Offline
  • From: UTC+1
  • Registered: 2012-10-14
  • Posts: 13,192

Re: Rme Digiface Usb/RayDat ADAT Latency Question

The RME drivers are generally better and digital transport over MADI and ADAT only has a few samples latency which depends on the sample rate. For example, forwarding of audio over MADI from a device to the next device (if some are serial chained) is exactly 3 samples at single speed (6 at double speed, 12 at quad speed).

Audio transport over ADAT should be similar.

But those 3 samples do not result in a high latency compared to your Audient interface where a setting of 8 samples introduced 4.1ms of latency at 48 kHz (including safety buffers of the driver and driver efficiency).
I mean the pure time for transfer of audio samples at the given sample rate.

On Windows with ASIO drivers, the lowest ASIO buffer size is 32 samples with HDSPe and MADIface drivers.
At 44.1 kHz (worst case, here the converters have their highest converter latency)
you still have RTL values under 3.06ms, see the table in this blog article.

https://www.tonstudio-forum.de/blog/ent … cts-en-de/

A few things to note here

1. I looked at the input and output latency in the DAW (Cubase) which takes the values from the ASIO driver
2. The values from RME ASIO drivers are accurate, therefore I didn't measure with RTL utility
3. The RTL values from the RME driver contain the converter latency for AD and DA on the card if there is analog I/O
If you connect a converter to such a pure digital card you can add converter latency from manual of the converter,
if they present you the values. Otherwise it is better to use RTL utility and to plug a loop to be on the save side.

If you compare with RME values, then keep in mind, my values are "worst case", at 44.1 kHz, where the converter latency is highest. Use the same sample rate for comparison.

According to my experience with UFX+ and UFX III playing guitar over VSTi is easily possible with ASIO buffer sizes up to 128 samples (at 44.1) where the RTL is still below 10ms. Starting with 256 samples buffer size the RTL is 13.1 ms, still playable but not so smooth as the delay becomes noticeably. But you are still able to play without slowing down and to stay in time.

BR Ramses - UFX III, 12Mic, XTC, ADI-2 Pro FS R BE, RayDAT, X10SRi-F, E5-1680v4, Win10Pro22H2, Cub14

4 Reply by ramses (edited by ramses 2024-09-19 09:13:32)

  • ramses
  • DAW Mastermind
  • Offline
  • From: UTC+1
  • Registered: 2012-10-14
  • Posts: 13,192

Re: Rme Digiface Usb/RayDat ADAT Latency Question

vinark wrote:

No extra latency on adat with RME, zero.

Sure? I ask because I remember MC telling something like only a few samples for digital transfer through ADAT.
But it's years ago when this came up on forum.

Maybe 3 like MADI (which is also digital transfer inside of a unit) or even a little less (1 or 2).

Either RME tells again or one would have to measure. For example, with a RayDAT and then take input and output latency reported by the ASIO driver and then substract the converter latency of a connected converter.
I would simply assume it is somewhere between 0 and 3 samples, which you can silently ignore.

EDIT: I found a reference
https://forum.rme-audio.de/viewtopic.ph … 442#p85442

Appears to something around 3 samples for ADAT and SPDIF (at single speed).

BR Ramses - UFX III, 12Mic, XTC, ADI-2 Pro FS R BE, RayDAT, X10SRi-F, E5-1680v4, Win10Pro22H2, Cub14