• Registered: 2010-07-12
  • Posts: 3

Topic: HDSP 9632 sample accurate looping problems

I sent this as a letter to support, but I wanted to post it here to see if anyone else had noticed the problem.


I own an RME 9632 HDSP PCI card and an AI4S-192 that I use in my home rig with a Presonus Digimax FS connected via ADAT giving me 14 inputs and 10 outputs. I was running some tests on my system a few weeks ago when I realized I can't get all the inputs to sync together sample accurately. The problem lies in the card misreading the latency on an input specific basis.

When the firmware for the card was updated with the 32 sample buffer for increased stability, it did not compensate for the ADAT inputs, so they are now 32 samples AHEAD of the analog inputs. The AI4S connected by the supplied ribbon cable is 13 samples AHEAD of the analog inputs.

I have tested this in both Reaper and Nuendo, and the results are the same. The drivers seem to be reporting the latency correctly for the analog out and in on the card itself, putting other inputs ahead of real time.

1. Looping a signal from the Analog out on the 9632 to the Analog IN on the 9632 gives you a perfect sample accurate locked signal.

2. Looping a signal from the Analog out on the 9632 to the Analog IN on the Presonus via ADAT gives you a signal that is 32ms ahead of the original signal. Ahead of "real time."

3. Looping a signal from the Analog out on the 9632 to the Analog IN on the AI4S gives you a signal that is 13ms ahead of the original signal. Once again, ahead of "real time."

4. Looping a signal from the Presonus Analog output via ADAT to any input increases this by yet another 32 samples ahead of the original signal.

This causes all kinds of problems with phase, etc, when using multiple inputs for a signal and is a pain to deal with in some cases. It seem obvious to me that the 9632 is reporting this new latency of 32 samples added some time ago and the ADAT is not being delayed by those 32 samples along with the Analog Ins and Outs. I'm not quite sure what would be causing the 13 sample offset of the AI4S, as this is bridged right to the card itself.

Any help on this matter would be appreciated. Also an update to fix these problems would be a big help to your user base, rather they realize it or not.

2 Reply by MC

  • MC
  • MC
  • Administrator
  • Offline
  • From: Germany/Thailand
  • Registered: 2007-04-06
  • Posts: 25,688

Re: HDSP 9632 sample accurate looping problems

There is no way to signal more than one latency under ASIO, so the analog one is reported. Has been like that in older drivers too. A digital loopback will have a negative shift then, fully normal again. Also mentioned in the manuals.

The analog board that you use has different converters, giving 13 samples difference. In real world this has no meaning, and without measurement you never would have noticed it.

Regarding 2: I hope you meant samples, not milliseconds. This will happen when the external converters are quicker then the 9632 ones, and again there is no way to change that except for a custom offset defined in your DAW software.

Regards
Matthias Carstens
RME

MC's Website

  • Registered: 2010-07-12
  • Posts: 3

Re: HDSP 9632 sample accurate looping problems

Well, I would argue that in my "real world" it's a bit annoying. I didn't just feel springy one day and run some random tests. I noticed problems and tried to track them down. Adding outboard compression was giving me a signal that was ahead of my project time, switching preamps on a dual miced guitar cab was ruining the alignment I had on the other preamp, etc.

Bare with me here as I try and understand the impossibility of it all. The AI4S aside, the release notes for the driver says..

- Better performance on playback. We added a small safety buffer of 32 samples on the playback side.

I find it odd that this is exactly how far off the ADAT inputs are from the analog inputs on a loop test. If the latency cannot be adjusted on an Input by input basis, then how are the analog outs being adjusted by this amount without effecting the ADAT?

I love RME, and believe your products to be far superior to most. I'm just trying to wrap my head around why it's impossible for this system to be in sync on a sample by sample basis.

Thanks again,
Billy

  • Registered: 2010-07-12
  • Posts: 3

Re: HDSP 9632 sample accurate looping problems

And yes, I meant samples. Sorry, I typed that up late at night.

5 Reply by MC

  • MC
  • MC
  • Administrator
  • Offline
  • From: Germany/Thailand
  • Registered: 2007-04-06
  • Posts: 25,688

Re: HDSP 9632 sample accurate looping problems

> I find it odd that this is exactly how far off the ADAT inputs are from the analog inputs on a loop test. If the latency cannot be adjusted on an Input by input basis, then how are the analog outs being adjusted by this amount without effecting the ADAT?

> I love RME, and believe your products to be far superior to most. I'm just trying to wrap my head around why it's impossible for this system to be in sync on a sample by sample basis.

Please check this forum, we had this discussion before and also over at Steinberg. Usually the ADAT ports are connected to converters, so there will be no negative delay. If the software does not support offset adjustment per track simply add a delay (VST Plugin) per track.

Regards
Matthias Carstens
RME

MC's Website