1 (edited by slohrisch 2012-03-08 23:15:20)

Topic: HDSP 9632 with AO4S-192 - Phase cancellation problem

Hi,

I sent an email to the general tech support address, but I thought I'd post this here in case the solution is helpful to others.

I'm running a 5.1 surround production setup using the HDSP 9632 and the AO4S-192 expansion board, and I'm experiencing a phase cancellation problem.  I'm feeding all 6 analogue channels into my sub (JBL LSR 4312SP), the sub is performing bass management on those 6 channels, and then the non-bass signals are passed to my L/R/C/LS/RS monitors (all are JBL LSR 4326). 

What appears to be happening is that AnalogueOuts 1/2 (balanced breakout cable) are out of phase (phase inverted) with AnalogueOuts 3/4/5/6 (AO4S-192 expansion board).  If I send the same stereo signal through all 3 stereo pairs, I get a dramatic reduction in bass once the same signal is active on AnalogueOut 1/2.  Whereas, if I have only have the signal going through Outs 3/4 and then send the same signal through 5/6, I get an appropriate boost in the bass signal produced by the sub.  If I delay AnalogueOuts 1/2 by 100ms or so, or phase invert that channel, the bass behaves as expected (I get a bass boost), so that's why I think there's phase cancellation occurring. 

I'm pretty confident that the sub, breakout cable and balanced cables are not the problem.  Specifically, the sub doesn't sum AnalogueOuts 3/4/5/6 in way that cancels the bass.  I also changed the XLR cables connecting Outs 1/2 to the sub (same problem occurs).  Furthermore, I switched to the non-balanced breakout cable for Outs 1/2 and saw the same problem. 

To rule out my software as the cause, I used Cubase 5 and I also used a 5.1 game audio engine called FMOD to produce the signal.  Both yielded the same results.

Is it possible that some setting for my RME cards is causing a phase problem between the HDSP 9632 and the AO4S-192 expansion board?  I should mention that if I route all 3 signal pairs (via the totalmix matrix) to AnalogueOuts 1/2, the summing seems to be fine (i.e. no phase problems occur at this point in the signal routing), so the problem would seem to be further down the chain. 


Some basics about my system:

Win7 64bit
8 Gigs of RAM
Intel i5 750 2.66 GHz
nVidia GTS 250
UAD-2 Quad Core PCIe card
Cubase 5

Let me know if you need any further info.

Thanks,
Sean

2

Re: HDSP 9632 with AO4S-192 - Phase cancellation problem

The 9632 and the AO4S-192 that you use most probably have different DACs, therefore have a different delay in the range of about 1 ms. You can check yourself by looking at the chips. There is a new AO4S board (for the AIO, using PCM 4104) and an old board (originally designed for the 9632, AD1852) which will match the corresponding boards/converters. The old board is discontinued for technical reasons. If you need that one please contact support.

Regards
Matthias Carstens
RME

Re: HDSP 9632 with AO4S-192 - Phase cancellation problem

1 ms would not really explain bass cancellation. Have a look at the matrix view of the mixer. Any channels phase-reversed there?

Regards
Daniel Fuchs
RME

Regards
Daniel Fuchs
RME

Re: HDSP 9632 with AO4S-192 - Phase cancellation problem

Thanks, guys. 

I looked at the components on the boards, and the AO4S-192 has one PCM 4104 DAC, and the 9632 has one AD1852 DAC.  So, I do have that mismatch problem. 

As for phase reversing channels in the matrix view, I'm not sure how to reverse the phase in there.  I took quick look through the 9632's manual and couldn't see an explanation on that.  Sorry if it's fairly obvious, but I missed it.

Thanks,
Sean

Re: HDSP 9632 with AO4S-192 - Phase cancellation problem

The minimal offset between these converters can not possibly cause even partial phase cancellation of bass signals. Please check the bass management settings. Also, please check your cables...


Are any routings in the matrix shown in blue? That would show switched polarity. Shift-click changes it back.


Regards
Daniel Fuchs
RME

Regards
Daniel Fuchs
RME

Re: HDSP 9632 with AO4S-192 - Phase cancellation problem

Thanks, Daniel.

No routings are shown in blue in the matrix.  Shift+click doesn't seem to toggle the polarity though.  It seems to have no special effect (same as just clicking normally; i.e. turns the routing ON/OFF).  Is there another way to do this?  Am I doing something wrong?  I'm using version 3.27.

I might just toggle the polarity in the matrix to work around the problem.  I've tried two different brands of cable & a different breakout cable, and I haven't observed a change in the behaviour.  The bass management hardware is processing the 4 outs of the AO4S-192 fine, but when I add in 1 or 2 of the outs from the 9632, the cancellation occurs.  Similarly, if I only have a signal on 2 of the AO4S-192's outs, and I add 1 out from the 9632, the cancellation also occurs.  So, that indicates that the bass management isn't likely the problem.  I can't prove that with 5 or 6 identical signals that it will process them correctly (since I only have 6 outs, and 2 seem to be phase inverted), but it  handles 4 identical signals ok, as long as they're all coming from the AO4S-192.  It's possible that I've overlooked some permutation though.  Is it likely that one of the cards is defective?

Thanks,
Sean

Re: HDSP 9632 with AO4S-192 - Phase cancellation problem

Pleae try routing/copying a mono signal in TM to one output of the 9632 and one of the AO4S. Then record these two to an external device's L/R channels and compare - are they really polarity reversed? Or just offset by a number of samples?

Regards
Daniel Fuchs
RME

Regards
Daniel Fuchs
RME

Re: HDSP 9632 with AO4S-192 - Phase cancellation problem

Thanks for the quick response.

Here's the experiment I ran:
- Routed a cable from one of the 9632's ouputs to the 9632's RightInput.  Routed a cable from one of the AO4S-192's outputs to the 9632's LeftInput.
- Had Cubase send the same mono signal to the two above mentioned outputs.
- Had Cubase record two mono tracks (one for the 9632's LeftInput and one for the RightInput).
- Played back the two recordings simultaneously and also mixed them together in a wave editor.  I saw about a 10dB reduction in volume.  I also looked at the 2 waveforms closely in the wave editor and saw that the peaks/valleys were inverted relative to each other. 

When compared to the source file, the signal coming from the 9632 appears to be correct (not inverted), whereas the AO4S-192 is inverted relative to the source file.  Therefore, it looks like there's a problem with the AO4S-192.  Might I have installed it incorrectly?  All I did was add it to a card slot and connect the ribbon cable.  I believe the existing drivers simply worked automatically with it.

In the interim, I would very much like to make use of the 'polarity switch' in TM, so that I can at least work around the problem.  Do I need a specific version of TM to use the shift+click polarity switch functionality?

Thanks,
Sean

Re: HDSP 9632 with AO4S-192 - Phase cancellation problem

Please mail me a very short (uncompressed PCM) clip of the resulting test (stereo file or two mono from the exact same range).

The cable to the AO4s can't really be installed incorrectly. If the channel assignment is correct, it is correct anyhow... Did you buy the expansion with the 9632 or is it newer? What is the serial number?

The polarity switch should work with the 9632, but I am not 100% sure, to be honest... MC might confirm (or not).

Regards
Daniel Fuchs
RME

Regards
Daniel Fuchs
RME

Re: HDSP 9632 with AO4S-192 - Phase cancellation problem

RME Support wrote:

The polarity switch should work with the 9632, but I am not 100% sure, to be honest... MC might confirm (or not).

There's no Polarity switch for the HDSP 9632!

Re: HDSP 9632 with AO4S-192 - Phase cancellation problem

Dandruff wrote:

There's no Polarity switch for the HDSP 9632!

Thanks, wasn't sure... HeadScratch

Regards
Daniel Fuchs
RME

Re: HDSP 9632 with AO4S-192 - Phase cancellation problem

The files have been sent.

I bought the 9632 around Dec. 2009, and the AO4S-192 in Jan. 2012 (serial# 23143320), so the expansion is much newer.

Thanks,
Sean

13

Re: HDSP 9632 with AO4S-192 - Phase cancellation problem

I have no such phase error in my setup. An easy explanation for the effects that you have would be that the balanced breakout cable is soldered wrongly. Just swap + and - inside both out XLRs and the problem is gone.

To check the real problem you need to use a polarity test signal. Can be self-made, just draw a positive peak of some samples length in an editor. Now play this and record it with the balanced breakout cable having XLR plugged directly into XLR. The recorded signal must be a positive peak again.

If so then out and in are either correct or both inverted. You then need to use a different breakout cable to check again.

If you have the unbalanced breakout of the 9632 then you could do the same test with the positive peak signal and see if it also matches.

Regards
Matthias Carstens
RME

Re: HDSP 9632 with AO4S-192 - Phase cancellation problem

I did the above mentioned recordings on my balanced breakout cable, but I also tested my unbalanced breakout cable to rule out that the breakout cable is the culprit (but I didn't do a recording test).  I saw similar phase cancellation with the unbalanced breakout cable.  This test with the unbalanced breakout cable also required me to replace two XLR-to-XLR cables with unbalanced cables, so that should rule out that the 9632's XLR cables are phase inverted.   However, I have yet to replace all 4 of the TRS-to-XLR cables that I have running out of the AO4S-192 and perform the same recording test.  I'll see if I can dig some up (I only bought what I needed).  It seems crazy that all 4 would be phase inverted, but maybe the manufacturer had a bad batch. 

Thanks,
Sean

15 (edited by slohrisch 2012-03-15 04:58:31)

Re: HDSP 9632 with AO4S-192 - Phase cancellation problem

As a final test, I bought another brand of TRS-to-XLR cable and it doesn't exhibit the problem, so I somehow managed to get 4 bad cables!  Sorry for wasting your time with this.  I should have known better than to think that RME might have produced a defective product.  I'm glad to at least have this figured out now.  Again, sorry for the wasted time and energy, and thanks for the great help provided.

Sean