The rear output is delayed by about 8 samples at 48kHz on my SBLive! CT4832

Tril · Jun 18, 2006, 07:37 PM

I use an headphone amp I built that is called "Dynamid" on Head-Fi. It has balanced inputs and ouputs. What that means is that it needs four inputs. I use the kX drivers for this. It also has four outputs, two for each earpiece.

The DSP is used to produce the balanced signal by using the Phase plugin.

In the DSP :

I connect the left and the inverted left to the analog front left and analog front right.
I also connect the right and the inverted right to the analog rear left and analog rear right.

It always sounded as if the sound was not centered. I thought it was because of a volume imbalance between left and right so I was using a Vol plugin. It never sounded totally centered.

I was beginning to think that my headphones were broken.

Today, I connected a TheSmallDelay plugin to the analog front outputs and I played with it. I found out that I need to delay the front output by about 8 samples to get the sound in sync with the rear ouput of the card.

What that means is that, on my soundcard, the rear output is delayed by about 8 samples (0.167 ms) compared to the front output. I need to delay the front output by the same amount to get the front and the rear in sync in my headphones.

I have no idea if it's like that with other soundcards using the emu10kX chip. If it's like that, it means that TimeBalanceV2, the plugin I wrote to delay the channels to get the sound from all the speakers to you at the same time, makes slightly incorrect calculations for the affected cards.

**dj_stick** · Jun 18, 2006, 08:10 PM

this could be due to the different DACs used for the front and rear outputs - the phase variation would cause 4.1 speakers to lose a lot of the low end…

Tril · Jun 21, 2006, 06:23 PM

I reinitialized my DSP and rebuilt it almost the same as before and it's now the back analog output that needs to be delayed to be in sync with the front. It's still the same delay.

Lex Nahumury · Jun 21, 2006, 07:51 PM

>>The rear output is delayed by about 8 samples at 48kHz on my SBLive! CT4832
That doesn't really surprise me.
Front uses AC97 codec through AC-Link protocol,
while Rear uses different codec through I2S protocol.

Even if you would apply exact same signal in DSP straight to those bare I/O port registers,
the resulting audio signals will most likely have phase delays.

/Lex.
[edit]
On second thought, I think the protocols AC-Link/I2S are the same.
See docs for details.

Russ · Jun 21, 2006, 11:07 PM

Another thing to consider is the plugin ordering. From your description it sounds like your setup is something like the following:

Surrounder(fL/fR) -> epilog(fL/sR)
Surrounder(fL/fR) -> Phase -> epilog(fR/sL)

Under that scenario, if Surrounder was ordered before epilog, but Phase was ordered after epilog, then you would have a delay of 1 sample between the inverted/non-inverted signal which might cause some problems.

It might be better to make a 2 input / 4 output plugin, instead of using the Phase plugin, to make sure the path is equal to epilog for all 4 signals (and then the plugin ordering should not make a difference).

i.e.

Code:

input in_left, in_right;
output out_left+, out_left-;
output out_right+, out_right-;

macs 	 out_left+,  in_left,  0x0, 0x0;
macs 	 out_left-,  0x0,  out_left+,  0x80000000;
macs 	 out_right+,  in_right,  0x0, 0x0;
macs 	 out_right-,  0x0,  out_right+,  0x80000000;

It is probably not the cause here (and maybe you allready considered it), but it could be a contributing factor. Just something to think about...

BTW: I know you can make your own plugin, I just figured I would post example code for others, etc.

Lex Nahumury · Jun 22, 2006, 06:55 AM

That's why I said;
"..Even if you would apply exact same signal in DSP straight to those bare I/O port registers..."
hence, ruling out any possible DSP induced delays.

I remember I did such tests and IIRC there will always be phase diff/delays because of diff. codecs used.

Russ · Jun 22, 2006, 11:51 AM

Quote:

Originally Posted by Lex Nahumury

That's why I said;
"..Even if you would apply exact same signal in DSP straight to those bare I/O port registers..."
hence, ruling out any possible DSP induced delays.

I remember I did such tests and IIRC there will always be phase diff/delays because of diff. codecs used.

Sorry Lex, I did not realize that is what you meant with that statement.

Chester01 · Jul 8, 2006, 08:41 AM

is this an issue on other cards? Such as the SB0350? I have no idea how I would test the delay caused by the DACs without more testing equipment...

Tril · Jul 8, 2006, 04:22 PM

Don't worry too much about it. If you take into account the speed of sound in the air, 8 samples at 48 kHz is equal to a distance of about 6 cm.

Place your rear speakers 6 cm closer or farther than their current position and you get a delay of 8 samples.

I only noticed it because I'm using headphones. I listen with one ear from the front output jack and the other ear from the rear output jack.

Chester01 · Jul 8, 2006, 06:08 PM

I am currently using the front and rear channels to deliver balanced audio into a DBX Driverack PA... Therefore I MAY be delaying one channel (left or right) by 8 samples... This is why I was slightly concerned... I used TheSmallDelay plugin to delay the cold and then the hot outputs by 8 msec and swapped the physical outputs into the amp (so I would use the rear cold with the front hot...) I have found that there are phasing issues when attempting to delay to compensate for the possible 8 msec delay caused by the output. Therefore it is not an issue with the SB0350

Tril · Jul 9, 2006, 01:43 AM

I am talking of a delay of 8 samples, not 8 ms.
8 samples = 0.167 ms

Chester01 · Jul 9, 2006, 06:29 AM

Sorry, that is my error in the post, I actually did use 8 samples (or 0.1666666667 milliseconds) TheSmallDelay is a 2-2400 sample delay. Sorry