TOPIC:

I've tried dozens of times, but I can't get Dirac to stop applying a 0.25/0.28 ms delay on my 2.0 system (DDRC-24). I put the mic at the exact center (tape measured lots of times), but Dirac still applies delay, IDK why. There are no strong reflection points apart from a side wall 30cm from the left speaker, nor any upstream processing.

The only way to get it to zero (or near zero) input delay is by moving the mic about 5 cm to the right (off center)

I don't want any delay on my system because I will be seating at the center spot and I wouldn't be able to keep my mind in peace knowing that one channel is delayed.

So, I believe I have two options:

1) place the mic 5cm off-center to trick Dirac to set zero delay (with potential downsides to equalization and individual SPL adjustment)
2) keep the mic centered and compensating the left channel Dirac input delay by applying the same value to the right channel's output delay.

Would option two work? Any potential downsides?

I can't move anything else in my room and I've checked and measured everything as best as I could.

Thanks

Edit: apparently, the PC version of Dirac allows delay adjustment. Why can't this be implemented on the Minidsp version?

You've talked about measuring left to right distances but I wonder if one of your speaker may be closer by being a bit further from the front wall?

Ultimately if Dirac is measuring a difference like this at the MLP then it makes more sense to me to apply this rather than try to cancel it out.

Ultrasonic wrote:

You've talked about measuring left to right distances but I wonder if one of your speaker may be closer by being a bit further from the front wall?

Ultimately if Dirac is measuring a difference like this at the MLP then it makes more sense to me to apply this rather than try to cancel it out.

Hi,
They are set exactly at the same distance from the front wall.. IDK why this is happening.
The right speaker is indeed meters away from the side walls (no reflections), while the left speaker is only 30cm apart from it's nearer side wall. But still it seems strange to me that this first reflection would trick Dirac into applying a 0.25ms delay..

flaviowolff wrote:

They are set exactly at the same distance from the front wall.. IDK why this is happening.
The right speaker is indeed meters away from the side walls (no reflections), while the left speaker is only 30cm apart from it's nearer side wall. But still it seems strange to me that this first reflection would trick Dirac into applying a 0.25ms delay..

I don't think the first reflection is having any effect at all. I say this both as it makes no sense to me that it would and because I have a speaker about the same distance from a side boundary and don't see the same effect that you do.

You can cancel the correction out using your option 2 but I wouldn't as I think there is something causing a real difference that Dirac is trying to correct.

You could use Room EQ Wizard to make a measurement independent of Dirac to see what is going on if you wanted. Use one speaker as the acoustic timing reference for separate measurements of each speaker and then compare the impulse responses.

flaviowolff wrote:

They are set exactly at the same distance from the front wall..

How square are the walls in your room actually? If the front wall is at a slight angle to the side walls, or not flat, then each speaker being the same distance from the wall could actually still mean they are different distances from your listening position. Checking the latter would be another thing you could do.

I also had one set of Dirac1 9 measurements calculate with 0.25mS delay on one channel. I knew that was wrong. My room setup is extremely symmetrical and centered.

I can use a tape measure from each tweeter to the nose of the mic and then adjust the mic to be less than 1/4" (5mm) difference.
I can use another mic/program to prove the acoustic delay difference is zero. OmniMic shows the smoothest treble and greatest extension to 20kHz whenever the mic is perfectly centered.

Since I would then switch mics back to UMIK for Dirac, I eventually setup a crosshair laser between the speakers and pointed to the "perfect" mic position. This way I could be sure that both mics stay centered for testing.

In my last and best Dirac1 session, I achieved an acceptable calibration and both channels were set to 0.0mS.

For the OP's situation, perhaps you can put something thick and soft against that nearby sidewall. Even a blanket held up with a broomstick should dampen the reflection. Maybe then Dirac would show a low acoustic differential.

Philmeister wrote:

I also had one set of Dirac1 9 measurements calculate with 0.25mS delay on one channel. I knew that was wrong. My room setup is extremely symmetrical and centered.

I can use a tape measure from each tweeter to the nose of the mic and then adjust the mic to be less than 1/4" (5mm) difference.
I can use another mic/program to prove the acoustic delay difference is zero. OmniMic shows the smoothest treble and greatest extension to 20kHz whenever the mic is perfectly centered.

Since I would then switch mics back to UMIK for Dirac, I eventually setup a crosshair laser between the speakers and pointed to the "perfect" mic position. This way I could be sure that both mics stay centered for testing.

In my last and best Dirac1 session, I achieved an acceptable calibration and both channels were set to 0.0mS.

For the OP's situation, perhaps you can put something thick and soft against that nearby sidewall. Even a blanket held up with a broomstick should dampen the reflection. Maybe then Dirac would show a low acoustic differential.

I have just done a new set of measurements and this time I was able to get a more acceptable 0.06ms delay.
Since I don't want to keep measuring until I can get 0.00, would there be any harm by setting a 0.06ms output delay on the opposite channel?

0.06mS = 60 microseconds, which is about 20mm or 3/4" time of flight difference. That is easily measured with a tape measure, as you already know.

If this was happening to me, and since I have another time-based measurement system, I would go to my other system and use the treble quality method I described earlier. The only way to know if your issue is a technique problem, or hardware/software problem is to get a "second opinion" without a completely different measurement system.

Philmeister wrote:

0.06mS = 60 microseconds, which is about 20mm or 3/4" time of flight difference. That is easily measured with a tape measure, as you already know.

If this was happening to me, and since I have another time-based measurement system, I would go to my other system and use the treble quality method I described earlier. The only way to know if your issue is a technique problem, or hardware/software problem is to get a "second opinion" without a completely different measurement system.

Considering that the FR measurement is probably fine because its just a 20mm deviance (and there are other 16 averaged measurements contributing to the result), I'm thinking about just setting the opposite output delay and forgetting about it. What do you think? Doing a recalibration is a big hassle in my daily routine ATM

250uS = 3-3/8" difference.

You could find the first reflection spot on the nearby wall, and then measure the distance from the tweeter to that spot, and then from that spot to the microphone. Does that add up to ~4" greater distance than the direct path to the mic?

The wall is very close to that speaker, and so not only is the path length not much longer, but the attenuation due to the extra distance is not very much either.

I agree with the hassle of another measurement session. Each acquisition and calculation takes up 1/2 hour. Perhaps the next time you do a session, try the blanket against the first reflection spot. That will attenuate the reflected signal at high frequencies (where the Dirac uses to determine times of flight).

Philmeister wrote:

250uS = 3-3/8" difference.

You could find the first reflection spot on the nearby wall, and then measure the distance from the tweeter to that spot, and then from that spot to the microphone. Does that add up to ~4" greater distance than the direct path to the mic?

The wall is very close to that speaker, and so not only is the path length not much longer, but the attenuation due to the extra distance is not very much either.

I agree with the hassle of another measurement session. Each acquisition and calculation takes up 1/2 hour. Perhaps the next time you do a session, try the blanket against the first reflection spot. That will attenuate the reflected signal at high frequencies (where the Dirac uses to determine times of flight).

Thank you for your input. I'll try that when I can. Meanwhile, I will try inserting the corresponding output delay to the opposite channel and check if I can hear anything.

To be clear, is your left speaker 30cm from the left-side wall, and Dirac puts 0.25mS delay also to the left channel?

I really am pretty confident the nearby boundary is having no effect at all here. The reflected signal will be delayed and lower amplitude than the direct signal so I see no scope for confusion.

flaviowolff wrote:

I have just done a new set of measurements and this time I was able to get a more acceptable 0.06ms delay.
Since I don't want to keep measuring until I can get 0.00, would there be any harm by setting a 0.06ms output delay on the opposite channel?

I'm sure that would be fine. Realistically I seriously doubt you'd notice any difference between leaving the 0.06 ms or deliberately cancelling it out with a compensating delay. Why don't you try comparing to see?