Is Dirac's DAP screwing up my absolute speaker polarity and/or relative phasing?

[wgscott]

The other day I recognized something was amiss with my speaker phasing. I switched around the wires for my left bass, and that seemed to improve things. This morning was the first time I had a chance to do measurements, using AudioTools and my iPad mini.

 

cf: Speaker Polarity | Studio Six Digital

 

Briefly, here are my results:

 

With Dirac completely out of the chain, the way I had everything wired up a week ago was correct. Each speaker cone gives a "+" absolute polarity. In the case of the woofers, I can see they go outward instead of inward.

 

With Dirac's DAP in the chain, two things happen:

 

The polarity of everything becomes globally inverted, i.e., all "+" readings become "-" readings with one exception ...

 

One of the woofers remains "+".

 

So the relative phase of the bass changes, so that the two woofers are out of phase. One goes in while the other goes out. I can see it.

 

Note that this is playing through the DAP with one of the filters I made, but the filter itself it turned to "OFF" in the interface.

 

What should I do? Is this an artifact of the filter I made, even though it is set to "OFF", or is it an artifact of the DAP itself?

[wgscott]

If I delete all the filters from the DAP interface, (rather than simply turning them to "OFF"), then the polarity is preserved. So at least the DAP itself isn't doing anything to the polarity (or phasing), i.e.

 

 

I will re-measure and re-create a filter before I conclude it is anything other than user error.

[wgscott]

I think I have a better idea of what is going on now. The mid/treble channels are phase-inverted; the others are not.

 

I believe this was a consequence of Dirac compensating for excessive speaker toe-in on my part. Hopefully I will get a chance to remeasure (and thereby test this explanation) in the near future. I've adjusted the speakers to cross slightly behind the listener, and without Dirac, the imaging is much better, so it will probably be a better starting point.

[LBob]

I think I have a better idea of what is going on now. The mid/treble channels are phase-inverted; the others are not.

 

I believe this was a consequence of Dirac compensating for excessive speaker toe-in on my part. Hopefully I will get a chance to remeasure (and thereby test this explanation) in the near future. I've adjusted the speakers to cross slightly behind the listener, and without Dirac, the imaging is much better, so it will probably be a better starting point.

 

So did Dirac invert the phase?

[wgscott]

I can't say until I do the remeasure, but I am guessing that having the speakers cross-fire in front of the listening position (even if only slightly) caused Dirac to try to compensate. I am not sure though.

 

Lots of things invert polarity, including airplay. The weird thing is that the relative phase of the mid/treble differs from the bass.

[flak]

Hello Wgscott,

 

is it possible to see what happens with one channel only, with and without Dirac?

(speaker wiring as designed by the manufacturer)

 

is there a difference between the left and right channel?

 

Ciao, Flavio

[LBob]

 

The weird thing is that the relative phase of the mid/treble differs from the bass.

 

That's what got my attention. Why/how would it do that for only one part of the frequency range?

[wgscott]

Well, it is reproducible. Here is what I did:

 

Toed out my speakers so that they cross several ft behind the listener's position. (The listener's position is where I make the calibration measurements). [The reason I did this was to eliminate the possibility that Dirac is compensating for speakers being too aggressively toed in. I had them exactly pointed to the listening position the first time around, but you never know unless you test it.]

 

Turned off my subwoofer to remove any complications.

 

Verified the polarity of each driver was "+" without Dirac DAP, and then remained so with an empty DAP.

 

Performed the usual room calibration measurements. Created the default filter correction. Loaded the filter.

 

With the filter loaded in the DAP, but turned off, verified each speaker cone was still "+".

 

Turned the filter I just made on while keeping everything else constant.

 

The treble left and treble right go from "+" to "-".

 

The bass left and bass right remain at "+"

 

Turned the DAP filter off again, and everything is restored to "+".

 

I'm measuring each speaker cone's polarity individually. In the case of the bass cones, I can see them go out (consistent with the "+" measurements).

 

There is no difference between the left and right channel (my early report was due to my own confusion, having switched wires around). Everything is back in the normal wiring setting.

 

The crossover frequency between the bass and the midrange is 350Hz.

[wgscott]

I should, however, add that it is entirely possible Dirac is behaving exactly as it should, and I am just finding the results surprising. The critical thing to keep in mind is that I am measuring speaker polarity right up next to each speaker cone, but what is really relevant is the interference pattern you get when you are seated in the listening position. I've spent most of my professional life thinking about phase and interference effects, and it still can be very confusing.

[Miska]

It may be result of time domain correction, trying to tidy up the step response...

 

Modifying phase and amplitude independently as function of frequency is simple operation...

 

(looking at step response with/without correction would tell more)

 

 

P.S. Change in vertical position of the measurement mic (ear level) may have fairly large effect on traditional multi-way speakers, especially when listening position is close to speakers.

[dallasjustice]

That's what I thought too. It's prolly the tweeter mid step. That's normal for drivers to be out of polarity. It could be the way the speaker is designed.

 

Btw, Acourate uses step response more than impulse for TD correction. :-)

 

 

It may be result of time domain correction, trying to tidy up the step response...

 

Modifying phase and amplitude independently as function of frequency is simple operation...

 

(looking at step response with/without correction would tell more)

 

 

P.S. Change in vertical position of the measurement mic (ear level) may have fairly large effect on traditional multi-way speakers, especially when listening position is close to speakers.

[Liam]

Yes the filters are affecting phase. I was planning to check this out as a matter of routine, having fairly recently confirmed that everything was in phase using the same tool you are using wgscott. You caused me to get on with it just now. I'm using my iPad for it here.

 

I'm using a modified filter here at present, as I felt the base was a bit light with a 'standard' reading/filter. I have bumped up the base from 100hz, as Dirac suggests might be preferable. That filter inverts both channels. I can reverse it again to + phase by inverting phase with my AMR Dp-777.

 

The 'standard' filter inverts one channel. I'm called for evening meal now but will check in afterwards.

[flak]

Hello Wgscott and Liam,

 

Dirac does not screw up phase

Dirac corrects the impulse response for each channel separately and possibly that can "look" as if some channel is inverted since the polarity is checked next to the speaker and the impulse response correction is in effect at the listening position, which may be quite asymetrical in some cases.

 

Ciao, Flavio

[wgscott]

Dirac corrects the impulse response for each channel separately and possibly that can "look" as if some channel is inverted since the polarity is checked next to the speaker and the impulse response correction is in effect at the listening position

 

v.s. ...

 

I am measuring speaker polarity right up next to each speaker cone, but what is really relevant is the interference pattern you get when you are seated in the listening position.

[Liam]

Hello Wgscott and Liam,

 

Dirac does not screw up phase

Dirac corrects the impulse response for each channel separately and possibly that can "look" as if some channel is inverted since the polarity is checked next to the speaker and the impulse response correction is in effect at the listening position, which may be quite asymetrical in some cases.

 

Ciao, Flavio

 

Hi Flavio,

 

Thanks so much for responding. So Dirac is more about correcting what is happening at the listening position than directly at the speaker. That is self evident I guess from the fact of how the mic readings are taken, but it is still nice to know that conventional readings may be disrupted by Dirac to achieve the result. Thanks for that.

 

If one makes adjustments to the initial result produced by Dirac, say for example like below, would you be surprised if I said, that whilst at both speakers the polarity showed as being inverted, that I preferred the sound when I reversed polarity in my AMR DP-777 dac so that the reading indicates the speakers are back in 'normal' polarity again. The sound seems more focused I feel. Would this surprise you or can you advance a reason why I might feel that way. In other words can adjustments made to the Dirac initially calculated result cause an unpredictable change, when altered by the user, such that when the frequency response is altered that the impulse correction gets messed up, or is it the case that everything is properly taken care of when the new filter is optimised in Dirac Calibration tool? Hope I have been clear. I will post the correct images in my next post....I messed up.

[Liam]

Flavio,

These are the correct images for the post above.

 

Original Dirac result.

 

Original in Close up.

 

My adjustments in close up.

[Liam]

"If one makes adjustments to the initial result produced by Dirac, say for example like below, would you be surprised if I said, that whilst at both speakers the polarity showed as being inverted, that I preferred the sound when I reversed polarity in my AMR DP-777 dac so that the reading indicates the speakers are back in 'normal' polarity again. The sound seems more focused I feel."

 

Mmm, quoting myself above, I have reverted back to Dirac's calculated result. On listening back and forth I have concluded I formed a false conclusion. Reversing Dirac's polarity at the speakers results in a reduction in timbre or 'richness' which mislead me into thinking focus was better, but was probably due to some information not coming across as well. I should heed my advice to myself to count to some high number before I come to conclusions about changes in my system.

[wgscott]

I also experienced the reduction in timbre, but had attributed it to the neighbors running chain saws whilst I was listening.

 

If you reverse polarity, you are changing the phase 180° arbitrarily.

[Boris75]

I also experienced the reduction in timbre, but had attributed it to the neighbors running chain saws whilst I was listening.

 

Thanks for allowing me to start the day with a good laugh.