Loopback-time align

[Wull]

jtalden, when I took those REW sweeps before I've since found my front right 15 had disconnected itself from circuit So I have run some more sweeps. Looking at the results I may have done something wrong as option 1 'all sub's and speakers in phase' looks better than 'subs or main's reversed'. Either way, both option 1 and 2 still look really good. Option 2 looking the better option.

Going forward, I would like to choose the best option. Then use REW EQ to supply filters for my miniDSP HD to flatten the LFE response. I will then run Dirac EQ. Then use REW EQ to configure a house curve if this is possible?

 

[jtalden]

Those don't look right at a quick glance. One or more of the phase traces are all not correct. I defined 3 options. I am not sure why you provided 4 measurements.
All the changes I provided were based relative to the initial timing settings applied in Post 33. Is this what you did?
[It would have been better if I had ask for those actual settings. that way I could provide actual delay settings instead of changes.]
If there is any question as to the correct delay settings, I suggest you set delays for Option 1 and measure L, R, FIB, RIB, F15, R15. I will rerun the analysis on that data. If you also provide the actual delay settings you used I will confirm/correct the values and also provide the settings for the other 2 options.

 

[Wull]

I do apologize for any confusion on my part.


Attached both options 1 & 2 set up as per post 46. Using a 120 Hz XO.

 

[jtalden]

I had intended to invert R-IB an to add 6ms of delay to it for both Options 1 and 2. This provides close timing of the 4 subs. The data in post 53 did not reflect that.

There also appears to be a small difference in the 4sub group settings as well. I therefore started fresh. The reference settings are those settings applied in Post 53 option 1.

Description:

Option 1, provides the closest phase tracking of the direct sound. It is the option that I would normally suggest.

Option 2, has the 4sub group arriving ½ wavelength earlier and crosses the phase near the XO frequency. This option appears to provide slightly better phase tracking considering the room impact on phase at the LP.

Both options provide similar SPL support in the XO range.

Below is a summary of 2 options:

 

[Wull]

Option 1. I added 3 ms to the main speakers. And added 6 ms to the R-IB and inverted it's polarity.

Option 2. I added 7 ms to the main speakers. And added 6 ms to the R-IB and inverted it's polarity.

 

[Wull]

I'm really not sure what to make of those last results. I have 'reset' put all delays back to zero, and re run REW sweeps on all subs and L&R to see if there is any drift going on (attached).

The previous option number 2 seemed to give good results. My PC I use to run REW is in my office. My office has a door that leads into my movie room. When I ran REW Option 2 last night it made my internal door wobble. Good sign.. So I will see how this sounds over the weekend and report back.


Thanks for all your help jtalden.

 

[jtalden]

Post 55 new option 1..mdat has the SWs reverse polarity and different mains to SWs delays compared to my Option 1.

Post 55 new option 2.mdat has the SWs polarity correct, but the has different mains to SWs delays compared to my Option 2.

Below is showing the step response of the your new option 1 vs L, R traces of the intended Step response. This is how I know there is still a problem. Similarly, the step response of new option 2.mdat was used to confirmed the wrong timing.

​

I will use Post 56 new subs L and R.mdat to make a table of delay settings and polarity settings for both option 1 and 2. You indicated a reset and delays set at 0 ms in that file. Presumably all polarities were also set positive and all phase controls were set to 0°.

 

[jtalden]

Here are the results:

 

[Wull]

Carried out sweeps as per post 58. Option 1 I did an extra sweep. Details in REW comments box.

 

[jtalden]

I reconfirmed Opt-1 values of Post 58 by starting with Post 56 mdat. If the setup is the same and the new delays and polarities are entered:

> The resulting overlay chart of the impulse positioning of the 6 driver measurements will look as shown below.


> The resulting overlay of the step response of the L + 4subs and R + 4subs will look as shown below.


This is a good quick check of your setup and measurement.

Since you mentioned that you intend to apply Dirac., I suggest Opt-1 is the better setup to use. That setting provides accurate timing for the direct sound. I'm thinking that is the probably the better starting point for Dirac.

 

[Wull]

Thanks jtalden.

I hope to run Dirac tomorrow. My version of Dirac is limited to EQ only. Features such as 'Balance all audio channels to same level & Calculate and enable channel delays' have been disabled. If I get time I will run Dirac on both options and report back.

Option 1. With all but the R-IB having Negative phases. I guess I should set the Surround and Back speakers (7.1 system) with a Negative phase too?

 

[jtalden]

Have you tried taking several measurements of the same speaker and confirming the impulse is located at the same position for each measurement? Zoom in on the time scale and the impulses should all fall on top of each other within maybe 0.01 ms.

Post 58 measurements still do not match the target for polarity and timing. All or most all of the polarities are the reverse of mine. If I invert your impulse it looks more like mine except the 4subs timing is arriving about 4 ms too early. Below is the overlay of your L, R measurement with my L target. I inverted my L target to and aligned the impulse peaks at 0 ms for clarity. [There is not enough info in your measurement to know if the 4subs are all timed correctly. There is a hint that they may not be.]

 

[Wull]

Attached - Right speaker 5 x sweeps

 

[jtalden]

I just saw you last post. That is what is needed. I will do the comparison.

 

[jtalden]

Post 63 file not attached?

 

[jtalden]

Looks good - no problems with repeatability. Sweeps 2-6 (R channel) are identical.
Sweep 1 (L channel) timing is only off by 0.01 ms. That means the mic was very well centered.

 

[Wull]

OK. So is there any more information you need from me or am I OK to proceed to Dirac Option 1 ?

 

[jtalden]

If you want to sort out the timing and polarities, attached is a file with the impulse timing of all 5 driver in the left channel. If you set the Opt-1 timing all the impulses your measurements should fall on top of the ones in this mdat.
You can load this mdat file and measure L, F-IB, R-IB, L15, R15. The overlay charts of each of drivers should be on top of each other. Adjust timing or each as needed until they do. Note that there may be a small offset/shift to all of them as I may not have guessed exactly the 0 ms point you Ref signal will. It should be close however.

 

[Wull]

OK. Having run those sweeps again all subs and speakers where out of phase. Using Option 1, where you said 'Negative' I 'inverted' the Phase, and Positive 'Invert'. I have reversed all these changes to match your 'impulse'

attached mdat

 

[jtalden]

Yes, polarities and timing all now match perfectly - ready for EQ.

 

[Wull]

Perfect. I will run Dirac tomorrow and report back.

Thanks for being patient jtalden

 

[jtalden]

Oops, I compared to my targets by mistake.
The polarities are indeed correct.
The timing is still wrong for some reason. It is probably not a major problem as with 4 subs however.

I see in my target timing file I accidentally forgot to deleted the last 2 measurements that are not used for this purpose. If you decide to adjust the timing just delete the last 2 measurements.

The offset of each sub from your L impulse should equal the offset of each sub from my L.

 

[Wull]

Sorry. I noticed just before that I hadn't imputed the delays in the miniDSP correctly for the 'back IB and L&R 15's. Attached a corrected mdat

 

[Wull]

option 1 & 2 Dirac results.

I've also attached the Dirac before & after plot. Seem to be loosing a lot of output 25Hz-40Hz?

 

[jtalden]

Sorry. I noticed just before that I hadn't imputed the delays in the miniDSP correctly for the 'back IB and L&R 15's. Attached a corrected mdat

The polarity and timing is now identical to the target for Opt-1.