Home Theater Forum and Systems banner

1 - 20 of 64 Posts

·
Registered
Joined
·
91 Posts
Discussion Starter #1 (Edited)
I've been trying to use a timing reference, but keep getting a frequency response plot that looks altered. I initially had lots more interference until I disconnected my amp from the XLR connection, but after figuring that part out, the freq plot looks decent, but is skewed only when using a timing reference loop back.

Timing Reference Output:
HDMI 5 (XMC-1 via ASIO4ALL); HDMI 5 = left rear speaker of my 5.2 system.

Timing Reference Input:
Presonus AudioBox Input 2 from XMC-1 RCA preout for left rear speaker (disconnected amp for left rear due to some interference I was getting initially).

ECM8000 non-calibrated mic on Input 1 of Presonus AudioBox

I plot the attached when doing this measurement, but what is strange is low end is elevated when using the loopback!?! I checked "set loopback as timing reference" under analysis and set the timing references as noted above.

I did all this because the normal method I was using (UMIK-1 calibrated USB mic and ASIO4ALL HDMI out), was showing that my subs were not time aligned to my speakers. The excess group delay plot showed about a 12 ms difference! I am using the Emotiva XMC-1 Dirac full version, and a lot users report time aligned issues with subs. This loop back method is confirming my subs (HDMI 4) are out of alignment from my left speaker (HDMI 1). Getting a 10.4 ms difference.

If you are curious, here are some measurements and system details (UMIK-1 calibrated USB mic and ASIO4ALL HDMI out for these files):
https://drive.google.com/folderview?id=0B1J0a4OV_WGLS2lYYl85N1BRNkE&usp=drive_web

Pics:
https://drive.google.com/folderview?id=0B1J0a4OV_WGLREI3eDVXd0xBM2s&usp=drive_web

BONUS QUESTION:
How to fix the sub delay and still use Dirac on the XMC-1 since I cannot adjust distance settings!?! I thought about buying a miniDSP and adding a 50 millisecond delay to the subs, then rerunning Dirac, then doing the loop back reference measurement again and then lowering the 50 millisecond delay.

I've also thought about bumping up the phase on each of my two subs (they are matched and wired in a mono configuration to a single sub output), rerunning Dirac, then decreasing the phase knob on each sub after doing the loop back measurement.
 

Attachments

·
Premium Member
Joined
·
1,562 Posts
Possibly you are expecting that ideally the peak of the SW IR should be closely aligned with the Peak of the Mains IR? That is a common misconception.

My understanding that it is ideal to have the phase of the SW and Mains track closely with each other throughout the XO range, maybe 50-150 Hz. I think this is the common ideal target that results in the closest working relationship of the drivers, I would expect that this is the target for a DL setup. In that case the peak of the SW IR will lag the peak of the Mains by several ms.

There are other commonly used alignments that provide almost equally smooth SPL reinforcement. One of those provides much closer alignment of the 2 IR peaks. There is little likelihood that that any of the good alignment will sound significantly better, but many can apparently hear a noticeable difference and thus may prefer one to another.

From the charts above, the XO SW timing alignment looks very good. There is no suggestion that there is a error in the timing. It is likely that the Dirac setup properly did it's job and I would not be surprised if it was ideally timed. That said, It is impossible to tell for sure with the charts above and it is always comforting to confirm it.

I can do this for you given just 4 simple measurements using either mic and no loopback is needed. This method is a difficult to easily explain so I may not be able to clearly explain the process to you. The advantage is that it is the easiest for you measure and get quick confirmation that the timing is correct or what adjustment is needed to do so.

You may well prefer to work through the process using the loopback setup and I can help with that also. It may be more of a process, but it is a more conventional one and may be easier to understand. The results will be the same.

Below is my short thoughts to your questions in case they are helpful for your process.

I've been trying to use a timing reference, but keep getting a frequency response plot that looks altered. I initially had lots more interference until I disconnected my amp from the XLR connection, but after figuring that part out, the freq plot looks decent, but is skewed only when using a timing reference loop back.
See below.

Timing Reference Output:
HDMI 5 (XMC-1 via ASIO4ALL); HDMI 5 = left rear speaker of my 5.2 system.

Timing Reference Input:
Presonus AudioBox Input 2 from XMC-1 RCA preout for left rear speaker (disconnected amp for left rear due to some interference I was getting initially).

ECM8000 non-calibrated mic on Input 1 of Presonus AudioBox
I think that is a good loopback setup.

I plot the attached when doing this measurement, but what is strange is low end is elevated when using the loopback!?! I checked "set loopback as timing reference" under analysis and set the timing references as noted above.
Not sure, but suspect the problem here is that the XMC-1 mode may be the issue. If using surround mode, mono mode, etc the redirected SW output will be adjusted for the several active main channels resulting in higher level for the SWs than for the single measured main. This is no problem for this type of timing analysis. The amplitude does not impact the phase.

BONUS QUESTION:
There is a very good chance that all is well so let's hold on this.
 

·
Premium Member
Joined
·
1,562 Posts
I should have mentioned that to confirm timing via the phase tracking method:

It is necessary to measure FL, SWs, FL+SWs using only channel #1. You should not measure using the LFE, channel #4. That way the XO filters are applied properly for FL channel and the timing and levels will also be correct. The right channel (#2) is then done similarly and so forth for any other channels you want to confirm.
 

·
Registered
Joined
·
91 Posts
Discussion Starter #4 (Edited)
Possibly you are expecting that ideally the peak of the SW IR should be closely aligned with the Peak of the Mains IR? That is a common misconception.
Yes I was!

My understanding that it is ideal to have the phase of the SW and Mains track closely with each other throughout the XO range, maybe 50-150 Hz. I think this is the common ideal target that results in the closest working relationship of the drivers, I would expect that this is the target for a DL setup. In that case the peak of the SW IR will lag the peak of the Mains by several ms.

There are other commonly used alignments that provide almost equally smooth SPL reinforcement. One of those provides much closer alignment of the 2 IR peaks. There is little likelihood that that any of the good alignment will sound significantly better, but many can apparently hear a noticeable difference and thus may prefer one to another.

From the charts above, the XO SW timing alignment looks very good. There is no suggestion that there is a error in the timing. It is likely that the Dirac setup properly did it's job and I would not be surprised if it was ideally timed. That said, It is impossible to tell for sure with the charts above and it is always comforting to confirm it.

I can do this for you given just 4 simple measurements using either mic and no loopback is needed. This method is a difficult to easily explain so I may not be able to clearly explain the process to you. The advantage is that it is the easiest for you measure and get quick confirmation that the timing is correct or what adjustment is needed to do so.

You may well prefer to work through the process using the loopback setup and I can help with that also. It may be more of a process, but it is a more conventional one and may be easier to understand. The results will be the same.
I would like to try using the loopback cable, but I don't mind seeing which plots you would look at for the no loopback setup as I already have a REW for the no loopback measurements here: https://drive.google.com/folderview?id=0B1J0a4OV_WGLS2lYYl85N1BRNkE&usp=drive_web

Not sure, but suspect the problem here is that the XMC-1 mode may be the issue. If using stereo mode, mono mode, etc the redirected SW output will be adjusted for the several active main channels resulting in higher level for the SWs than for the single measured main. This is no problem for this type of timing analysis. The amplitude does not impact the phase.
The plots in my original post were just of "HDMI 1" where HDMI 1 is crossed to the subs at 80Hz. HDMI 1 corresponds to the front left channel.


There is a very good chance that all is well so let's hold on this.
This is great news! I was only concerned as someone on the AVS forum suggested an issue wrt to sub timing after looking at my excess group delay plot. I read over the REW manual, and it also suggests using EGD plots to check time alignment, so then I became worried that Dirac was doing something wrong.

On my google drive, there is an REW file I made using the UMIK usb mic (no loop back measurement using ASIO HDMI outputs), and it has measurements like this for the left speaker:




These measurements look very good, and as you suggest seem to indicate Dirac is picking a sub timing that results in the smoothest possible sub transition wrt to amplitude. In the XMC-1 the 80Hz crossover is limited to 12dB slope for the speakers, but you can select a 24dB or a 12dB slope for the sub. Currently I have a 24dB slope for the subs, which luckily gave the pretty results above on my google drive. I did try changing this to 12dB slope for the sub low pass filter and inverting the phase on the sub. It gets rid of one phase shift in the plot, but causes dips in the amplitude, although the excess group delay plot shows a much lower timing difference. I put it back like I had it originally for now as you suggest, and will hold off on randomly changing things.
 

·
Registered
Joined
·
91 Posts
Discussion Starter #6 (Edited)
Here is the file for the left channel. Contains nine measurements, all use Dirac. HDMI 1 channel only for all (had to turn off subs or amp to get different measurements). Subs are share common signal from XMC-1 Sub1 output.

For this file, I used the calibrated UMIK-1 USB mic and ASIO4ALL HDMI 1 channel out.

File size is 21MB. File is here: https://drive.google.com/open?id=0B1J0a4OV_WGLdXZ5UzkzN1RuR3M

Or you can view all measurement files here: https://drive.google.com/open?id=0B1J0a4OV_WGLS2lYYl85N1BRNkE

In chronological order:
The first three are with -24dB/octave low pass, no phase adjustments to the subs.
Second set of three use -12dB/octave low pass, no phase adjustments to the subs.
Last three measurements use -12dB/octave low pass, 180 degree phase switch switched for both subs.

EDIT: Thought it might be worth while to add a tenth measurement of HDMI 1 with no crossover, just the loudspeaker. New version of file is on google drive at the old link.

I should have mentioned that to confirm timing via the phase tracking method:

It is necessary to measure FL, SWs, FL+SWs using only channel #1. You should not measure using the LFE, channel #4. That way the XO filters are applied properly for FL channel and the timing and levels will also be correct. The right channel (#2) is then done similarly and so forth for any other channels you want to confirm.
 

·
Premium Member
Joined
·
1,562 Posts
Here is the file for the left channel. Contains nine measurements, all use Dirac. HDMI 1 channel only for all (had to turn off subs or amp to get different measurements). Subs are share common signal from XMC-1 Sub1 output.
This was the correct process for this particular analysis.

For this file, I used the calibrated UMIK-1 USB mic and ASIO4ALL HDMI 1 channel out.
Good, this will work.

Or you can view all measurement files here:
Since this is an analysis of the of the REW files and the measurements must be manipulated the charts are no help. The file you linked includes what is needed.

In chronological order:
The first three are with -24dB/octave low pass, no phase adjustments to the subs.
Second set of three use -12dB/octave low pass, no phase adjustments to the subs.
Last three measurements use -12dB/octave low pass, 180 degree phase switch switched for both subs.
This needs clarification. We need to analyze the condition that was used for the DL setup. Was that the first 3 measurements?

Moving an LFE filter to different settings should not impact the results, but it appears it did at first glance. So, maybe this pre-pro really change the SW to mains XO filter slopes as you suggested? That is not an option in the 3 pre-pros I have experience with. I can understand if this is indeed an option. I have just not run across it before. We just need to decide the setting to use and then run the DL setup and confirm the timing.

I will analyze the first 3 measurements on the left channel in case they were done with the correct settings. If you advise this is not the preferred settings then we can try again.
 

·
Registered
Joined
·
91 Posts
Discussion Starter #8 (Edited)
Thanks for your help!

The Emotiva XMC-1 is a strange beast.

According to Emotiva's engineers (links below), the crossover slope for the sub (e.g. low pass) is user changeable from -12dB/octave or -24dB/octave. The high pass slope is always set to -12dB/octave no matter what (well unless it is disabled of course). Also, Dirac does not use the crossover when EQ'ing. The crossover is set AFTER Dirac (I know this is backwards from every processor I've owned, but it is what Emotiva's engineers state in the thread below).

http://emotivalounge.proboards.com/post/728764/thread
http://emotivalounge.proboards.com/post/728893/thread

EDIT:
And some other really good points from KeithL on how they designed Dirac and bass management:
http://emotivalounge.proboards.com/post/744469/thread

"Just to clarify something - Dirac doesn't "handle" crossovers at all.

Dirac Live treats each channel as if it's full range, and corrects it as such (although it may not apply corrections above or below the frequencies which it considers to be the functional limits of the speaker, it doesn't block any frequencies form reaching the speaker). Dirac DOES NOT KNOW OR CARE what the crossovers in the XMC-1 are set to. The crossovers in the XMC-1 then act independently of Dirac Live to do bass management.

There also seems to be some serious misunderstanding about what a crossover does (in the quoted post).

You CANNOT "tell the XMC-1 not to play the test tone over the front speakers"; what you can do is to instruct the XMC-1 to apply an electrical filter that will alter the level of those test tones based on frequency.

For example, you can set the XMC-1 to apply a 24 dB/octave high-pass filter at 100 Hz.
If you do this, then the XMC-1 will leave all frequencies well above 100 Hz alone, apply a cut based on frequency that works out to -3 dB 100 Hz, -27 dB cut at 50 Hz, and approximately -50 dB at 25 Hz.
This is how a 24 dB/octave filter is SUPPOSED to work, and how they all do.

However, there's nothing to prevent the filters created by Dirac Live from applying additional boost or cut at some frequencies.
Likewise, if you're using the manual PEQs in the Speaker Presets instead, there's nothing to prevent them from making additional changes to the frequency response.
And, finally, these changes are applied to the frequency response of the speaker as it already is in your room.

As an extreme example, let's assume that your room and speaker (as a combination) have a nasty resonance that results in a 27 dB peak at 50 Hz.
If you add the response of our 100 Hz 24 dB/octave filter in the crossover to that in-room frequency response, the net result will be that the response will be 0 dB at 50 Hz (the peak in the room will cancel out the action of the crossover at 50 Hz).
This is, again, the way it is supposed to work.

If you want the crossover to cut in more suddenly, and do a more thorough job of blocking sound below its cutoff frequency, then you need a SHARPER FILTER.
Unfortunately, the sharper a filter you use, the more processing power it requires, and the more likely it is to produce audible unpleasant side effects.
For this reason, very few vendors offer filters sharper than the 24 dB/octave ones we use."

Moving an LFE filter to different settings should not impact the results, but it appears it did at first glance. So, maybe this pre-pro really change the SW to mains XO filter slopes as you suggested? That is not an option in the 3 pre-pros I have experience with. I can understand if this is indeed an option. I have just not run across it before. We just need to decide the setting to use and then run the DL setup and confirm the timing.

I will analyze the first 3 measurements on the left channel in case they were done with the correct settings. If you advise this is not the preferred settings then we can try again.
 

·
Premium Member
Joined
·
1,562 Posts
I reviewed the 3 sets of timings:
The first set (3 measurements) is a good alignment as seen by good SPL support in the XO range. The phase timing is good with a crossing phase situation near the XO frequency. From a sound quality perspective it is fine to use this setting and be assured that this is a good alignment.

The second set is poor as easily seen by the SPL sag through the XO range.

The third set is also good as is easily seen by good SPL support in the XO range. In this case the phase does not cross, but due to a slight timing error they track a little separated.

If we want to nitpick there is a more ideal setting in terms of optimizing phase tracking if that is our objective. The sound difference is not likely to be significantly better than the first or third series however. There is a little more SPL support through the XO if that is desired, but that may not be needed to achieve your target house curve.

The optimized timing is based on setup 3 (the third series) but with the SW delay reduced by 2ms. That is, the SWs polarity should be set to -180° and their delay reduced by 2ms. This may not be possible as you indicated the SWs distance setting cannot be adjusted after the DL setup is conducted? If the distance can be changed just increase the SWs distance setting by 2 feet. If not then just use the first or third settings as they are considered good by providing a reasonably flat SPL response. If it meets your desired house curve then either one should work great.

Selected charts below to illustrate:

SPL Support; Series 1,2,3 vs Optimized
SPL.png

Phase Tracking; Series 3 Current
Phase Series 3 Current.png

Phase Tracking; Series 3 Optimized
Phase Series 3 Optimized.png
 

·
Registered
Joined
·
91 Posts
Discussion Starter #10 (Edited)
Thanks for taking a look! How did you do the optimized timing plot? Is this an iterative process where you modify the subwoofer plot so that phase better matches the phase plot for the left speaker with sub turned off (e.g. reduce by 5 ms, then 4 ms, etc...) ?

Do you think I can trick Dirac by adding a miniDSP with an extra delay (maybe 30 milliseconds), and then that way I have 30 milliseconds to remove if needed to better optimize?

I'm a little confused on why 180 degrees setting on the subs works so well to be honest! In theory, this can't subtract from the DSP delay in the sub right? I do have phase knobs also on my subs, but was under the impression that these will never reduce time delay, but can add to the delay.

EDIT: I really think #3 sounds a tad more natural than before. Can't tell if this is a placebo effect though. Of course #1 was the best system I've ever heard and I lived with it for many months now which makes it really hard to beat.

The optimized timing is based on setup 3 (the third series) but with the SW delay reduced by 2ms. That is, the SWs polarity should be set to -180° and their delay reduced by 2ms. This may not be possible as you indicated the SWs distance setting cannot be adjusted after the DL setup is conducted?
 

·
Premium Member
Joined
·
1,562 Posts
How did you do the optimized timing plot? Is this an iterative process where you modify the subwoofer plot so that phase better matches the phase plot for the left speaker with sub turned off (e.g. reduce by 5 ms, then 4 ms, etc...) ?
Basically yes.
> Find the 'Current real' SW IR position relative to the FL IR position. Since loopback was not used that is not originally known. I have method to do that accurately using the step response chart, but is difficult to explain the process. Now that 2 forms of loopback are available in REW this complicated method is not needed. Just create measurements using loopback.
> Offset (shift) the SW IR (as you suspected) to find the best phase tracking. That may require inverting the SWs polarity to get the best overlay as it did in this case. It just depends on the XO filters being applied and the driver responses. There is no 'correct' SW polarity. It is just the one that provides the best phase tracking.
> Use 'Trace Arithmatic' FL + Shifted SWs to see the resulting response of the new alignment.

Do you think I can trick Dirac by adding a miniDSP with an extra delay (maybe 30 milliseconds), and then that way I have 30 milliseconds to remove if needed to better optimize?
I am suggesting that settings 1 or 3 is just fine. A 2ms (2 foot) shift from ideal for settings 3 is not likely to make a practical difference. I would not complicate the system with additional boxes. I would think there is a way to change the delay/distance in the current setup, but if not, then just use it as is. Adding a box to the SW output would allow additional delay, but you need less SW delay so you really would need to increase the delay in all the main channels instead - sounds complicated.

I'm a little confused on why 180 degrees setting on the subs works so well to be honest! In theory, this can't subtract from the DSP delay in the sub right? I do have phase knobs also on my subs, but was under the impression that these will never reduce time delay, but can add to the delay.
You are correct.
The 180º polarity setting just flips the polarity without changing the timing/delay. A variable phase control on a SW just increase the delay around the XO freq and that may help in some cases. It will probably be counterproductive for settings 3 as you need reduced delay. Settings #1 is pretty close to the correct timing for that crossing phase style of timing. I don't think the phase control will help much there, but if I recall the situation correctly, I think it is directionally correct. The phase tracking chart may then look marginally better, but I would expect no noticeable change to the sound quality. It would take a lot of work to find out for sure.

EDIT: I really think #3 sounds a tad more natural than before. Can't tell if this is a placebo effect though. Of course #1 was the best system I've ever heard and I lived with it for many months now which makes it really hard to beat.
:smile:

Either is very good technically. You may have preference to which sounds a little better to you so use the one you prefer.
 

·
Premium Member
Joined
·
1,562 Posts
This is the excess group delay measurement Markus on the AVSforum showed me. Pretty interesting to compare trial 1, 3 and no crossover at all (front left only):
Yes, the excess delay is increased a little with settings #1 vs #3. Excess delay is another (less sensitive) way to try to determine the if there is an issue with timing. The phase tracking chart is much more helpful for me. The excess phase chart does not identify that settings #1 is a good setup although that can be inferred from the SPL chart without looking at phase.
 

·
Registered
Joined
·
91 Posts
Discussion Starter #14
The idea is to add a box to give extra delay, run dirac, then have delay "padding" I can reduce to get the plot the way I want (versus right now where I can't reduce delay at all). Hopefully this would trick Dirac into doing what I want.

There really isn't a way to modify Dirac's delay settings, that's a big drawback to the XMC-1. Also, Dirac was added to the XMC-1 after it came out, and there is some limitation on the DSP they used, so Emotiva can never modify the XMC-1 to apply the crossover before Dirac. Dirac on the XMC-1 is fantastic for the price though. It blows away my Marantz AV8801 and I hear no hiss at my MLP.

Still, looking at how effective Dirac is, makes me really want to do the crossover outside the XMC-1 so Dirac would be applied to fix the entire crossover region (by setting the XMC-1 main speakers to large). I could even do the filter using the same miniDSP box being used to "pad" the delay. Not sure if it's worth the hassle though. My setup sounds truly fantastic the way it is, and it's not worth adding a second box just to fix that last phase shift.

If I could live with the extra 1% distortion from my main speakers below 80Hz, I'd do away with the crossover altogether which had the best looking phase plot, but I swear I hear inter-modulation distortion, especially on tracks with bass strings.

Thanks for your help. I learned a neat trick from you, and I greatly appreciate it!

I would think there is a way to change the delay/distance in the current setup, but if not, then just use it as is. Adding a box to the SW output would allow additional delay, but you need less SW delay so you really would need to increase the delay in all the main channels instead - sounds complicated.
 

·
Premium Member
Joined
·
1,562 Posts
I am surprised you can't change the speaker distances either before or after the DL setup. That sounds like a real limitation to me. Fortunately for you, there are 2 settings that provide good XO timing results even without that capability. Possibly the owners can lobby Emotiva to update the firmware to provide that option. I suspect that has probably been tried, but I had to say it anyway. It represent a significant oversight on their part in my opinion.

I would not expect the distortion to be impacted by either setting. Are you sure there were not other factors involved impacting the distortion results? All sorts of things impact the distortion level more than 1% from run to run in the low frequencies so it is easy for these effects to creep into the results. A minor phase shift between drivers is not a major factor so far as I have experienced.

I just looked at your distortion results. They look very good. Even the relatively poor series 2 setting only measure 2.4% at 68Hz where there is a small peak. The small improvement in series 1, 3 is more related the small increase in SPL at that frequency rather than a reduction of the measured distortion level. You may want to measure the noise floor level just before and after each distortion measurement and see if there is a peak at that freq range. If so then that is more the issue than the timing is. Note also that there are several other minor distortion peaks at other higher frequencies as well. These also may be related to noise floor limitations. Just thoughts FYI...
 

·
Registered
Joined
·
91 Posts
Discussion Starter #16
I've never measured the noise floor of my room, this is a great idea! In the file named "NOISE FLOOR TEST FRONT LEFT UMIK-1" there are some distortion measurements with several trials. I used the UMIK-1 USB mic and HDMI 1 channel. I turned the XMC-1 volume all the way down, then placed it on mute.

The first 3-6 chronologically are how the room would be during normal listening. For 7-9 I unplugged everything that was feasible to unplug while leaving on the amps, subs and XMC-1. Also did some RTA captures after doing sweeps 7-9.
https://drive.google.com/drive/folders/0B1J0a4OV_WGLS2lYYl85N1BRNkE

Seems there are some noise spurs in the mid-range region. Could be my UMIK-1 USB mic or could just be a little pink noise from my amps that I left on. I'll have to run a similar test next week when my calibrated EMM-6 comes from Cross-Spectrum. I'll also remember to do a test with the amps off (duh).

The distortion difference is when using only the front left speaker which will go below 20Hz on its own (e.g. with no crossover). This low frequency distortion is a little higher when not using the two subs, presumably since the subs are of very good quality and have larger drivers, and there is a better modal coupling of dual subs to the room versus just the front left speaker. However, the front left speaker by itself is very impressive. This plot of just the front left was plot 10 in the CROSSOVER TEST file.

I would not expect the distortion to be impacted by either setting. Are you sure there were not other factors involved impacting the distortion results? All sorts of things impact the distortion level more than 1% from run to run in the low frequencies so it is easy for these effects to creep into the results. A minor phase shift between drivers is not a major factor so far as I have experienced.

I just looked at your distortion results. They look very good. Even the relatively poor series 2 setting only measure 2.4% at 68Hz where there is a small peak. The small improvement in series 1, 3 is more related the small increase in SPL at that frequency rather than a reduction of the measured distortion level. You may want to measure the noise floor level just before and after each distortion measurement and see if there is a peak at that freq range. If so then that is more the issue than the timing is. Note also that there are several other minor distortion peaks at other higher frequencies as well. These also may be related to noise floor limitations. Just thoughts FYI...
 

·
Registered
Joined
·
91 Posts
Discussion Starter #17 (Edited)
I hope you are doing well.

I ended up buying the Dirac software and now do all processing on my PC (movies and music are on my server). This lets me use the distance settings in the XMC-1 to fine tune the subs.

To get rid of the distortion, I found that tri-amping and removing the passive crossovers helps a bunch (and sounds better too). I added two Ashly XR1001's.

However, now I am revisiting sub delay setting (since passive crossovers being removed made my EGD plot less than 2ms from 20-20000Hz for just the left channel and no sub and now the sub doesn't match). I made a spreadsheet that calculates a distance that yields an expected degree for the subs.

For 80Hz, this gives: C7 = degree difference = main left degree at xo - dual subs degree at xo.

(1/80Hz)*(C7/360)*(13503.9 inches/s) = expected distance needed to change distance setting in XMC-1.

I think I duplicated your method:
0. Turn off all processing/EQ.
1. Set crossover in AVP/AVR.
2. Measure left speaker with subs off (acoustic timing ref is right speaker)
3. Measure subs with left speaker amp(s) off (acoustic timing ref is right speaker)
4. Note degrees crossover frequency from step 2 and 3. Also set level of sub to be similar at crossover frequency to freq resp plot of step 2.
5. Calculate the needed distance to move the sub distance setting in AVP/AVR.
6. Set new distance = old value + calculated value.
7. Measure things and see if it works.

This seems to work well, but the impulse is off from 0 by about 9ms and the EGD plot shows a 12ms timing difference below 60Hz which makes me think I need to go 360 degrees from 4' 7 feet , but then the phase slope is opposite of the left speaker at ~18' 7" (4' 7" + 360 deg @80Hz).

I'm tempted to use 18 feet 7 inches even though this shows an impulse that is -2ms versus +9ms for 4 feet 7 inches. It sucks the phases increase and decrease in opposite directions at 80hz though? I know though that the impulse plot is not reliable for setting subs from the REW manual...

EDIT: I think the smart thing to do is since when I plot both left and right with no crossover, phase is not linear and smooth at 80Hz (presumably due to the room). Would be smarter to just use a crossover point where the sub and both front speakers have linear and smooth phase.


I am surprised you can't change the speaker distances either before or after the DL setup. That sounds like a real limitation to me. Fortunately for you, there are 2 settings that provide good XO timing results even without that capability. Possibly the owners can lobby Emotiva to update the firmware to provide that option. I suspect that has probably been tried, but I had to say it anyway. It represent a significant oversight on their part in my opinion.

I would not expect the distortion to be impacted by either setting. Are you sure there were not other factors involved impacting the distortion results? All sorts of things impact the distortion level more than 1% from run to run in the low frequencies so it is easy for these effects to creep into the results. A minor phase shift between drivers is not a major factor so far as I have experienced.

I just looked at your distortion results. They look very good. Even the relatively poor series 2 setting only measure 2.4% at 68Hz where there is a small peak. The small improvement in series 1, 3 is more related the small increase in SPL at that frequency rather than a reduction of the measured distortion level. You may want to measure the noise floor level just before and after each distortion measurement and see if there is a peak at that freq range. If so then that is more the issue than the timing is. Note also that there are several other minor distortion peaks at other higher frequencies as well. These also may be related to noise floor limitations. Just thoughts FYI...
 

·
Premium Member
Joined
·
1,562 Posts
If I follow correctly:
> The 3-way mains are now tri-amped using the Ashly boxes for the XO's. All is found well with this this change.
> The XMC-1 is providing an 80 Hz XO (mains to SW's). The unit can now also provide delay timing, i.e., speaker distance settings for the mains and SW's.
> A miniDSP Dirac box follows the XMC-1 and will provide EQ for the system. [This is an optional place for timing adjustments.]

You are looking now to set/confirm the proper XMC-1 distance settings prior to running the Dirac EQ.

This sounds like a reasonable approach to me.

I can help in confirming the distance settings. If you need any Dirac EQ help, others will need to provide that assistance.

So, concerning distance settings:
I don't really follow the logic for distance settings in Post 17. I'm also not comfortable trying to use EGD as a measure for that purpose. It may work just fine, but I no experience with that method so I have no good feel as to the its accuracy or repeatability. The phase tracking approach I use is likely to be a little more sensitive and thus at least as reliable however. So, If you want me to check your work I can do that. I will provide some charts with a brief explanation. If a distance adjustment is advised the needed change will be detailed.

Minimizing the needed measurements for a full analysis, I would need:
Setup
> Mic Centered at the LP
> Record and provide me the distance settings in the XMC-1 for these measurements.
> Set the XMC-1 desired 80Hz XO on using the chosen slope.
> MiniDSP Dirac box on, but no EQ, delays, or other active settings, i.e., set as a bypass mode.
> Engage either loopback timing or acoustic timing in REW (depends on USB or XLR mic type).
Measure
> SW's only
> FL main only
> FR main only
> CC main only (if desired)

From that mdat file of 4 measurements I will be able to calculate the SPL and phase responses of the 3 channels (SW's + FL, SW's + FR, and SW's + CC) using REW 'Trace Arithmetic' functions. I will also be able to shift the timing if needed to confirm the delay setting needed for optimal phase tracking.

Other Comments:
Phase and GD are related so either is usable for timing adjustments. EGD provides a cleaner trace than GD so that is why it works better. I assume you are just looking for a smooth transition from mains to SW's in the EGD trace. That should work pretty well given a relatively smooth trace. If there are room modes in the XO range that can make it more difficult. Using phase instead, the 'frequency dependent window' controls in REW help a lot in smoothing the phase curve so adjusting the timing is probably easier in that situation.
Using that method requires IR shifting and REW trace settings, so it is not quickly explained. Some idea of the process (a bit outdated now) is shown here.
 

·
Registered
Joined
·
91 Posts
Discussion Starter #19 (Edited)
EDIT: I'm thinking about a Yamaha SP2060 so as to align the mid-range and high range horns to the woofers at the MLP. I'm thinking I can still use REW to find the PEQ settings and time alignment, and that I'd just copy these over. If you'd like, please take a look at my tweeter and midrange in the REW below. Is time alignment worth exploring for the MF and HF drivers? I have a dip at 5kHz that was not present before. Might be 5" or so difference at the front of the speaker.

I will be using the PC version of the Dirac software. It shows up as a sound device to windows, but is really a virtual soundcard that binds to an actual soundcard output: http://www.dirac.com/online-store/

So signal chain is PC->XMC-1->Ashly XR1001 analog active xo->3 amps.

Might order a Yamaha SP2060, so PC->XMC-1->Yamaha SP2060 digital active xo-> 3 amps.


I ended up loading PEQ filters into the XMC-1 since REW lets me export them for the XMC-1. This was used to flatten out the curves (Ashley XR1001 is an analog active crossover). When I did this, I'm pretty sure my THD went up, but I'm assuming IMD will not be impacted.

Here is the link to the file. First measurement description has current XMC-1 distances I'm using (via the method I attempted to describe, which appears to be the same as you were doing in your other post except instead of inverting polarity, I attempt to offset phase at the crossover point by using the distance equation I posted previously).

This first measurement was my stab at the settings after many attempts, but the EGD plot is not as pretty as before when you did an awesome job last year.

I think adding the shelves and amps up front and removing the cinder blocks that helf up my center speaker has impacted room response. There is now a room mode at 70Hz for the subs I'm guessing.

File: https://drive.google.com/open?id=0B1J0a4OV_WGLUFdCZzJ6cXFsOG8

It will be interesting to see if you can come up with a better timing as I tried to learn from your posts from before and spent hours on this so far. I'm sure you will find something better though as this if my first time using these features in REW.

Thanks again for your other posts and wisdom.

If I follow correctly:
> A miniDSP Dirac box follows the XMC-1 and will provide EQ for the system. [This is an optional place for timing adjustments.]

You are looking now to set/confirm the proper XMC-1 distance settings prior to running the Dirac EQ.

This sounds like a reasonable approach to me.

I can help in confirming the distance settings. If you need any Dirac EQ help, others will need to provide that assistance.

So, concerning distance settings:
I don't really follow the logic for distance settings in Post 17. I'm also not comfortable trying to use EGD as a measure for that purpose. It may work just fine, but I no experience with that method so I have no good feel as to the its accuracy or repeatability. The phase tracking approach I use is likely to be a little more sensitive and thus at least as reliable however. So, If you want me to check your work I can do that. I will provide some charts with a brief explanation. If a distance adjustment is advised the needed change will be detailed.

Minimizing the needed measurements for a full analysis, I would need:
Setup
> Mic Centered at the LP
> Record and provide me the distance settings in the XMC-1 for these measurements.
> Set the XMC-1 desired 80Hz XO on using the chosen slope.
> MiniDSP Dirac box on, but no EQ, delays, or other active settings, i.e., set as a bypass mode.
> Engage either loopback timing or acoustic timing in REW (depends on USB or XLR mic type).
Measure
> SW's only
> FL main only
> FR main only
> CC main only (if desired)

From that mdat file of 4 measurements I will be able to calculate the SPL and phase responses of the 3 channels (SW's + FL, SW's + FR, and SW's + CC) using REW 'Trace Arithmetic' functions. I will also be able to shift the timing if needed to confirm the delay setting needed for optimal phase tracking.

Other Comments:
Phase and GD are related so either is usable for timing adjustments. EGD provides a cleaner trace than GD so that is why it works better. I assume you are just looking for a smooth transition from mains to SW's in the EGD trace. That should work pretty well given a relatively smooth trace. If there are room modes in the XO range that can make it more difficult. Using phase instead, the 'frequency dependent window' controls in REW help a lot in smoothing the phase curve so adjusting the timing is probably easier in that situation.
Using that method requires IR shifting and REW trace settings, so it is not quickly explained. Some idea of the process (a bit outdated now) is shown here.
 

·
Registered
Joined
·
6 Posts
The conversation here has gone over my head, but I don't see the graph I would use so I would mention it here:

Why not just run a full range measurement of your system then look at the Spectrogram: :laugh2:

A perfect impulse has a horizontal line as its spectrogram. You can tick the "Plot the peak energy curve" option to show the timing clearly.
[1st image]

When I first recorded my system it looked like this (also a synthetic impulse to illustrate more clearly):
[2nd image]

Two problems. Firstly a too-steep crossover was introducing excess group delay centered on the crossover frequency; secondly the sub was lagging the mains.

After lowering the crossover slope and delaying the mains, I got a measurement like this: (crossover frequency 70Hz)
[3rd image]

Finally, my sub has a +/- phase switch rather than phase knob (I wouldn't know what a knob does...) to set the right phase, I simply flicked the switch while playing tones around 70Hz and chose the position that gave louder tones (indicating reinforcement rather than cancellation).

Hope that helps :smile:
 

Attachments

1 - 20 of 64 Posts
Top