Large 2.2 system set-up assistance

Discussion in 'Official REW (Room EQ Wizard) Support Forum' started by Ed Zeppeli, Sep 11, 2017.

  1. Ed Zeppeli

    Ed Zeppeli Member
    Thread Starter

    Joined:
    Aug 30, 2017
    Messages:
    29
    Likes Received:
    0
    Hi all,

    I'm setting up my system and am digging in deep on the details here in my attempts to fully understand what's going on and maximize the gear I have. Here's part of the room used for audio...

    [​IMG]

    Behind the camera lense is another 22' of room.

    I'm trying to get best results of crossover between the horn modules and 15" woofer module (most importantly) and then to integrate the subs. I'm flexible on moving them around a bit and by using two I was hoping to eliminate some modes in the listening position.

    I have a Steinberg Cl1 soundcard, a Behringer measurement mic and a dbx Venu360 for processing. Separate amplifier channels are feeding the horn modules (2), the 15" woofers (2) and the 18" subs (2) for a total of six channels.

    I have 8 parametrics per band available as well as another 12 overall on the dbx input side.

    Initially I was hoping to use some voltage drive curves which I found for a very similar set-up but I can't get the slope right on the horn module because there is an internal crossover which will not allow it. I'd like to come up with my own slope equivalent to LR24db/octave in the 700-900Hz region but am not sure on how to accomplish this. Delays can also be set on any output in order to time align the drivers.

    Attached is an mdat file which includes three listening position measurements for each of four drivers. Subs have been excluded for now.

    The soundcard is set-up for loopback and I am familiar with the method for aligning the IRs in REW. Specifically I would like to know a good way of getting the most even alignment between woofer and horn in the crossover region. I saw an interesting idea using Arithmetic A-B and steepening the notch but I haven't wrapped my head around it fully and don't seem to be getting any notch whatsoever.

    If anyone wants to take a look at the file and offer a bit of direction it would be greatly appreciated.

    Thanks in advance,

    Warren
     

    Attached Files:

  2. Ed Zeppeli

    Ed Zeppeli Member
    Thread Starter

    Joined:
    Aug 30, 2017
    Messages:
    29
    Likes Received:
    0
    Just thought I'd add....

    The sweeps are full range which can be seen in the woofer measurements but on the horn modules you can see what the internal crossover is doing.

    Also the naming convention is LWOOF 1 = Left Woofer in microphone position 1 (of 3). SAM=the horn module aka Synthesis Array Module.

    Cheers
     
  3. jtalden

    jtalden Member

    Joined:
    May 22, 2017
    Messages:
    23
    Likes Received:
    8
    Impressive Equipment!

    First a look at the 3 potential LPs:
    After time aligned the woofers for each LP position, the phase tracking was considered for LWOOF vs. RWOOF for each LP position. We want the phase to track as closely as possible to so that EQ >300Hz can be common between them.
    It appears to me that LP2 (position 2) is the better choice. At LP1, and to a greater extent at LP3, the relative phase is significantly divergent through the operating range.
    If that temporary table in the photo was there for these measurements, its impact may have accounted for these differences so that may negate this comment.

    Regarding the XO timing:
    I do not use the process referenced for time alignment. Phase tracking is the better indicator of proper time alignment. [It still produces a null at the XO point if the phase is reversed on one of the drivers.]
    No matter the process used though, it is necessary to have the XO settings chosen and active for the measurements. This is because the XO settings impact the timing.
    Note that, the highpass XO filter (HPF) of the lower XO will only slightly impact the upper XO timing. The effect is relatively minor. I do suggest setting the lower HPF to something approximating the settings that will be eventually used, maybe LR-24 at 80Hz. Minor adjustments made later when setting final SW-Woofer XO will then have no significant impact on the upper XO timing.
    Conceptually:
    • Set initial Filters for both XOs
    • Measure, determine/adjust timing, analyze result
    • Adjust the XO filters as needed and repeat
    • Finished
    [We can adjust the XO filters for the lowpass and highpass to best approximate an acoustic LR-24 if we like. The closer we get, the closer the phase tracking will be. This additional work however will not necessarily provide a noticeable sound improvement to just using LR-24 electrical filters and accepting a gentle phase crossing situation if that occurs.]
    -----
    Just FYI:
    A process that avoids settings that at interact with future setting adjustments would follow a process like this:
    • Choose SW and LP locations based on the Vector average between the 2 SWs at the LP when the SW timing is aligned for each LP being considered.
    • Choose the mains locations considering phase tracking through the midrange bandpass.
    • Choose the lower XO with timing based on phase tracking through the XO range. [Try to stay in the range from 70-100 for the frequency and near LR-24 for the acoustic response shape.]
    • Set the upper XO with timing based on phase tracking through the XO range. [Adjust if possible to approach an LR-24 for the acoustic response.]
    Final EQ to a house curve is the final step.
    ----

    I will try to help with questions on any of this as requested.
    I can determine or confirm your upper XO timing given the correct measurements.
    For the upper XO that would be.
    Setup:
    • Both XOs active.
    • The delay timing is not critical. We just need to record what was used so we can adjust it as needed.
    • Mic at LP (or ~2m from the left baffle at LP height).
    • Loopback timing active
    • Full range sweeps
    Measure:
    • LH (Left Horn Module)
    • LW (Left Woofer)
    • LH+LW (both together)
     
    Ed Zeppeli likes this.
  4. Ed Zeppeli

    Ed Zeppeli Member
    Thread Starter

    Joined:
    Aug 30, 2017
    Messages:
    29
    Likes Received:
    0
    Thank you very much. I've moved things around a bit since last photo. Duplicate coffee table sub is now front and centre and utility cabinet sub is behind listening couch as shown.

    [​IMG]

    I've re-aligned subs, left woofer and horn modules.

    [​IMG]

    I'll update ASAP following your outline.

    Cheers,

    Warren
     

    Attached Files:

  5. Ed Zeppeli

    Ed Zeppeli Member
    Thread Starter

    Joined:
    Aug 30, 2017
    Messages:
    29
    Likes Received:
    0
    Ok. Here are the left woofer, left horn and both together. Crossover settings are in the notes for each driver. Delays are in.
     

    Attached Files:

  6. jtalden

    jtalden Member

    Joined:
    May 22, 2017
    Messages:
    23
    Likes Received:
    8
    Thanks, the measurements. They are just what was needed.

    Below, the current timing is acceptable as it provides good SPL support in the XO range. The phase crosses near the acoustic XO of ~1kHz. The XO range is roughly 400-3kHz so that is the range we are concerned with.
    SPL - Initial.JPG

    Below, the phase tracking in that range shows that the phase on both sides of the crossing is diverging at a relatively large rate. This indicates there is another delay setting that will provide closer tracking and better SPL support at the extremes of the XO range.
    Phase Tracking - Initial Timing.JPG

    The best setting was found with the Woofer delayed an additional 0.94ms (or the delay of the horns reduced that much). The Impulse chart below shows the delay adjustment made to a copy of the Woofer measurement. The second charts shows the resulting phase tracking. This delay adjustment is one full cycle at ~1kHz. With this setting the sound from both drivers (400-1.4k Hz) will arrive at the LP at the same time. With the original timing the sound from the 2 drivers at 1kHz has the Woofer leading the Midhorn by I full cycle. Note that the phase tracking is very close from 400-1.4k Hz. It then diverges rapidly to 3khz. This is the best that can be done with these XO settings. To get better tracking the SPL rolloff of the woofer would need to be at a higher rate. Possibly a 30 or 36 dB/octave LPF would help, but we don't normally find those filter choices available in a speaker management box. It normally jumps to 48dB/octave which is too much slope.
    Impulse - Woofer Shift.JPG
    Phase Tracking - Adjusted Timing.JPG

    Below, using A+B vector math of the WooferCopy+Horn is a comparison of the current Vs the new predicted SPL. We see some expected minor SPL improvement below the XO point , but little impact above it as the phase tracking there is not significantly better.
    SPL - Comparison.JPG

    Other Comments:

    • The delay change makes a more ideal alignment, but it may not be significant to the sound quality. The current setting provides almost as good SPL support in the XO range and that is most important criteria.
    • Either setting will provide a significant null at the XO point with one the drivers inverted. The phase there is aligned with either setting and thus when one driver is inverted they will be 180° out of phase and thus cancel.
    • The HF Horn is leading the MR Horn sound significantly; several cycles at the XO frequency. This is expected with a passive XO and the large offset between the 2 horn drivers. The XO there is high enough that this is not likely to have any impact on sound quality. I aligned the Woofer properly to the midrange horn without regard to the leading HF horn as those are drivers we are aligning. I only mention it because the positions of the woofer impulse to the horns impulse appears to too delayed when viewing the two new impulse locations on the overlay chart.
     
    Ed Zeppeli likes this.
  7. Ed Zeppeli

    Ed Zeppeli Member
    Thread Starter

    Joined:
    Aug 30, 2017
    Messages:
    29
    Likes Received:
    0
    Excellent. I really appreciate your assistance. I have lots of Xover options available to me. The Woofer Low Pass wasn't even a LR24 so I can easily remedy that.

    I need to figure out how you came up with that windowing on the phase portion that allowed you to see it the way you did.

    Gotta run right now but more to follow. Just wanted to say thanks much.

    Cheers,

    Warren
     
  8. Ed Zeppeli

    Ed Zeppeli Member
    Thread Starter

    Joined:
    Aug 30, 2017
    Messages:
    29
    Likes Received:
    0
    Ok I've replicated the window which allowed you to see the phase of the two drivers in a similar, and thus comparable vein. Very helpful.

    What I cannot figure out is where/how you came up with the .94ms delay? Was it trial and error or did you derive that number from somewhere else?

    Regards,

    Warren
     
  9. jtalden

    jtalden Member

    Joined:
    May 22, 2017
    Messages:
    23
    Likes Received:
    8
    Yes, trial and error.

    Basic steps:
    1. Make or load a copy of the lower frequency driver measurement.
    2. Set smoothing to 1/48 octave.
    3. Change the 'Window Ref Time' for 'all measurements' to the initial Impulse rise of the higher frequency driver. Remember the value used as it must be used again below.
    4. Set the 'Add frequency dependent window' (FDW) active with the needed width to clean up the phase trace (6-12 cycles is normally a good setting) [The FDW must be active or the Ref Time shift is not active and phase then cannot be easily read.
    5. Adjust the location of the lower frequency driver until the best tracking is obtained. (It takes some experience to recognize the direction and extent of the change needed to get there efficiently. It does not matter how many steps it takes us to find the best setting. [Note that after any shift of the impulse the Ref time set above must be loaded and activated in that measurement. This is because REW will normally change the Reference time by the amount the impulse is shifted. We must have a consistent Ref time for all measurement for phase comparisons. This can be particularly confusing as REW only shows one decimal place in the Ref time setting and for higher freq XOs we normally are dealing with 2 decimal place changes so we can't see the actual setting in the REW popup.]
    6. When the phase is best aligned in the chart, measure the distance the impulse has been shifted as shown in the 3rd chart in Post 6. That is the shift needed to the delay setting for that driver to actually achieve this new phase tracking condition.
    Regarding This Particular Analysis:
    • In Step 3 the Ref time was initially chosen as 100.62ms. The expectation was the phase of the SAM would then be relatively smoothly falling through the XO range. Instead, the phase was as expected above 7kHz, but steeply falling below that. It was clear that the HF horn in the SAM was leading the MR horn. After a couple tries it was determined that a ref time of 101.00ms would provide the more gently falling phase response in the XO range that is more appropriate for this analysis. This suggests the HF horn leads the MR horn by roughly .38ms. This is more likely a better estimate than the 0.94ms that was suggested in Post 6.
     
    Ed Zeppeli likes this.
  10. Ed Zeppeli

    Ed Zeppeli Member
    Thread Starter

    Joined:
    Aug 30, 2017
    Messages:
    29
    Likes Received:
    0
    Great. That worked for me and I now have a much better grasp of this. Incidentally, in the graphs I was experimenting with I had previously generated minimum phase. I used the first rise of the generated minimum phase as a guideline for the woofer delay alignment in the IR Window and that seemed to have put me in the ballpark.

    In the dbx app, there is the ability to adjust phase from 0 deg down to -180 deg. At what point would this be a more appropriate method of tweaking phase than using delays?

    Thanks again,

    Warren

    [​IMG]
     
  11. jtalden

    jtalden Member

    Joined:
    May 22, 2017
    Messages:
    23
    Likes Received:
    8
    I don't follow this thought. Can you provide a little more detail?

    I don't know.

    Long ago, I did investigate my DCX phase control briefly since it did not provide the XO filter slope needed to improve my tracking. I was never really completely clear on its action. It only appeared to work on lower frequency XOs and only provided a delay type of adjustment. It was easier to just adjust the delay and not use this control. I have seen others indicate that some phase controls change the phase rolloff rate in just the XO range with no impact to SPL. I assume this is an allpass filter effect. I have not personally seen this. Maybe that is what happened, but was just in the wrong direction to help me. I really don't remember anymore.

    To the extent that your phase control provides the change you need to improve phase tracking then it is just another way to improve your result. Try it out and let me know if it helps or not. Is its impact different from a simple delay change in your case?

    Just FYI, another option (more off the wall) is to use an extra EQ filter in the SPL rolloff range to modify the SPL rolloff rate and thus phase. Below, I show my 5 main speakers each with a midwoofer and tweeter where this method was used. Shown is the SPL rolloff dropping 40dB through the range from 500-7k Hz. Next is the Phase tracking through that range. Finally the extra EQ filter that was applied to achieve that. It is filter number 1 applied to the midwoofers to increase their rolloff (+8db, 10.3kHz, 1Q). I happen to be using that filter now. It actually provides pretty good tracking down 60dB from 400-10k Hz. It was done as a learning experience. Normally I don't use it as this is the least of my concerns. Good phase tracking for a 25dB SPL drop is probably way more than adequate. Any crossing at the XO point provides good SPL support and thus sounds very good so long as it is no more than 1 cycle or so from the ideal setting (it depends on the XO frequency I suspect).

    SPL.JPG

    Phase.JPG

    DCX EQ.JPG
     
  12. Ed Zeppeli

    Ed Zeppeli Member
    Thread Starter

    Joined:
    Aug 30, 2017
    Messages:
    29
    Likes Received:
    0
    REW will generate a Minimum Phase for us in the Impulse window under 'Controls'. When I compared my L Woofer Copy to the Minimum Phase I used the difference in peaks as a starting point for tweaking the delay for best summation at Xover and it seemed to get me pretty close. In the case of the photo below, I'd delay the woofer 0.475ms as a starting point.

    Does that make sense from an explanation point and from a technical stand-point? I'm really not sure but I thought I'd try it.

    [​IMG]
     
  13. John Mulcahy

    John Mulcahy REW Author

    Joined:
    Apr 3, 2017
    Messages:
    184
    Likes Received:
    17
    The soundcard cal for those measurements behaves pretty oddly at the top end, suggest you remove that.

    Here is a view of the BOTH measurement from the forthcoming 5.19 beta 7 that might help with timing alignment decisions:

    wavelet.jpg
     
  14. Ed Zeppeli

    Ed Zeppeli Member
    Thread Starter

    Joined:
    Aug 30, 2017
    Messages:
    29
    Likes Received:
    0
    Well that looks interesting!

    When can we expect to try it out?
     
  15. jtalden

    jtalden Member

    Joined:
    May 22, 2017
    Messages:
    23
    Likes Received:
    8
    No, not that I recognize. There are various schemes and I may miss your point. It's true that, If both impulses have their initial rise fall close to each other in time, then that normally suggest the timing is close to ideal already.

    My general thoughts are:
    • The exact timing relationship between the 2 impulses must be maintained for any phase evaluation.
    • We can shift both measurements the same amount back near 0ms, or we can leave them both as measured to see the phase relationship.
    • In either case we need to be careful to set the appropriate reference time and an FDW. If REW has different reference times the relative phase shown by the chart will be misleading.
    • The chart you posted shows the woofer near 0ms. If REW offered the IR shift value, the default condition is to place the peak of the IR at 0ms. The minimum phase impulse location is nearer the initial rise of the impulse as shown in your chart. If we unselect 'set t=0 at IR peak' in 'preferences/analysis' then REW will shift the IR peak near the initial rise of the IR and thus the both IRs will look to be more closely aligned.
    • Again, It is true that it's good starting point to locate the initial rise of the 2 impulses near each other to achieve a close starting point. The best delay setting often looks like that. The exact relative positon changes a little however with different acoustic XO conditions so the ideal position is best set based on the phase tracking of the direct sound from the 2 drivers.
     
    Ed Zeppeli likes this.
  16. Ed Zeppeli

    Ed Zeppeli Member
    Thread Starter

    Joined:
    Aug 30, 2017
    Messages:
    29
    Likes Received:
    0
    Thanks gain jtalden.

    As I work through this I'll post my results to keep this thread complete.

    "Just FYI, another option (more off the wall) is to use an extra EQ filter in the SPL rolloff range to modify the SPL rolloff rate and thus phase."

    I had actually tried this initially by importing some curves provided by the designer for bi-amping the system and then EQing to them as Target curves. For various reasons (namely the built-in horn crossover) I couldn't match the more gradual slope provided so I abandoned that idea. I did, however get a very smooth, two-stage Low Pass slope off the woofer so the method is sound.

    Thanks also for pointing out the UHF timing. It may be fairly simple for me to pop open the back and disconnect it for impulse measurements at some point.
     
  17. John Mulcahy

    John Mulcahy REW Author

    Joined:
    Apr 3, 2017
    Messages:
    184
    Likes Received:
    17
    Aiming to get the next beta out by month end.
     
    Ed Zeppeli likes this.
  18. jtalden

    jtalden Member

    Joined:
    May 22, 2017
    Messages:
    23
    Likes Received:
    8
    Thanks John. I should have noticed that.

    Your posted chart clearly shows the same offset between the 2 horns near 6kHz that I noticed. There is hint of the change needed at the 1k XO as well. The magnitude of that change may be harder to determine.

    Attached is the mdat containing the 'A+B' for of the adjusted timing in case you want to compare it to the 'Both' measurement.
     

    Attached Files:

  19. Ed Zeppeli

    Ed Zeppeli Member
    Thread Starter

    Joined:
    Aug 30, 2017
    Messages:
    29
    Likes Received:
    0
    OK, Here's where I'm at...

    I disconnected the UHF on the left horn module and have omitted the sketchy soundcard.cal for these measurements.

    There is no system EQ or Delay in these measurements. I've flipped the polarity on the SAM. Upper High Pass is set to 650Hz LR24 and Woofer Low Pass is set to 592Hz LR24 giving me an acoustic XOver of near 750Hz.

    [​IMG]

    [​IMG]
     

    Attached Files:

  20. jtalden

    jtalden Member

    Joined:
    May 22, 2017
    Messages:
    23
    Likes Received:
    8
    The alignment as shown is ideal. No further fine tuning needed.
    The LR-24 LPF increased the rolloff and helped the phase tracking above the XO significantly.
    Did you shift both Impulses the same amount?
    0ms delay on both drivers for these measurement?
     
  21. jtalden

    jtalden Member

    Joined:
    May 22, 2017
    Messages:
    23
    Likes Received:
    8
    I posted too quickly, There may be a better setting. I will investigate further and advise.
     
  22. Ed Zeppeli

    Ed Zeppeli Member
    Thread Starter

    Joined:
    Aug 30, 2017
    Messages:
    29
    Likes Received:
    0
    Here's what is looks like with the only change being to move the woofer (offset) forward in time .25ms. Trace arithmetic shows a pretty deep notch of 28dB at xover.
    [​IMG]

    Phase overlay now...

    [​IMG]
     
  23. John Mulcahy

    John Mulcahy REW Author

    Joined:
    Apr 3, 2017
    Messages:
    184
    Likes Received:
    17
  24. jtalden

    jtalden Member

    Joined:
    May 22, 2017
    Messages:
    23
    Likes Received:
    8
    Using the file from post-16, I get the best tracking with the woofer inverted and delayed 0.4 ms.
    I used a 6-cycle FDW. Post-23.jpg
     
  25. Ed Zeppeli

    Ed Zeppeli Member
    Thread Starter

    Joined:
    Aug 30, 2017
    Messages:
    29
    Likes Received:
    0
    Thanks John. Is this data from my freshest data in post #19? This info correlates with the nasty dips I have at 258Hz and 429Hz. I will have to think about how to deal with those.
     

Share This Page