How does the 3rd sub time align and PEQ look?

[Usher99]

The CAO (chief aesthetics officer) OK'd a 3rd sub (KEF kc 62) which was placed near a 3rd corner along with a passive sub (sub 2) and my Paradigm UC 10 (sub 1).
In a bonehead move somehow lost my data (older file was open and no room for the new) by accidental deletion. I can't get the data back, sigh. I can redo the measurements on another channel in the minidsp next weekend if necessary.
I had gain matched the subs.
Anyway, I do have a final screen shot and the delays and PEQ are fine in the minidsp. I used the time align feature of REW to optimize the delays to sub 2 and then the delay to sub 3 using the virtual combined (sub 1&2) rather than brute force experiment as before with two subs. The measured response wasn't quite as good after the delays in the combined virtual sub with all 3 going.
Ended up inverting the 3rd sub (KEF) with a 5msec delay. Not sure the augmentation was that good. I also changed the MLP. Perhaps I should input a shelf filter to boost the output? The new 3rd sub almost looks better by itself. I am a Newbie at this.
I may try MSO though looks complicated with Java drivers--after at least get set up with MLP at least. Still have to run ARC on the Anthem. Appreciate any comments.

 

[Usher99]

The subs were gain matched with a test tone at 50 Hz. The yellow highlighted curve is for the 3 sub virtual sub after time aligning and the PEQ filters loaded into the minidsp. Isn't it fishy that the output of the combined 3 subs is so low at 50 Hz? Any comments most appreciated.

 

[JStewart]

Isn't it fishy that the output of the combined 3 subs is so low at 50 Hz?

Yes. If the measurements are valid and the subs aligned in phase then the combined response must be greater than any of the individual responses.

 

[Usher99]

Thank you for commenting. The PEQ filters cut the gain, but it seems excessive. Perhaps I should start from scratch and run more sweeps being sure the predicted optimal delays with the REW alignment tool worked ok.

 

[JStewart]

Well then, if eq has been applied to reduce output in that region then not fishy.

If the question is should predicted results in REW for time alignment and peq results = actual measured results afterward, the answer is yes. If not then something isn’t right.

 

[Usher99]

Thank you for looking.

Perhaps the PEQ filters somehow were overzealous. Perhaps I should clear the filters and shift the house curve higher and see what happens. I wish I had not lost the sweeps. It sounds fine with music but a bit underwhelming for the one movie we watched since the redo. I don't want to lose the output at 40-60 HZ. I'll also do a sweep of all 3 subs pre and post alignment with no EQ to be sure nothing went askew. I realize with wildly different subs it won't be optimized under all that many conditions. The KEF also has non -modifiable dynamic EQ but I don't think that kicks in until it gets stressed.

 

[JStewart]

I don't want to lose the output at 40-60 HZ.

 

[deercreekaudio]

This white paper outlines a simplified approach to Home Theater Multi-Subwoofer Setup and also contains excellent references for looking deeper into areas such as delay calibration.

https://deercreekaudio.com/p/home-theater-multi-subwoofer-setup

Deer Creek audio is an authorized mini DSP dealer.

 

[Usher99]

Thanks for the link. The mini DSP 2X4 HD is awesome. A bit intimidating to start but Hometheaterguru Ep 7 demystifies most issues and emboldens the non-engineer enthusiast to venture forth.

 

[Usher99]

I redid everything and have attached the data file. I am puzzled about a few things. First how low the output was on sub 3 (KEF kc62). I had gain matched in the position of sub 2 (passive) with a 50 Hz tone. It is about 11.5 ft from the MLP and sub 2 about 2 feet. Sub 1 is about 10 ft and in front of the room. I used the alignment tool in REW to estimate the delays but then did a few sweeps (didn't save most of them) to optimize a bit better. Perhaps I made poor choices as the aligned/PEQ'd subs is arguably not substantially better than all 3 subs with no delays or EQ? As expected, the 2 big nulls did not respond to any boost on the input side. I ended up boosted sub 3 by 8dB in the minidsp but it didn't do that much. Should I proceed to integrate Mains and subs or try something else with the subs?

This was pre 8dB boost for sub 3.

 

[welldun]

I have some additional questions that can help figure out which approach you could take. Here are the questions along with some general strategies that I use when using multiple subs.
1. How many seats/ listening positions do you have in this room, and how many are you optimizing the system for?
2. What are you using to drive the passive subwoofer?
3. It's unclear to me whether you have all 3 subs co-located in the same 3rd corner, or if each sub is placed in it's on separate location. Can you clarify?

Now as for the subs themselves, since they are not all the same models, and have different amplification driving them, it follows that the delays would be different on each because of the different electronics through which the signal is being channeled before arriving at the drivers.

In general, when combining multiple subs via a miniDSP, I first get the subs aligned to each other, before moving on to integrating them with the main speakers.
One way to do this is by clearing out the miniDSP and making sure that there is no EQ or filter set at the AVR. You want to be able to get a clean signal to the subs and see what their individual response is before combining them. When it's all said and done, you want the AVR to detect the combined subs as if there is only one large and more efficient subwoofer in the system. That allows the AVR calibration software to use that subwooferresponse as the only one it has to now integrate with the main speakers.
There is more to the process but I don't want to write a whole bunch here without 1st knowing how you are driving that passive subwoofer. This is because the process for integrating the subs would be different if you are using the AVR itself to drive the passive subwoofer as opposed to some Independent amplifier.

 

[Usher99]

Thank you so much for thinking about this. I have a couch along a long wall on the side and two chairs directly in line with the Ctr channel. I put the U-Mik1 at ear level at one of the chairs
(My seat) and designated that the MLP. The couch is close so took a measurement there. I hope to use MSO too at some point after getting a nice setup for the MLP. I can save that a separate config in the mini dsp as I understand it.
I had a local tech go over and recalibrated My NAD 3140 and bridged it for mono to power the passive sub (sub 2). It is write on spec and the bit of crackling on start up is gone. Sub 1 (Paradigm Ultracube 10) is in front left corner on same plane as TV.
Sub 2 (Passive) is in corner behind MLP. Sub 3 (KEF kc62 ) is not too far from right rear corner--there is a fireplace with the wall jutting out and a "corner" is there. So all three subs in own location. I wanted to even out the response in the room. My alignment and EQ seem questionable. The AVR sees one virtual sub. I have not run ARC yet as am not sure I did an adequate job with the subs.

 

[welldun]

I took a look at the MDAT file you shared and noticed a few issues with the measurement results.

1. First thing I noticed is that the frequency response/measurement seems to be cut off around the 120Hz mark. This tells me that you might need to adjust the crossover setting on each sub (open it fully) as well as in on the AVR in order to allow a full measurement sweep from say 5Hz to 200Hz. The purpose of doing so is to see what the actual output of each sub is. For this, you want to bypass any limit that a low crossover setting can impose. There is something odd about your combined measurements in that they show the subs output extending down to around 10Hz yet each sub on there own can at best go down to 25Hz.
2. The next thing I noticed is that the output of sub 3 is severely limited compared to sub 1. The 8db boost that yo added for the subsequent measurement made close to no difference. This tells me that the gain on that sub amplifier is not sufficiently high or something else not set properly with that sub. *Make sure that the MiniDSP does not have any active PEQ, Crossover, and also that the output on each channel is set to 0 (zero), that way each sub is getting the same signal level sent to it amplifier.
3. I understand that you want to use MSO in the end, but I recommend that you first get each sub playing optimally before using MSO or any other type of DSP.
4. If you've gain matched the subs properly and the signal going to each is the same, you should not be seeing such difference between sub 1 and sub 3.
5. This brings me to the actual measurement volume that you are using. If you are routing the laptop to the AVR via HDMI then you want to make sure that the AVR has all DSP settings disabled and the channel levels for all speakers and sub out are set to 0 as well. Once that is done, in REW you can use the GENERATOR to produce Pink Random noise for the Full Range and route it to your Left Main Speaker while using the SPL Meter (with your measurement mic at the MLP) to set the AVR volume knob at a level that reads 75db on the meter. Take note of what that volume setting is, and use that moving forward. This will ensure that the measurement sweeps are all being done at an equal and adequate volume.
6. Once you've figured out the above setting, you can send pink noise to each sub (one at a time) to see what the output reading is at the MLP. you want If all 3 subs are close to the same distance from MLP, then you could use the gain control on each sub to set them to the same output at the MLP. Hopefully this does not max out the gain on any of them. There is a way of gain matching the subs by placing each sub in the same location and sending a signal to each (one at a time), and adjusting the gain for each sub so that the output level matches. You can do that if you prefer, then move the subs back to their respective locations and then level match them by sending the same signal again to each but this time adjusting the gain via the miniDSP. Ultimately, you want to start out with similar output level from each sub you can best assess their individual and combined responses. *Episode 7 of HometheaterGuru has good info on all of this, so you might want to watch it again in case you missed something.
7. If the couch on the side is too far from the MLP, then trying to calibrate for both locations will be difficult. For the purpose of getting things set properly this first time around, I would focus on getting the MLP set up correctly first. Once you have that as a good baseline you can experiment down the road with things like MSO to try to broaden the sweet spot. Get the basics right first and then, when ready, go ahead and experiment. I use MSO and can tell you that it is not an easy program to use if you don't have the basic down.
8. Again I would say to you that if you get the above sorted out, then you can use the great walkthrough on Episode 7 of Hometheater Guru as well as the explanation he gives on episode 9.

There is still more that I can write, but I want to make sure that you have the basics down first so that you don't waste time chasing errors. I'm a hands on type of guy, and could probably help you better if we were communicating in real time as you were doing this. If you'd like to set up a call, that might be more fruitful. Let me know.

 

[Usher99]

Thank you for looking at the data and taking the time to write back. I think I had the crossovers on the two powered subs (sub 1 and 3) to bypass, but I'll check they didn't get jostled by me when I did the gain match or my cat for that matter. I couldn't for the life of me get ASIO drivers to work (Flex or ASIOforall) so had to use the java ones like in Ep 7. I set the main xover to 250 Hz on the AVR as directed. I can send pink noise to each sub via the AVR directly.
2. I am stumped by that! The KEF is a little itsy sub but has a 1000W amplifier and gain matched with gain on the sub set at 50%. It is tucked away in an odd position. With my first go around weeks ago I did put a xover starting at 30Hz or so for the passive sub as much below that the distortion went through the roof. I think the Paradigm (sub 1) rolls off at about 20HZ but I got the distortion down by a modest extra roll off below that too. It did not effect the output, but I can bypass for testing.
3 I agree totally and I think my subs are wildly far from optimized.
4. I used a 50 HZ tone from a youtube file with my UMik-1 at the MLP.
5. I have never run ARC (Anthem room correction) yet so I don't think the AVR is messing it up.
6. I did not level match as thought the gain match was sufficient. Something is up with sub 3 that I do not understand. It has some output (purportedly) down to 11 Hz.
I have no engineering background but am a lifelong enthusiast and persistent, so perhaps I'll eventually get this right. I never would have attempted this w/o the clear nuts and bolts approach of ep 7. Using REW time align feature to get in the right ball park for the delays and ?of inverting is much faster than a ton of sweeps. My sweeps didn't always match precisely---trust but verify. ;-)
7. Couch is not that far from the MLP and I hoped the 3 subs would even out the response some. MSO with Java drivers looks challenging. I have the program and spent a couple hours reading but won't attempt that until the MLP looks decent. I should fix any inadvertent crazy errors and then perhaps a call would be great, though I make no presumptions using your time.

 

[welldun]

Sounds like you've put in some good hours trying to get it all to play nice. It will all pay off in time as all of these tests and trials will deepen your understanding of what the you want and how to get it. Sometimes what may seem obvious to some just doesn't click for others right away. So we might overlook the basics and jump right into something more complex without fully understanding why it all works. In my case, for years I didn't fully understand that having multiple subs was primarily for the purpose balancing out the overall response so that it creates a broader sweet spot. I totally missed the point of having the subs aligned so they work as one efficient unit as opposed to multiple units trying to do different things. Once I understood that part, it was like a vail was lifted and I saw things clearer. Things like the MiniDSP made more sense after that revelation.

In any case. getting back to your set up... I would recommend spending a few minutes checking the AVR settings just confirm that none of the DSP settings are ON. You want to make sure that all EQ settings are off, as well as any other type of enhancement that is available. Once you confirm that, you can move on to the MiniDSP and make sure that it too is zeroed out. Then Make sure that the Phase setting on all of the subs is set to zero.
Would be good if you can get the ASIO4ALL working for you. I've had good success using ASIO4ALL_2_14_English.exe with both windows 10 and now windows 11. Once you install it, you can use it with REW V5.20.5 . It could take some tries, but once it locks in, it works perfectly. Page 36 of the Getting Started with REW: A Step-by-Step Guide (Rev 6.2, June 23, 2021) has some troubleshooting steps to help get ASIO4ALL working. I've attached a copy of that manual just in case you don't have it.

In the measurements that you attached it looks like SUB2 is inverted (based on the Impulse response). I'm not sure how the settings for that sub are setup in the amp, but make sure those are neutral also.
With that said, since all of the subs are different, I would recommend placing your best sub upfront where it can give you the best solo performance, and then use the other subs to fill in the gaps. If the Kef is the strongest on paper, then bring it up front next to sub1 and see if performs better than sub1. If you move it upfront and the output is still anemic, then check all of its settings again to make sure that you don't have something limiting its output. You can even swap the two subs connections to see if the signal going to sub3 is the problem.

 

[Usher99]

Here is more information. I could work from home today and had a few minutes only. The low output in sweep sub 3 mystery was investigated with a modification of your suggestion. I didn't have time to get out the Umik-1 etc but used my phone as a rough dB meter. I used the white noise generator in the AVR with all the miniDSP settings as is including the 8dB boost for sub 3 that didn't seem to do much in the REW sweeps.

With phone mic at MLP sub 1--about 51 dB (Paradigm UC-10)
sub 2 64 dB (Passive, vented)
sub 3 64dB (KEF kc-62)

All 3 subs---about 67 dB

To at least in part to divorce the room effects I took a reading 1 foot from the subs with just one going at a time. The passive sub is vented and I put the mic about a foot from the vent.
Sub1 68 dB
Sub 2 68 dB
Sub 3 80 dB

Seems like that boost in the minidsp is doing something. Perhaps something is fouled up with my REW settings.

I had purposely inverted sub 2 and 3 and could get a bit better summation using the alignment tool in REW.

I realize the phone app mic is very crude but should be uniformly crude for these measurements just to see what is going on.

 

[welldun]

The 80dB nearfield vs 64dB at the MLP suggests that either sub 3 is sitting in a null or it is quite far from the MLP. They say that for every doubling of distance from the source of the sound, the SPL drops by about 6dB.
All in all, I recommend just starting from scratch with all settings zeroed out. Use 1 main location to set the SPL of each sub and then, using that same location for each sub, check the SPL at the MLP. If you get them all equal using output 1 from the minidsp, then do the same with output 2 and output 3. That will help you rule out the minidsp. you can do the same with the signal cables... use one cable to get a measurement reading, then swap that cable with another one to make sure you get the same reading from the other cables. If that all checks out fine, then its time to consider a new location for sub 3.
If you care to, send me the dimentions of the room. We can use a room mode calculator to figure out if the locations being used for the subs and MLP are poor choices.

P.S. In REW, you can change the number of measurements that you want to save per session. Open Preferences, then click on the VIEW tab. Then, on the right hand side of the lower section label INTERFACE, you will see "*Maximum measurements:" just change the number to 100 and you should have room for plenty.

 

[Usher99]

I understand. Thank you for thinking about this again. Here is a quick sketch. Sub 3 is 11 ft from the MLP. The CAO largely dictates the allowable play in the sub locations. I have maybe 6" of flexibility.

I may have left out a couple critical measurements. MLP ear level is close to 2.5 ft from rear wall and 2.5 ft from side wall. A majority of the rear wall goes back 19 inches more for a window seat. We have shades on them that should dampen some reflections.

With the pink noise from AVR sub 1 has larger decrement (near field-MLP SPL) than sub 3. All very mysterious to me. The cables are SVS. I still have the minidsp configured with all the settings from my last alignment attempt and the measurements are with those in place including the 8dB boost to sub 3. I hope to have some time this weekend.

 

[welldun]

I see. With the couch up against a wall like that, it will be difficult to get a similar response to that which you can get at the MLP. Also Understand that if MLP is not centered in the room (at least left to right), then the output from the speakers furthest from the MLP will be driven higher in order to make up for the drop in SPL that comes with the increased distance. If you are able to set the MLP closer to the centerline you will get better imaging from the speaker pairs.
In any case, getting back to the subs... you mentioned that you are still using the previous (old) settings in the minidsp, this tells me that there is a chance that you have some PEQ's or filters set up in there which could be negatively affecting your current setup attempts. So again, I recommend that you reset the AVR and the minidsp, or use a different config in the minidsp (which has 4 available). Just make sure that when you do, both the PEQ sections and the Crossover sections are bypassed. This will allow the minidsp to passthrough the signal without coloring it. Then take new raw measurements for each sub using the left main speaker as the acoustic timing reference. This will give you a true view of how each sub is performing at the MLP as well as its distance from the MLP in comparison to the Left Main Speaker. from that point on you can start to make the necessary tweaks to get the system working optimally in that space.

 

[welldun]

Not sure if you've seen this room mode calculator before, but using your room dimensions (had to guess the height), It shows you all of the spots in the room where you can get Peaks and Dips for the frequencies that most affect your room based on the room dimensions. The top of the arch is were that frequency is at it's peak and each leg of the arch touches the spot in the room where that frequency would be in a null.
The way you read the chart for each dimension (Length, Width, and Height) is from left to right. The Length section starts at 0ft which is your front wall (the one you face when watching tv) and 15ft at the rear wall (closer to your MLP). For the width,0ft is your left wall and 13' 6" is your right wall. and for the Height, 0ft is your floor and 9ft is your ceiling.
Using those estimated figures. you can calculate where to place speakers and subs as well as your seats. For example, using the Length info above, your MLP is at the 12 1/2 foot mark which puts you at a null for the 113Hz frequency. In that case, moving your MLP up by 1/2 foot, could improve that 113Hz for you. In a perfect world you would want to avoid sitting in a null just much as you would want to avoid a peak. Notice how all of the frequencies peak as they approach the walls.
They also say that if you place a sound source (speaker or sub) at a null and that source is reproducing that frequency, it fixes/ balances out the null.
all of this is in theory since real world situations tend to be slightly different, but it gives you a good starting point.

 

[Usher99]

Thank you for taking the time to do that and explain the results. The MLP is only slightly off ctr--my drawing was lousy. With your previous suggestions and the good output from sub 3 in the near field measurements using the filters and inversion on the minidsp, I wonder if the sub just needed to be rebooted. I have never done a timing measurement and my modest reading on MSO suggests that is a bit difficult with the java drivers. One kind forum member spelled it out for me--have to hook the minidsp to the preamp out on a channel in the AVR and unplug the other main channel, IIRC. It is easier said then done given the cramped quarters in the cabinet.

I'll use a fresh minidsp config this weekend and try again. Job after that is a retry with ASIO drivers using the debug suggestions you provided. I don't have ARC engaged at all so the manual distance readings I inputted into the AVR shouldn't do anything to the timing, I don't think. I don't think the manual said much about that though. The surround sound is that great yet and haven't even used Atmos. First step is getting the subs working together and positioned as best possible. I suppose there are much much worse problems, but this is important to me. Thank you again.

 

[Usher99]

I redid everything. Sub 3 (KEF kc-62) I think was the culprit. It is a few months old but some sweeps at lower levels resulted in intermittent oddly low output responses. I boosted it by 5dB and all sweeps were consistent. Perhaps I should have it looked at under warranty, but it sounds OK. When I sent a 50Hz tone to it, sometimes it seemed not to start up right away. The other subs didn't do that.
Anyway the final sweeps with PEQ in place is below. I put in a smidge EQ at the nulls and it did make them a bit less deep but not much. Perhaps a waste of amp power. The nulls are only in the composite results.

Another thing I learned using the REW alignment tool is the results seem a bit better if align the first two subs and use the actual measured results not the theoretical results to align the next sub.

Here is the result at the couch--doesn't look too bad??

Here is my house curve:

Appreciate any suggestions. I want to get an approximate setup with ARC then fine tune and maybe try MSO especially if can get ASIO drivers working. Not sure if can trust ARC to integrate the subs and mains. Thanks for looking.
I attached the data too if it would be helpful.