Parallel Sub Configuration / Simulation Question

[whoareyou]

I've been experimenting with two subs in parallel, and like the overall results and I believe system sounds best with this technique. For me much better results than using AL two sub solution with each sub separate.
But I have question (as usual ), this time regarding simulation of the parallel configuration.

At 40hz area simulation shows huge dip, which to me doesn't make sense based on the measurements shown in frequency response. Why this dip when other areas with dips show much better results in simulation?
Of course, verification would be to measure actual results, but I have yet to figure out a way to measure actuals via my setup (Jriver / ASIO dac / USB Mic).

Absent ability to measure actual results, does anyone have a suggestion as to why I see these results in the simulation (attached smoothed and actual measurement)?

Also, anyone have success with measuring a multichannel setup via JRiver / REW /USB DAC / USB mik (umik-1)? .

 

[Ofer]

Are there clock issues when using the UMIK1 with asio dac? I see your DAC has USB inputs. Can you plug the mic to the DAC usb input? Do you have an analog input for a mic?

 

[whoareyou]

Are there clock issues when using the UMIK1 with asio dac? I see your DAC has USB inputs. Can you plug the mic to the DAC usb input? Do you have an analog input for a mic?

From measurement to measurement the speaker timings are very consistent. Unfortunately, DAC (exasound e38) only has analog output with USB only for computer interface purposes.

But looking at this further, I think AL may be having issue with determining start of the subwoofer's IR.
Looking at the manual noise removal for one of the subs, AL seems to change the frequency response (blue - top graph), which I've never seen before, and has removed a portion of the measured IR response (red).

 

[whoareyou]

So, issue was out of phase subs.
These are two very different subs. One is a svs sealed 3000 micro, and the other is a ported HSU stf2, I suppose I shouldn't be too surprised.
I had assumed that AL would take care of this, but I guess there is only so much that room correction is capable of in this situation.
Flipping the phase switch on the HSU and remeasuring, and completely different results with the correction.

Although, the correction ends up with a couple of dips (in new areas), 14DB filter correction mostly takes care of this and I am very happy with the result (I could probably fine tune the subs via the SVS phase control, but good enough for now)..

Seems to all prove out in the listening sessions which has distinct improvement in the bass for both 2 channel and multichannel.

.

 

[Ofer]

Glad it is solved. Can you add a legend as to which curve is which?

 

[whoareyou]

red = SVS, blue = HSU Grey = target Pink = Simulation

 

[Ofer]

Seems quite a difference between the simulation and the actual measurements.

 

[whoareyou]

The filter has a 14DB correction. My room is fairly difficult, and I have large variations in the bass regions below around 200hz, and also a difficult null at around 80hz (you should have seen what that looked like before I added a modest amount of room treatment),

Attached image is a bit busy but I think you can get idea of what is going on.

Overall, I end up with fairly flat response with my target curve - and it sounds great. I'm amazed that AL provides results that it does.
It's one reason why I switched away from another DSP solution and now go with AL. Without the customization that AL provides I could not achieve anywhere near the results I achieve for this room.

 

[Ofer]

Busy indeed. Can you remove the simulations and leave just the target and actual measurements? is this a 5.2 system? did I get the number of speakers right? Also how are the subs and mains crossed? it seems that the subs and mains overlap above 30hz.
wonder what modest treatment constitutes

 

[whoareyou]

Here you go. This is a 5.1 system with two subwoofers used in parallel.
My crossovers are mostly being used to fill in some of the nulls, which i.e. is why you see one of the sides at 90.

 

[Ofer]

OK. according to the table you cross the mains at 50hz, but on the FR it looks like everything (mains and subs) roll off at around 30hz. Which coinsides with the expected FR of your front speakers. BTW how do you measure? with the UMIK1? A

 

[whoareyou]

Umik1 - Yes

 

[Ofer]

Umik1 - Yes

And no clock issues as reported here often? also if you can measure AL with UMIK1 why can't you use it for REW?

 

[whoareyou]

Only issue is WASAPI which always provides bad timings. For me it just doesn't work, so I work around this by using direct mode, drift correction and separate stream options. Measurement to measurement the timings are very consistent. Once in a while a measurement will fail with bad dynamic range, but I don't think that is anything unusual.

At this point, only UMIK-1 thing I wonder about is accuracy of the factory's calibration file.

 

[Ofer]

I get you. Did you try to overlap the sub with the mains? This is something I really like to know if possible on AL. Bertt answer was somewhat vague and then he disappeared. BTW what DSP option did you leave for AL?
I see your speakers can be bi-ampted, do you use that feature?

 

[whoareyou]

By overlap do you mean doubling up the bass?

Otherwise, doesn't XO naturally provide overlap between sub and speaker with overlap amount controlled by width of the XO frequency being offloaded, and the bass offloading is then summed appropriately to correct levels?

With what I know about AL (and compared to others, I just scratch surface with some of this stuff), that is what I would try to do, and either measure or listen for impact to bass.

 

[Ofer]

By overlap do you mean doubling up the bass?

Otherwise, doesn't XO naturally provide overlap between sub and speaker with overlap amount controlled by width of the XO frequency being offloaded, and the bass offloading is then summed appropriately to correct levels?

I am not sure what you meant with your explanation so I will sum it up with numbers. Assuming your speakers that play all the way to 30hz. Can you set the system in the following way:
Front (Left and Right) 30hz-20Khz
Subs (summed or individually) 20hz-100hz.
This is an overlap in the region of 30-100hz. I can do it in my miniDSP as each output channel is completely independent. Still miniDSP bears many other limitations.

 

[whoareyou]

Sorry, I'm not familiar with MiniDSP, so when you say MiniDSP overlap, are you duplicating the bass or is it an XO? If you are duplicating then wouldn't some bass be doubled up at many frequencies i.e main playing 40hz and sub playing 40hz at same level for same portion of track?

But in AL maybe if you set XO somewhere between 20 and 100hz together with very wide XO point the sub and main overlap over a wide portion of that frequency range and levels summed properly so that no particular frequency is too high a level per target curve. The bass doesn't just cut off at a particular XO frequency like brick wall.. It's a gradual roll off and you can control the roll off via width of XO filter, and there will be overlap of frequencies between the two speakers.
Have you tried this and observed the interaction between subwoofer and mains? Just create a few different filters with various XOs / widths and check the simulations to see what they look like.

 

[Ofer]

Now I understand what you mean by double up. Yes the point is to double up the base frequency range, in the case above form 30hz-100hz. By doing this you can compensate for nulls in the speakers with the sub(s). When levels and timing (delay) is summed properly, as you said, this will result in positive summation in the bass frequency of choice, 30-100hz in our hypothetical case.
Indeed you can choose different XO slopes and overlap using a shallow slope but that is not what I mean.
"Have you tried this and observed the interaction between subwoofer and mains? Just create a few different filters with various XOs / widths and check the simulations to see what they look like."
Well I haven't yet acquired AL, its a big move and I am trying to gather as much information as possible. Being such a niche product this forum is probably the only place to learn about it and good people like you are what I rely on.

 

[whoareyou]

All I can say is that for me the bass (and rest of system) has never sounded better. This is coming from other potential DSP. I'd say it's mostly because of the flexibility with target / XO / filter configurations that are simply not possible with the other solutions with rest of the improvement is from the TTD with XO version

I think you really just need to try AL. If you already have the equipment required to measure your system, there's really nothing to lose as there is a limited money back guarantee. I think it's a 3 month period but you'll need check with Bernt @juicehifi. In my case, it was really no brainer to try because I had measurement equipment.

From my experience you need the time to experiment with the software.

If you can put up with the outlay of funds with refund guarantee, that is probably your best bet.

 

[Ofer]

Thanks for that. I do have the UMIK1 and bought a RME 8 channel interface. It means re-wiring all the system from the miniDSP based system to a computer based system. It will happen eventually. Did retain Mithch's services when you started? Did you think about my question regarding overlapping the subs and mains?

 

[whoareyou]

Did retain Mithch's services when you started?

No, but it would have helped

Did you think about my question regarding overlapping the subs and mains?

You mean overlap sub and mains with AL?
Pretty sure you can't do this with AL.
If I wasn't using AL, I might try something similar to what you are doing, but the thing is, with AL I don't need it.

I think my room is somewhat difficult with nulls, and I deal with them by combination of the XO's and the filter's correction boost.

The frequency response is pretty much within +/- 3db for most of the target. You can see this response in above attached image showing the simulation for all speakers together with the filtered frequency responses.
I only listen to music with this setup and it is not lacking by any means.

If anything, the bass seems more pronounced and focused. No one frequency is overpowering another frequency, and if you are using TTD, the timing of the frequencies are aligned.

 

[Ofer]

That's the thing. When I look at your actual measurements it seems the mains roll of at about 30hz. On of the subs, I think it's 1, runs form 20-300hz more or less and the other from ~10hz-100hz. They in fact do seem to overlap with the mains. Nothing seems to roll of or start at 50hz as the xo setting imply. I guess you achieved that with the filters, boost and TTD as you said. Interesting. Can you show and explain which filters you used and how you chose them? Thanks for you patience.

 

[whoareyou]

It's not necessarily about where the mains roll off in my room. BTW - Both subs should be rolling off at 150 (look at LFE setting in image above). I'm not sure why both don't show the same high frequency roll off (question for Bernt), but I suspect it is just a programming error.

You can probably best see with the side speaker, that is crossed at higher frequency to eliminate the 60hz null (green). Without XO simulation I'd still have null over there, yet simulation is flat. Then from XO point and down, the two speakers overlap according to the XO width. You should get the idea of how the subwoofer and side speaker are blending together by following the XO lo/hi pass curves. Where one cuts out completely, I couldn't tell you. But does it matter?
Only you can answer that question based on your existing speakers and subwoofer i.e. main speaker may be much better sounding than subwoofer at frequency you are trying to cross... other factors.

I chose the speaker's XO by testing against other possibilities until I had best results per various AL simulations. Some people may not agree with my XO point choice, but that's what seems to work best for my situation.

Also, notice how the side speaker rolls off in low end, yet simulation is flat to 20hz? These speakers (Revel 105's) don't really have great low end output below 40hz or so, but what difference does it make? The subwoofers fill in that bottom end.

No audible distortion and sounds great.

For the main speakers which do have pretty good low frequency response, I XO at 55. I I'm remembering this correctly, I had a couple of nulls to deal with that the 55hz XO corrected. Then from 55hz and down, the subs and main are splitting duty playing the low frequencies, but I'd imagine the main speakers are still doing their fair share for a decent part of the lower range to around 45hz. I'm not sure where the main speakers completely cut out, but it really doesn't matter to me since I elimnated a couple of nulls and extend the bass a bit deeper for those audio tracks that have something to play down in that very low region. And it sounds very good.

 

[kevinzoe]

Hi Whoareyou,
You wrote " Also, anyone have success with measuring a multichannel setup via JRiver / REW /USB DAC / USB mik (umik-1)? "

I'm running JRiver -> exaSound 8channel DAC-> pre+power amp to main speakers and 2 more outputs from the DAC to 2 subwoofers. So, 4 channels in total. I use the Dayton Audio UMM-6 mic with Audiolense, and also own the Dayton Audio OmniMic measurement mic+software.

I was noticing how much filter insertion loss your target curve was experiencing and wonder if there's a way to reduce it, somewhat? Do you use the correction filters to try and align the volumes of all your amps (e.g. gain matching)? I'm assuming you use an amp for your main and side speakers, and separate amps for each sub? I found that if one amp shows a higher correction filter than the other amps then this adds filter insertion loss. If this higher correction is very near a XO point then I'd play around with XO frequency and slope width to see if it'd reduce OR turn up the volume on the amp that is related to the higher correction. The correction filters are like a teeter-totter found at the kid's playground - turning up the vol on one amp causes the other amp's volume to fall because the Y-axis is a relative measurement and not an absolute measurement. Try and get the teeter-totter horizontal, so to speak . . .