Project M2 DIY Thread

[sguttag]

So, picking up on page 6 when using a QSC Q-SYS CORE as the DSP processor (CORE 110c, in my case).

I've created a Q-SYS Component using the parameters supplied by "pos" (for QSYS by limiting to 4096 coefficients and using an additional filter block on the LF to finish the job) and exporting from rePhase. The only change I made was to raise the gain up to nominally "0" but kept the offset in level between the HF and LF sections.

In my component below, I show an LF and HF delay sections that are bypassed (just there in case alterations will be needed during final tuning (in a home-theatre environment). Likewise, the gain and HPF sections are bypassed. I'm not completely comfortable not putting a HPF below the point where the M2 is really playing but, for now, it is bypassed too.

My concern in looking at the results in QSYS is the phase response on the LF section starting at 400Hz. That just doesn't seem right to me.

Mind you, I don't have my listening room completed yet (home theatre) but the LF seems a bit "tubby" to me. The ports are tuned to 27Hz (using AC voltmeter/resistor method with generator). I'm wondering if I have the internal insulation right. Is there a consensus on what type, how thick and what surfaces should be covered? I have tried to find what people are doing and I've seen pictures were the middle 1/3, where the LF driver is was untreated but the upper/lower thirds are treated and other pictures have shown all surfaces BUT the front (driver mounting) board are treated. I have no doubt that POS has the answer. But with the number of pages in the two threads here that I know about, I'm not finding anything concrete other than a reference to POS recommending a particular type. So any advice there would be appreciated.

Thanks,

-Steve

[bubbleboy76]

POS can hopefully tell you if something is wrong.
You know about the concept "unwrap phase" when looking at phase? Try it, if possible.
Changing the gain in the FIR-filters from POS is not wise, I think. They were a 0dB maximum level, I guess, to avoid digital clipping.

[pos]

Hello Steve,

bubbleboy76 is right about the level: I told you via PM that you could raise it a bit if you were confident you would not exceed 0dB anywhere, but if you go straight to 0dB you are some 8dB above 0dBFS in some frequency ranges.
It all depends on your digital gain structure: if you are handling the volume digitally before (or inside) the processor then you should be good: simply remember that above -8dBFS you at risk of clipping if your signal has energy in the UHF (not very common) and LF (much more common).
IIRC the original Crown preset is at +6.6dB, but the volume is handled before the DAC.
I preferred to play it safe in the rephase presets and aim for 0dB, because many people will control volume after the DAC, or at least after the DSP where clipping might already have occurred.
Once again it all depends on the gain structure and volume control implementation, but one needs to understand the consequences of one choice or another.

Regarding phase behavior on the LF section, as I told you you don't need to worry about it: it is just a consequence of the delay that is part of that FIR.
Here is the long answer:
When loading a FIR the QSC software has no way of telling what the time reference is, so it has to choose an arbitrary one in order to represent phase.
Many software will choose taps/2 because that is the traditional centering for linear-phase (and FIR is traditionally used for linear-phase corrections, for better or worse), others will try to "guess" based on the peak position (HOLM for example), and some will simply use 0 as a time reference as if the FIR was a causal, phase minimum, correction.
(Of course these choices only apply when visualizing phase, and have no consequence on the actual correction itself)
The later is what your QSC software is doing. This works out perfectly for the HF correction as it is minimum-phase with 0 delay.
The LF part is also minimum-phase, but the 270.8us that it embeds (which is an essential part of the settings) implies the phase shifts that you are seeing.
This is pure delay, and normal.

[pos]

To illustrate the above explanation, here is the LF FIR loaded into HOLM:


blue curves are with a time reference set at 0 sample, red ones are with a time reference set at 13 samples (270.8us at 48kHz).
The impulses (lower graph) clearly show the delay difference.
In the upper graph magnitude curves are the same, but phase curves (the dashed ones) do differ: the blue one reflects what you see in the QSC software, whereas the red one reflects the theoretical minimum-phase correction that you see in the google doc.
Both represent the exact same correction and will result in the same transfer function.

Just for the fun of it, here are the same curves with a linear frequency scale:

Delays are materialized by a linear phase curve.


So, back to the short answer: don't worry about it

[sguttag]

First off, thanks for the long answer! I do hope you know how much I (and I would have to presume others) appreciate the work you put into this.

One reason I also posted where I am publicly is so any other Q-SYS based systems get to set the net results (the previous posting didn't show the phase relationship).

On the 270.8us delay. This is to match the BSS/Crown delay on the HF? Could this not be done in the QSYS DSP or is it better than it is done in the FIR filter to precisely match the BSS setting and remove the DSP variances out of it?

As for gain/volume, that will be done in the Q-SYS up-stream from the Crossover section so if it gets loud in the filter, it will be quite loud in the room. Each speaker will have a maximum SPL of 105dBc and reference will be carefully set. In my current testing sending in a -20dBFS signal outputs approximately -20dBFS (input/output meters for each section, set to peak reading track with each other. This is true for both normal signal (music) and test signals (pink noise).

How about the insulation within the cabinet? Recommended surfaces to cover and how much?

Thanks again,

-Steve

[pos]

I merely duplicated JBL's original settings, that's all

On the 270.8us delay. This is to match the BSS/Crown delay on the HF? Could this not be done in the QSYS DSP or is it better than it is done in the FIR filter to precisely match the BSS setting and remove the DSP variances out of it?

This is exactly 13 samples at 48kHz, and can be done in the FIR or anywhere else in the processing chain. Just set centering to 0 in rephase and add 13 samples of delay in the QSC and you will get the same result.

How about the insulation within the cabinet? Recommended surfaces to cover and how much?

I can help for the DSP correction, but not too much for the cabinet construction I am afraid
I am currently struggling to get a pair of decent clones myself, to replace the hastily hacked together cabinets presently sitting in my living room

[sguttag]

See now...my clones are supposed to be "hastily hacked together"! Well maybe not that. They are supposed to be no-frills versions. No bling, no fancy veneers, no badge...just sound. With luck, I won't see them for years (until the screen is replaced).

The one we call the "prototype" was about a 4.5-hour cabinet to test some concepts, make sure we would have success before banging out 5 "real" ones. For instance, a change between the prototype is that the front plate is designed to be removed to allow access to the D2...in the real ones, that will be changed so that a panel may be removed from the rear to allow access to the driver, if it ever needs it (like the L250, if you are familiar with that speaker).

Things I don't get to do with just one speaker is listen to things like stereo imaging and the like. However, this does let me focus on the detail of various recordings. I've been very impressed with how well the D2 handles being "pushed." That is, if I raise any treble...that is it, it just produces more high end but none of that nasty metallic strain I'm used to from most compression drivers (or ribbon drivers when they get "pushed.") So far, my only negative is in one room where the bass seems a bit tubby (and it measures so on an RTA) but I'm blaming the room for that. I'm going to do some close mic tests on it and maybe see if I can give it a better test listening area. I'm rather sure it will do well for my needs and I'll give them gobs of power to play with.

[sguttag]

I was able to do some more listening tests this weekend and I tried both forms of delay...in rePhase and done in the Q-SYS and, as predicted by pos, I detected ZERO difference in sound. The RTA measured zero difference too.

I'm electing to go with the delay in the Q-SYS and also to put any attenuation in the Q-SYS. So the LF section was set to 0.0 in rePhase and the HF section was -0.6dB in rePhase. I can see on valid reason to make the crossover block in Q-SYS inherently attenuate the signal. I turned it VERY loud today and was never in danger of clipping the signal, digitally in the DSP and had a little more headroom in the DSP than the amplifier did though things seem pretty well matched there.

FWIW, I'm finding that the LF section of the M2 to have a 93dB 1W/1m sensitivity using C-weighted PN. JBL claims 92dB. My room might be accounting for the extra dB or JBL was being conservative. Adding in the HF section adds another 2dB to the overall SPL.

[lesliew]

Its good to see someone else using the Q-sys core, also looks like I have a bit of reading to catch up on.

I'll be starting a new build on 2 more M2's very soon.

FWIW the closest I could get to identifying the insulation used in the M2 is insulation known as duct liner. However I ended up using denim insulation purchased from HomeDepot since I couldn't find anywhere in Florida who sold the duct liner at a reasonable cost. There is a photo showing the inside of a real M2 in one of the threads around here and it looked to me as though both sides and the back where insulated top to bottom.

[lesliew]

Can someone please confirm the thickness of the additional baffle board?
I think it is 16mm, but I also read 18mm.

Pos

My curved baffle certainly ended up being about 16mm. Originally started with 3/4" mdf (19mm) and by the time the CNC machined the curve the highest point was right around 16mm

[sguttag]

Duct liner is pretty commonly used in cinema behind the screen. Manville Linacoustic is the stuff.

https://www.jm.com/en/hvac/duct-liner/linacoustic-rc/

It is good stuff to know about for actual duct lining too, which can be key to keeping HVAC noise down (its original design) as well as keeping the HVAC ducts from being the audio conduit between rooms (or cutting it way down).

My five M2s are now "in prouction."

So you are building two more? Adding another listening room?

Thanks for the info.

-Steve

[pos]

My curved baffle certainly ended up being about 16mm. Originally started with 3/4" mdf (19mm) and by the time the CNC machined the curve the highest point was right around 16mm

Hi Leslie,

You lost me here: I thought that baffle was flat
Are we talking about the same baffle? I am referring to the additional baffle plate that comes around the woofer frame and under the waveguide lip.

[sguttag]

What is everyone doing about a gasket for the 2216nd? The exploded view doesn't show a rear gasket and I can't believe that the front rubber gasket is expected to really seal up between the cabinet and driver.

I plan to use the JBL 360-70002-01 rear gasket that is used on other differential drive 15" woofers.

-Steve

[lesliew]

Hi Leslie,

You lost me here: I thought that baffle was flat
Are we talking about the same baffle? I am referring to the additional baffle plate that comes around the woofer frame and under the waveguide lip.

My mistake, thought you where talking about the curved section.

Pretty sure that flat piece was also 16mm

[lesliew]

What is everyone doing about a gasket for the 2216nd? The exploded view doesn't show a rear gasket and I can't believe that the front rubber gasket is expected to really seal up between the cabinet and driver.

I plan to use the JBL 360-70002-01 rear gasket that is used on other differential drive 15" woofers.

-Steve

I used Parts Express Gasket tape, also used this to stop any leakage from around the Waveguide