JBL Master Reference Monitor

[pos]

You mention that biquad implementations of IIRs are often imprecise. Do FIRs not suffer from this at all? More to the point, is it safe to assume that rephase predictions will be exactly reflected at the output of an OpenDRC unit?

Hi Synthetic
FIR precision is linear and depends on the number of taps relative to the sampling frequency.
It is also predictable and rePhase will show you exactly what you will get with the red curve, which is basically an FFT of the FIR.
So if the convolution engine does its job correctly (which is the case with the openDRC which implements a very straightforward time based convolution), then that is exactly what you will get on the output.

The rephase settings above have format set to "32 bits floats mono (.txt)". I could be mistaken, but to get miniDSP to load it I think it needs to be changed to "32 bits IEEE-754 mono (.bin)".

Yes, my bad, I used the txt format to export the impulse to HOLM and do the comparison screenshot.

Gotta do the LF part next !

[Synthetic]

Hi Synthetic
FIR precision is linear and depends on the number of taps relative to the sampling frequency.
It is also predictable and rePhase will show you exactly what you will get with the red curve, which is basically an FFT of the FIR.
So if the convolution engine does its job correctly (which is the case with the openDRC which implements a very straightforward time based convolution), then that is exactly what you will get on the output.

Awesome. Thanks for the education. Looking forward to the updated LF.

[pos]

LF part done, HF part slightly modified to compensate for the AD/DA response.
Thanks again grumpy for the measurements!

This should be pretty accurate now.
The relative levels between sections might not be exact, I cannot verify this right now, but it should be within +/-0.1dB

Old settings in red, new ones in green:


The HF section had +/-1dB errors, whereas the LF sections had +/-0.5dB errors.
The way BSS constant Q EQ implementation differs from theoretical one is a bit of a surprise/disappointment.
Shelving Q/slopes are also not what they should be, as can clearly be seen.

HF part:

rePhase settings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LF part:

rePhase settings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=

[pos]

I updated the google doc with these new settings, and added a few other things and a "cloning" section.
I might as well withdraw this document as it is only really usable if you have an original BSS/Crown unit, or a FIR-capable unit... (and as of today that basically means an openDRC)

[DS-21]

I might as well withdraw this document as it is only really usable if you have an original BSS/Crown unit, or a FIR-capable unit... (and as of today that basically means an openDRC)

The new miniDSP 2x4HD is also advertised as FIR-capable. I don't know enough about FIR to know if it has enough taps for this project, but if so it's an inexpensive, small, and quiet one-box solution.

https://www.minidsp.com/products/min...minidsp-2x4-hd

[pos]

Not enough taps for the LF section, so it has to be assisted with a few biquads.
miniDSP uses proportional Q EQs, and you also have the possibility to load your own biquad coefficients.
It should be pretty accurate, but again...

https://www.minidsp.com/forum/opendr...start=45#24625
It would probably be a better idea to convert to proportional Q the modified constant Q values found in the new rePhase settings for those 8 EQ points.

I *hate* IIR crossovers and their noisy/unreliable/capricious biquad implementations
FIR is the future, for sure, and not only for linear-phase crossovers.

[Synthetic]

The high frequencies look better now.

Before

After

[pos]

There is still a dip around 6kHz, but less pronounced indeed.
Do you hear an improvement?

[Synthetic]

There is still a dip around 6kHz, but less pronounced indeed.
Do you hear an improvement?

I think it sounds noticeably better, but it could definitely be a case of "I swapped this interconnect and it's a night and day difference!"

[pos]

new settings in red, old ones in blue.

on-axis ~1m



off-axis ~1m, ~30°



The overall response got better/flatter, but the dip around 6.5kHz is still there.

It is also there in these JBL measurements (in black) but it is almost completely gone in the averaged one (in red) :

[johnhere]

I had a nice conversation with G.T. over the New Years…The latest generation of waveguides/horns are "prefect". They don't pinch at the top end like the "almost perfect" M2 and 4367 do. I would imagine they will be thoroughly "protected" and "patented". I believe the first examples will be coming out in some in-wall systems.

Still waiting for this new waveguide which surpasses M2 & 4367.

[bubbleboy76]

Still waiting for this new waveguide which surpasses M2 & 4367.

Why are you waiting?!
Take a look at the SCL-3 and the SCL-4 in-wall speakers.

[10 Watt Street]

Take a look at the SCL-3 and the SCL-4 in-wall speakers.
Links added.

[johnhere]

My god, my god it's already there, at first I just thought it will be a new separate waveguide which we could buy it to diy (with the d2430k driver)... I'm now kinda depressed, this new waveguide looks so nice and big, and it could compete with the Array vertical waveguide, the 1400 Array vertical waveguide has the best imaging in all JBL lines. But the compression drivers in 1400 is not top of line, my current dream is a brand new deisgned vertical (or the SCL kind) waveguide that suits with the 2430.

[Synthetic]

The LSR 7 series has pattern control out to 20kHz, owing to its use of a 1/2" exit CD with the M2 waveguide technology. What is "perfect[ed]" about the in-walls, which use the same CD as the LSR 7 series? They should have no problem with pattern control at the HF anyway.