Analog crossovers, only using DSP to treat issues the analog system cannot fix

[john61ct]

If I have Main L/R comprised of a mix

2- 3-way speakers + midbass couplers + stereo trueSubs all co-located

Those subs easily switched to mono if that turns out to test better, there are also other trueSubs already in mono thus freely placed, can be treated as a single sub-group in the DSP used for taming the room modes and other room compensation EQ

The crossovers / bandpassing being ANALOG, 4th order Linkwitz-Riley and everything sounds great, REW shows smooth summing / blending on the frequency side


What if I then decide to use DSP to specifically do per-speaker compensation EQ, based on as anechoic measurements as possible

say to the midbass speakers - not lowering their frequency (which is analog - bandpassed anyway) but flattening or targeting a curve within specific ranges?

Or, if I see with impulse testing, that the subs have timing delays at LP.

Changing the physical placement only goes so far, makes the living room awkward and poor WAF.

I want to keep the existing crossovers, is it possible to use DSP just to successfully "spot treat", handle these specific finer point issues, treating the analog system "as given"?

All constructive feedback welcome, but please abstain from advocating a particular proprietary-box "all in one solution", I'm looking to use PC / RaspberryPi based platforms

and committed for testing purposes to start off using DSP in a minimalistic way.

keyword "Modularised audio DSP"

@Keith_W

 

[voodooless]

The question is: why?

An not, it won’t work very well. Your analog filters are designed for the speaker as is, warts and all. If you remove the warts, the filter aren’t a good match anymore.

There are several reasons to combine analog and active crossover components, but it has to be done properly as a holistic solution.

 

[john61ct]

Yes that is exactly what I strive to arrive at, through incremental testing.

Including various analog crossovers, nothing is "designed for" any particular speakers,

except for the passive networks internal to the 2- 3-way speakers, which I am not messing with, treating as a single unit per channel, not bi- tri- amping

 

[voodooless]

Okay, so you are willing to tweak the analog crossovers?

Then still, the question remains: why?

Maybe give us a bit more information on the system, it’s components and what you’d like to achieve?

 

[kemmler3D]

I want to keep the existing crossovers, is it possible to use DSP just to successfully "spot treat", handle these specific finer point issues, treating the analog system "as given"?

As I'm reading this, if I understand the goal correctly, it's definitely possible if you'd like to correct deviations from the target curve that are left over after implementing an analog crossover. This is basically the idea behind "anechoic EQ" that comes with some JBL speakers, for instance.

The key is to make sure your measurements are really as anechoic as possible. I think the way to do it is take gated measurements with as much empty space around the speaker as you can get.

 

[john61ct]

Yes, anything short of a raising the enclosure with a forklift out in the back paddock

 

[kemmler3D]

Yes, anything short of a raising the enclosure with a forklift out in the back paddock

I think that would actually be pretty good, except for the forklift itself causing some reflections... see how they measured the Purifi speaker:

 

[john61ct]

Okay, so you are willing to tweak the analog crossovers?

Well they generally aren't very tweakable other than frequency selection.

More like have the DSP treat them as given, both the external active units, and the passive networks internal to the 2- 3-way speakers.

Overall I want to implement the system as described in OP, plus a front Center channel (probably two enclosures)

and a rear surround pair that will likely be left plain.

But really the short term priority is not just that system, more so my learning, hands-on testing - especially the DSP aspects to the extent they are needed

so very few of the components are settled.

 

[voodooless]

Okay, so let me get this straight, you have:

- subs, I suspect they are already active?
- midbass enclosures with a passive band pass?
- 2 or 3 way speakers with passive sub.

What is the amp arrangement here?

Is you have separate amps for speakers and midbass, I think it would make sense to make the midbass band pass and speaker high-pass active. This will give more flexibility, and performance will probably be better. If speaker and midbass use one amp, just EQ the whole.

You seems to have a surround system. Any half decent AVR should already offer ways to linearize speakers and sub responses. No need to buy anything extra.

Well they generally aren't very tweakable other than frequency selection.

There is a lot more that is possible with passive crossovers.

 

[john61ct]

combine analog and active crossover components

Note the relevent distinction is DSP-based or not

the analog crossovers I plan to test are primarily electronic, hence active

but I might try some passive units also, apparently the Harrison Labs stuff is pretty clean

for example if used to HPF the 2- 3-ways, treated as a fixed given, a DSP might be useful to "perfectly blend" the transition to the LPF of the midbass coupler units just below.

 

[voodooless]

Why use analog active crossovers and then add DSP? Why not use the DSP in the first place? Skip the analog active crossovers

 

[john61ct]

Okay, so let me get this straight, you have:

- subs, I suspect they are already active?

They will be actively bandpassed

None of my speakers nor subs will have a built-in power amp, so passive in that sense.

All the LF units only one driver per enclosure.

 

[john61ct]

Why use analog active crossovers and then add DSP? Why not use the DSP in the first place? Skip the analog active crossovers

Yes that is the modern mainstream protocol.

I want to begin with the tech from decades ago, and learn DSP by surgical modular application only where it is required.

Then comparing to your full-DSP approach.

As I stated, the testing / learning process is the primary goal, not any one resulting implementatiin.

 

[john61ct]

midbass enclosures with a passive band pass

No, active / electronic for sure, the testing will be between analog vs DSP based

 

[Oddball]

Yes that is the modern mainstream protocol.

I want to begin with the tech from decades ago, and learn DSP by surgical modular application only where it is required.

Then comparing to your full-DSP approach.

As I stated, the testing / learning process is the primary goal, not any one resulting implementatiin.

Oh well that's cool. Good luck with the the journey. That's like you would try to compare original iPhone to the current one. Not a fair game at all.

My suggestion is get the best DSP you can afford, and if there is still budget left take a trip to some destination you always wanted, but have never been to. Time better spent on a trip than on comparison that is bound to fail - in favor of the modern top range DSPs like Dirac ART or Trinnov Optimizer.

 

[john61ct]

What is the amp arrangement here?

The mono enclosures mostly get their own power amp. Some stereo pairs share one.

 

[john61ct]

Oh well that's cool. Good luck with the the journey. That's like you would try to compare original iPhone to the current one. Not a fair game at all.

My suggestion is get the best DSP you can afford, and if there is still budget left take a trip to some destination you always wanted, but have never been to. Time better spent on a trip than on comparison that is bound to fail - in favor of the modern top range DSPs like Dirac ART or Trinnov Optimizer.

None of that is desired, even if within my budget, see the end of my OP.

Even a dedicated PC convolver only if necessary, strategically placed RaspberryPi units more likely.

Ideally as much FOSS as possible.

 

[john61ct]

You seems to have a surround system.

Yes but the rear pair will be optional, depending on the music content's compatiblity with the processing.

For film, sure but that's pretty rare compared to music.

No decoding of proprietary "standards", see XKCD #927.

 

[voodooless]

Note that convolution isn’t ideal for video applications due to inherent delays, especially on high tap count. A Pi can be used if it will also do video. If you need to capture -> process -> output sound, it will also add quite a bit of propagation delay. Not ideal for video.

If the goal is only audio, this is obviously less of an issue.

I feel we’re being fed ever more titbits of information. Still the full picture is far from clear.

 

[Oddball]

None of that is desired, even if within my budget, see the end of my OP.

Even a dedicated PC convolver only if necessary, strategically placed RaspberryPi units more likely.

Ideally as much FOSS as possible.

I think that is the problem - we don't know what you desire really and what your budget is. Or really what you system is? So pretty much nothing after all these posts.

But my recommendation still stands even the waters are muddy as they can get.