This question goes to those who have experience with ANALOG active crossovers. Please don't turn this into a DSP thread.
Can I prototype my crossover in DSP to get the frequencies, Q, etc for a filter and then just build the analog circuit equivalent, or will the phase/delay be different with DSP than with an OP-Amp based filter?
In the end, I want an analog an not DSP crossover, but I also don't want to have to build breadboards for each iteration.
Again, I know there are a ton of DSP superfans on here, and that is not the direction I'm going.
Thanks!
Can I prototype my crossover in DSP to get the frequencies, Q, etc for a filter and then just build the analog circuit equivalent, or will the phase/delay be different with DSP than with an OP-Amp based filter?
In the end, I want an analog an not DSP crossover, but I also don't want to have to build breadboards for each iteration.
Again, I know there are a ton of DSP superfans on here, and that is not the direction I'm going.
Thanks!
After decades with analog active XO (which is a giant improvement over passive XO (passive hardly deserves anybody's serious attention on this forum any more)), went to DSP a dozen yrs ago and love the incredible further powers it provides.
You concept sounds smart and logical. But in using the DSP, are you thinking of prototyping the speakers in-situ or are you going to model in the abstract?
After long hours pleasantly tweaking the knobs on my Behringer DCX2496 to get the best sound out of my gear, I can hardly imagine learning anything of much value just doing the math in the absence of the room.
B.
You concept sounds smart and logical. But in using the DSP, are you thinking of prototyping the speakers in-situ or are you going to model in the abstract?
After long hours pleasantly tweaking the knobs on my Behringer DCX2496 to get the best sound out of my gear, I can hardly imagine learning anything of much value just doing the math in the absence of the room.
B.
Hi, I'm not sure that answered the question. The process I'm proposing is such:
Build Speaker
Use my Clio to measure the speaker.
Load measurements into a DSP
Determine transfer functions and crossover frequency in an environment where I can cahnge things by clicking the mouse.
Take the crossover functions from the DSP and replicate in Analog.
Build Analog Crossover, add balanced input and maybe output.
What I'm asking is if I establish a transfer function in DSP is the result the same as building the same transfer function in circuitry?
thanks!
Build Speaker
Use my Clio to measure the speaker.
Load measurements into a DSP
Determine transfer functions and crossover frequency in an environment where I can cahnge things by clicking the mouse.
Take the crossover functions from the DSP and replicate in Analog.
Build Analog Crossover, add balanced input and maybe output.
What I'm asking is if I establish a transfer function in DSP is the result the same as building the same transfer function in circuitry?
thanks!
While there are certainly things you can do in the digital realm, as long as you use standard crossover filters in your digital prototype, there is no reason your approach can not be successful. Having done both, I simply kept the digital part, and left out the soldering, but that part is up to you.
I did the same thing on my two way BBC style diy speakers. First I simulated various slopes and X-over frequencies with a software plug in. Then I did it in hardware using Opamps in sallen key configuration. Sound is much cleaner and natural compared to the software. My recommendation: Just try it, you will be surprised.
This depends on the DSP implementation.
But an Infinite Impluse Response (IIR) implementation is common because it makes least demands on the hardware and behaves essentially like an analog circuit in respect to phase shift.
There are some subtleties about quantisation and numerical stability that I didn't pay attention to in class but I think in a commercial product they should be below a level that would affect you.
I think that any Q you are likely to need will be easily done with a decent op amp.
Best wishes
David
I guess the OP is willing to make these efforts. Decent analog active designs take these things into consideration normally.
To the original question: I once purchased some DSP evaluation boards from Analog Devices exactly for doing what you want to do. I use them with the Sigma Studio software. It is very convenient to do it like that.
How exactly the DSP response matches that of an analog circuit depends on the frequency however. There is an effect called warping which causes IIR filters to behave differently in terms of amplitude- and phase- response compared to their analog counterparts. The closer your poles and zeroes get to the Nyquist frequency the more pronounced it will be.
Nonlinear relationship of frequencies in bilinear transformation, Assignment Help, IIR Digital Filters
Regards
Charles
Thanks, I think that last post is what I was looking for.
To the last response, I assume that any design will need a range of crossover tools to get the job done. Occasionally the shape of the speaker and the selected drivers makes it easy, but Murphy's law usually wins out in some aspect.
I think what I'm going to do is use DSP to get the X-over Freq and slopes. Once I get that I will redesign with some software aid the active analog design, but I won't necessarily rely on the exact transfer functions of DSP being ideal.
I think the only software tools that can be used for active crossovers are Soundeasy, VitiuxCAD, and Leap(no longer and option)
I personally don't like Soundeasy (but I do own a copy). Has anyone designed a speaker with VitiuxCAD? I don't want to sink a bunch of time in SW that doesn't work.
Does FINE X-over work for active? Seems like the price is even more ridiculous if it can't even do what freeware can..
To the last response, I assume that any design will need a range of crossover tools to get the job done. Occasionally the shape of the speaker and the selected drivers makes it easy, but Murphy's law usually wins out in some aspect.
I think what I'm going to do is use DSP to get the X-over Freq and slopes. Once I get that I will redesign with some software aid the active analog design, but I won't necessarily rely on the exact transfer functions of DSP being ideal.
I think the only software tools that can be used for active crossovers are Soundeasy, VitiuxCAD, and Leap(no longer and option)
I personally don't like Soundeasy (but I do own a copy). Has anyone designed a speaker with VitiuxCAD? I don't want to sink a bunch of time in SW that doesn't work.
Does FINE X-over work for active? Seems like the price is even more ridiculous if it can't even do what freeware can..
I have used my DCX2496 to prototype passive crossovers several times. It's a fairly straight forward process that saves a lot of time and building. Should work just as well for line level active, maybe better.
DSP works for me - but I do understand the desire to retain an all analog signal path.
DSP works for me - but I do understand the desire to retain an all analog signal path.
Thanks Charles, this is the bit I didn't pay attention to in class. Bilinear transform - Wikipedia
As I understand it, a commercial produce should already have factored in the non linear transformation in the software and the IIR filter should have essentially the same (minimum) phase behaviour as an analog filter.
And it's mostly an academic point substantially below the Nyquist frequency, where any reasonable crossover will be.
Hence my comment that it will be "below a level that will affect you" [the OP].
Best wishes
David
How could we forget:
Behringer CX-3400 V2 Super-X Pro Crossover CX3400V2 B&H Photo
That's the obvious solution, an adjustable analog XO. Works like a charm. I am still using my old one in a secondary system. Certainly answers my biggest challenge to OPs problem which is, what do you do if you want to change the hard-wired analog XO next month?
Funny thing, newbies need to take care in using this unit because they couldn't get L and R to match by just eyeballing the tiny continuously variable knobs. But piece of cake for OP to fine-tune the knob settings with REW sweeps.
Or a simple matter to fabricate your own clone.... and skip steps 1, 3, 4, and half of 5. (Or just go DSP with a Behringer DCX2496.)
B.
Behringer CX-3400 V2 Super-X Pro Crossover CX3400V2 B&H Photo
That's the obvious solution, an adjustable analog XO. Works like a charm. I am still using my old one in a secondary system. Certainly answers my biggest challenge to OPs problem which is, what do you do if you want to change the hard-wired analog XO next month?
Funny thing, newbies need to take care in using this unit because they couldn't get L and R to match by just eyeballing the tiny continuously variable knobs. But piece of cake for OP to fine-tune the knob settings with REW sweeps.
Or a simple matter to fabricate your own clone.... and skip steps 1, 3, 4, and half of 5. (Or just go DSP with a Behringer DCX2496.)
B.
Last edited:
why do you need to?
measure the drivers (response, T/S)
pick target curves.
transfer function needed = target/driver
What ctrlx said. The first step is to create .frd files for the transfer functions from your passive crossover design program. WinPCD and XSim can do this directly. I also use Bagby's Excel based PCD - in that case you have to save the transfer functions to a picture format and use a graph tracer program (e.g. FPGraphTracer) to create the .frd files.
That used to be my work stream. Establish slopes, compensations, what is needed, with DSP, then replicate in analog circuitry. I happen to have an example laying on my desk:
It has some other tricks on board as well, but this is the general idea (this a three way 24 dB/oct. LR filter with adjustable bass slope on the low end).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.