• These commercial threads are for private transactions. diyAudio.com provides these forums for the convenience of our members, but makes no warranty nor assumes any responsibility. We do not vet any members, use of this facility is at your own risk. Customers can post any issues in those threads as long as it is done in a civil manner. All diyAudio rules about conduct apply and will be enforced.

Hi-end DSP based multi-channel integrated Preamp/Crossover/DAC project

Hi Tranquility Bass

How are you planning on implementing the decimation, more precisely the antialiasing filter?
Have you read the 2007 AES paper from four audio? They describe the decimation process used in the HD2, which is very clever but requires compensation in the filter target response (shallow low pass using an Albrecht window). It requires fewer taps and has less potentially adverts effects than a brickwall filter.
I am currently (in the process of...) implementing this compensation in rephase, so if you use the same low pass filters that would make it easier to do something a little more generic.

Thanks for that info.

One way to do it is to first apply your low pass IIR crossover filters which run at 192K before you decimate downwards for the proceeding FIR filters so in that way you get a free lunch as far as the anti-aliasing filters are concerned.

regards
 
Thanks for that info.

One way to do it is to first apply your low pass IIR crossover filters which run at 192K before you decimate downwards for the proceeding FIR filters so in that way you get a free lunch as far as the anti-aliasing filters are concerned.

regards

Alternatively, one can apply an IIR to woofer and apply FIRs to the corresponding high-pass part. This will ensure phase alignments throughout different crossover;

Woofer: IIR low pass at freq 1
Mid low: IIR high pass at freq 1 + {FIR low pass at freq 2}
Mid high: IIR high pass at freq 1 + {FIR high pass at freq 2 + FIR low pass at freq 3}
Tweeter: IIR high pass at freq 1 + {FIR high pass at freq 2 + FIR high pass at freq 3}

Multiple FIRs can be combined into one FIR, which denotes as {} above. Woofer output is close to symmetric enough. I don't see much difference between FIR and IIR. Above approach can cut down number of FIR taps to reasonable level.

This is why I'd like to have combination of IIR and FIR.
 
One way to do it is to first apply your low pass IIR crossover filters which run at 192K before you decimate downwards for the proceeding FIR filters so in that way you get a free lunch as far as the anti-aliasing filters are concerned.

You need either a steep slope or a low corner frequency in order to get a >100dB rejection at the target Nyquist frequency. In IIR that means large phase shifts in the passband that will need to be compensated for together with the magnitude.
You can ignore the magnitude compensation if you know the final low pass will get you enough attenuation there, but you cannot ignore those massive phase shifts. That means FIR generation tools will *have* to handle this compensation in order to generate filters for your module.

I imagine you could also simply use some brickwall ASRC if you have these available (I think the miniSHARC has them for example, albeit they do not use them at all).

I am still not sure how (where) to implement this compensation mechanism in rephase. This is more a problem of UI and logic for the user that a technical issue, but of course it would help and make things easier/clearer if all decimation solutions out there used the same low pass schema.

I think Thaden's way of doing this (Albrecht windows used as low pass) is as good as it gets. I can send you the paper if you want it, as well as the required Albrecht implementation.
 
Last edited:
You need either a steep slope or a low corner frequency in order to get a >100dB rejection at the target Nyquist frequency. In IIR that means large phase shifts in the passband that will need to be compensated for together with the magnitude.
You can ignore the magnitude compensation if you know the final low pass will get you enough attenuation there, but you cannot ignore those massive phase shifts. That means FIR generation tools will *have* to handle this compensation in order to generate filters for your module.

I imagine you could also simply use some brickwall ASRC if you have these available (I think the miniSHARC has them for example, albeit they do not use them at all).

I am still not sure how (where) to implement this compensation mechanism in rephase. This is more a problem of UI and logic for the user that a technical issue, but of course it would help and make things easier/clearer if all decimation solutions out there used the same low pass schema.

I think Thaden's way of doing this (Albrecht windows used as low pass) is as good as it gets. I can send you the paper if you want it, as well as the required Albrecht implementation.

By all means please do send me the paper. ;)

If you have a 4th order low pass filter centered at 100Hz you are already down 80dB at 1KHz and 160dB at 10KHz. You could afford to decimate down to an effective sampling rate of 24KHz without any adverse affects. That's 12.5% of the original DSP work load which makes any proceeding FIR filter easier to implement ;)

cheers
 
By all means please do send me the paper. ;)
PM send

If you have a 4th order low pass filter centered at 100Hz you are already down 80dB at 1KHz and 160dB at 10KHz. You could afford to decimate down to an effective sampling rate of 24KHz without any adverse affects. That's 12.5% of the original DSP work load which makes any proceeding FIR filter easier to implement ;)
A 24kHz sampling rate with a more aggressive and linear-phase filter would let you use a larger frequency range.
And for the <100Hz range you could as well use a much higher decimation factor.

It probably does not get any better than the decimation strategy used in the HD2.
 
I am overseas at the moment but I have found some time to finish off some board designs that I brought with me on the laptop ;)

Before I left I got the Asynchronous Sample Rate converter going and feeding it from the Toslink output of my Creative Sound Blaster external sound card on my PC. Was going to post some pictures but I ran out of time.

cheers

David,

Any better estimates of the timeline when the HW+SW product is ready for shipping out to buyers?

thanks
Joji
 
David,

Any better estimates of the timeline when the HW+SW product is ready for shipping out to buyers?

thanks
Joji

When I get back from overseas I will start work on the dsp firmware and PC interface software. I have some ideas for this but this will take some time to get right. The hardware will probably be finished before the software.

cheers
 
Hi,

Instead of reading all the latest pages I'll just ask(hope you don't mind). Will this unit have digital outputs (of some description) for each channel (or some pads where we can get the required signals)as I'd like to feed my already present R-2R dacs which are built into my 4 amps. Say yes, lol.

Thanks,
 
Hi,

Instead of reading all the latest pages I'll just ask(hope you don't mind). Will this unit have digital outputs (of some description) for each channel (or some pads where we can get the required signals)as I'd like to feed my already present R-2R dacs which are built into my 4 amps. Say yes, lol.

Thanks,

Yes it's possible with another board revision without the onboard DAC's. Something to look at in the future if all goes well with the current board
although I am not keen on a DSP environment that adjusts filter coefficients on the fly every time the incoming sample rate changes. That could become really messy depending on the implementation of the dsp signal processing.

cheers
 
Last edited:
Hi TB,

I hope the next revision isn't too far away, lol, as I'd love to try it out when I get my third set of amps built.

I'm not sure what you're saying in your second part of your answer. I guess I just assume(d) that most of these DSP FIR filters are adjustable on the fly for crossover point and filter steepness(db). What does that have to do with have digital outputs and no onboard dacs? Am I missing something here in the discussion?

I guess they don't really need these to be adjust on the fly.

Thanks and keep up the good work.
 
Last edited:
Hi TB,

I hope the next revision isn't too far away, lol, as I'd love to try it out when I get my third set of amps built.

I'm not sure what you're saying in your second part of your answer. I guess I just assume(d) that most of these DSP FIR filters are adjustable on the fly for crossover point and filter steepness(db). What does that have to do with have digital outputs and no onboard dacs? Am I missing something here in the discussion?

I guess they don't really need these to be adjust on the fly.

Thanks and keep up the good work.

I assumed you were talking about the Soekris R-2R DAC which has its own re-clocker and expects to be fed with data in its own native sample rate and not a constant sample rate from a sample rate converter - although it will still work with the latter.

cheers
 
Last edited: