Ground Sound- Interesting active filters?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I wonder too............

Hello all.:D

We are still developing a new crossover, but things have changed a bit/byte.:rolleyes: The working title of the crossover is DCN28DAV, which indicates 2 input channels, 8 output channels, digital in and analogue in and volume control. I think we are nearly there, but we still have some issues.:xeye:

In the meantime you can read a review of our Chassis 1500 in the next issue of Hobby HiFi from Germany and a review of our power amplifier module PA3CC in Elektuur/Elektor September issue.:)
 
Input & output

Will it have single ended analogue inputs?

The analogue inputs will be balanced. If you want to connect a single ended signal you just short the inverting input to ground or you can make a balanced cable with a small 100R resistor in the RCA connector at the pre/processor end. The 100R resistor emulates the output impedance of the pre/processor and has to be connected between inverted wire and ground. I call it “pseudo” balancing and this way the differential input will balance out the noise from the cable just as an active version would.

The digital inputs options will be 2 S/PDIF and 1 balanced AES. We might make a possibility of optional optical TOS link.

The outputs (only analogue) will be "pseudo” balanced.
 
It is almost freely configurable with a GUI running on your PC. :cool:

One of the great advantages of digital x-overs is the possibility to achieve transient-perfect crossovers. A nice topology to achieve that would be the subtractive-delay type (After Lipshitz & Vanderkooy, nice spreadsheet by John K available).

Robert: When are you going to implement this topology ?

Regards

Charles
 
IIR - FIR filters

The processor will be capable of doing both IIR and FIR. I’m not sure how the configuration will be as FIR filters calculations demands roughly 20 times more that IIR does of processing power. If FIR is used one will have to settle for less eq points etc. I would say that one has to focus on the weakest point in the system and I hope everyone can agree that it’s the drivers/loudspeakers. Focusing at the speakers will dictate measurements of the system and not to forget the room too. The software we provide for the DCN23 and the new crossover is supporting measurements import, which we find crucial for the system performance. We don’t consider using only FIR filter as the best solution. A blend of IIR and FIR will give some advantages.

The subtractive delay type filter presented in an AES paper from 1981 is a bit obsolete theory from my point of view. It focuses on the filter response and function, which doesn’t make much sense when considering that ex. tweeter and bass doesn’t have the same impulse response and also mounting will influence on the result. It is beautiful is theory, but lack performance in the real world. In the 1980ies I actually implemented a system with the subtractive delay type filter design, but it did not outperform a “normal” filter design at all – of course it was analogue.
I will still claim that the key to a good system performance will be to focus on optimization of the speakers from good measurements.

:) :( :eek: :D ;) :xeye: :cool: :bawling: :hot: :apathic: :dead: :devilr:

Sorry - I just had to ad some smilies:D
 
It focuses on the filter response and function, which doesn’t make much sense when considering that ex. tweeter and bass doesn’t have the same impulse response and also mounting will influence on the result. It is beautiful is theory, but lack performance in the real world.

But that accounts for any other filter type as well. If implemented with a "real" delay (i.e. not an allpass) - this filter as such is veeeeeeeeeery hard to beat. And the differing responses of the tweeter and woofer are NOT A PROBLEM but a TASK (sorry for this manager babble :D ).

I could manage to take driver response into consideration with the classic subtractive-type crossover of which also all theorists claim that they don't work in practice.

And the subtractive-delay topology as such would use even less computing power than seperate IIR highpass and lowpass (it would need some more memory of course and soem more EQ-ing).


Regards

Charles
 
Okay Charles - I'm making a crossover it will be able to do IIR and maybe FIR filtering. Based upon the DCN23 sound quality - I'm quiet satisfied with IIR filtering and what it will do and the DSP platform only allows IIR and FIR. The big difference I experience is when measurements are employed in the XOverWizard and the difficult part often is to make good measurements. That is why I say that focus should be on the speakers - not the filter technique.

Robert: When are you going to implement this topology?
Maybe a later day - but today I will focus on the measurement import/optimization, which I think is the key to success.
Why don't you put a system/software to the marked? Why is everyone not using the subtractive- delay filter technique? Maybe because:
it would need some more memory of course and some more EQ-ing
and may I ad - latency in the system, which might not be acceptable, just like FIR ads latency to the system. Latency gets bigger as the FIR filter has to process lower and lower frequencies. Again that’s why I’d say a blend of IIR and FIR will be better.

There is progress – the near future will be very busy for me. Very exciting.
 
How large will the FIR filters be? With just a few tabs, it will be more ore less useless. You'll need propper bandpass ripple, and a decent stopband to make them really worth while. That means long filters and lots of processing power..

To bad that the outputs of the new module are not truly balanced :(
 
How many fliter tabs will be available? And will you be able to choose your own coefficients?

About the amp: I can always connect then semi-symmetric, but that would not be what I want. The DAC's have symmetric output, why go to the trouble of converting to single ended ;)
 
Mark, I see - the ”contact” link – it’s just to fancy. I have fixed the contact page last night. ;)

You could make a PSU of your own at 110V~ for the DCN23 module. It’s a bit difficult to get 115V~ transformers in the right housing here in Europe without buying very large stock, but I’ll see what is possible.:(

The future DCN28DAV will be possible to get for 115V~ mains. We still have some issues to attend to before we can release DCN28DAV, but we are getting closer…:)
 
Hi Robert,

If you want to send me a private email with the specifications necessary for a US sourced transformer, that would be great. You'll also notice that I sent a PM to active@... with a couple of questions.

It appears that JustMLS with an E-Mu 0202 (I have an Earthworks M-30 mic) would be a good combination for measurements. Any comments on those two products-- or do you have a better choice for a reasonably priced sound card?

Thanks.

Mark
 
Hi Mark,

You'll get a PM later. ;)

I have no experience of JustMLS or E-Mu 0202 or Earthworks M-30 mic. The only thing I can tell is that some measuring systems have problems measuring the phase right, because they aren’t using a reference signal or people who are doing the measurement haven’t enabled this feature. I only know for certain that MLLSA and LspLAB work, if used the right way. :(
 
I just downloaded the software for the DCN-23, and I could easily import my old JustMLS data aswell the measurement data of the Praxis measurement system. Measurements of both systems also include phase data.

So I can voice for the suitability of the program to import real world data from Libery Instruments Praxis and the JustMLS program (incorporated with LspCAD).
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.