DSP Xover project (part 2)

Jump to Latest
Z

zoula

Member
Joined 2006
Hi,

I have a question for those who know both the Najda ans the miniDSP. I need to convert a crossover for the LX521 for the Najda from the miniDSP.
It use some low shelf and high shelf but the "Q" from the miniDSP don't seems to be the same "Q" in the Najda.

In the miniDSP they use a "Q" of 0.5 and the lowest possible value in the Najda is 3.0 The Najda don't seems to use the "Q" the same way as miniDSP. How can I achieve the same crossover with the Najda??

Thank you
 
Z

Zonneschimmel

Member
Joined 2013
I had an all FIR 4-way with crossover at ~100, 300, and 2500 with Najda.

with 1000 taps at 48KHz you should be able to do LR2 at about 100-150Hz. Don't know the lowest crossover point of the LX521.

In any case it was just a suggestion, I'm a big fan of linear phase systems and with Najda you get to try and listen to both. See what works best for you.
 
dahlberg

dahlberg

Member
Joined 2005
dahlbergaudiodesign.com
Zonneschimmel said:
I had an all FIR 4-way with crossover at ~100, 300, and 2500 with Najda.

with 1000 taps at 48KHz you should be able to do LR2 at about 100-150Hz. Don't know the lowest crossover point of the LX521.

In any case it was just a suggestion, I'm a big fan of linear phase systems and with Najda you get to try and listen to both. See what works best for you.
Click to expand...

I have come to the conclusion that 96khz is my prefered samplerate. FIR is interessting and most likely I will end up there as well some day. My three way system is crossed over at 65 and 350hz/2ord, also in need of a subsonic filter at 15hz/4ord. Some dipole compensation for both sub and midbass and preferably also some input eq. I'm having a real hard time finding something that can fix this in FIR. The sub is as I see it not possible at all. The Nano sharc is most likely what comes closest when they get there software fully compatible.

It would be interessting to see your project file for the 4-way, is it something you like to share ? I'm using "RePhase" for fir filters, should be able to open it there I think.
 
P

paalj

Member
Joined 2005
zoula said:
Hi,

I have a question for those who know both the Najda ans the miniDSP. I need to convert a crossover for the LX521 for the Najda from the miniDSP.
It use some low shelf and high shelf but the "Q" from the miniDSP don't seems to be the same "Q" in the Najda.

In the miniDSP they use a "Q" of 0.5 and the lowest possible value in the Najda is 3.0 The Najda don't seems to use the "Q" the same way as miniDSP. How can I achieve the same crossover with the Najda??

Thank you
Click to expand...
Hi,
I have both miniDSP, Najda and LX521. I did a check with Arta on both units and ended up with Najda Q of 7 = miniDSP Q of 0.5 on the shelf's.

/Paal
 
R

Rainair

Member
Joined 2015
FIR-Filters by Zonneschimmel

Hi Zonneschimmel,

could you be so kind to explain your way to the 4-way-FIR Filters for Najda ?

1. Am I right, you made it with Rephase ?
2. Minimum-Phase or Linear-Phase ?
3. Linkwitz-Riley 24 db each ?
4. 1023 taps for Low, 1023 for Lo-Mid, 1023 for Mid, 500 for Hi ?
5. FFT length 16384 or 8192 ?
6. centering middle ?
7. Rate 48 kHz each ?
8. Did you measure the several chassis and did you correct it ? If yes, ARTA / REW ?
9. Did you do some input-correction ?

So, this are many questions, but I think it could be very helpful for many guys here, in particular me !

Thanks in advance,
Rainair
 
Last edited:
F

fluid

Member
Joined 2009
Zonneschimmel said:
Why not use the awesome FIR capabilities of Najda instead of IIR?
Click to expand...

To keep the crossover amplitude and phase response as designed. There is around 20dB of boost in the very low bass as part of the dipole compensation so it would need a lot of taps if realised in FIR. If you wanted to eliminate the phase turn from the crossovers then rephase could achieve the same thing without needing as many taps or introducing as much latency with it's feature to undo the phase turn of specific crossovers rather than creating the entire crossover with FIR filters. Would only work with a PC that can run a convolution plugin or use a minisharc or similar before the Najda though.

Zoula, I have Najda crossover files for LX521.3 and 521.4 if you want them PM me. They are exactly the same slopes as required verified by ARTA output.
 
Last edited:
Z

Zonneschimmel

Member
Joined 2013
Rainair said:
Hi Zonneschimmel,

could you be so kind to explain your way to the 4-way-FIR Filters for Najda ?

1. Am I right, you made it with Rephase ?
2. Minimum-Phase or Linear-Phase ?
3. Linkwitz-Riley 24 db each ?
4. 1023 taps for Low, 1023 for Lo-Mid, 1023 for Mid, 500 for Hi ?
5. FFT length 16384 or 8192 ?
6. centering middle ?
7. Rate 48 kHz each ?
8. Did you measure the several chassis and did you correct it ? If yes, ARTA / REW ?
9. Did you do some input-correction ?

So, this are many questions, but I think it could be very helpful for many guys here, in particular me !

Thanks in advance,
Rainair
Click to expand...

From your questions I strongly suggest you read the very well written Najda under control manual and the rephase documentation (there is also a good writeup at the minidsp site). For more information about my system or further discussion I kindly point you to it's own thread. http://www.diyaudio.com/forums/multi-way/287929-achilles-fir-4-way.html

1) yes
2) linear phase
3) 12dB for 100Hz 24dB for 300 and 2500
4) I used 1023 for lower bass 1023 for upper base 500 for mid and 150 for tweeter
5) I let rephase decide that setting based on number of taps
6) yes
7) This is the sample-rate you set in the DSP, all FIR filters need to have the same sample-rate. If you make filters for 48KHz and load them in a 96KHz DSP they will be shifted by 1 octave :)
8) yes, REW
9) no
 
C

cookiemonster77

Member
Joined 2012
Perhaps this was mentioned in the replies above, but for dipole compensation and general EQ minimum phase is the way to go, as it also corrects phase. Where you want linear phase correction is in the crossover. Take a look at how Grimm Audio did it with their speaker. In basic terms: give every driver a textbook response using IIR, time align to ensure flat combined response (pure delay), and last reverse crossover phase shift using FIR.

Linear phase dipole boost for example, will result in worse phase response than minimum phase. So use FIR where its useful. Of course IIR filters can be made with FIR but its a waste of taps..
 
Y

yoshisoakland

Member
Joined 2010
Ryssen said:
I wonder what gear are you folks using with the analog input?
What do you think of the sound?
Have you changed the opams at the analog in,what opamp did you use?
(I have only so far used it with the XTZ messauring stuff)
Click to expand...

I use the analogue in with an old ProJect turntable. Sounds good to me, but the turntable is mediocre. I have compared an XMOS Dac at the analogue in with SPDIF in using 96kHz Linn test albums, because I am a bit suspicious about the 24kHz filter at the digital inputs (who knows the filter might ring). However I could not identify higher res material in a proper foobar abx tester test also at the analogie in without filter. I also thought the reclocked XMOS signal might be interesting because the lack of a stable clock in Najda (it uses a jittery phase locked loop of the codec) is a weak point - of course the PLL still has jitter after the XMOS DAC. I prefered the digital input of Najda, three DA AD DA conversions seem to detoriate the sound more than any filter advantage or jitter issues.
 
Last edited:
Y

yoshisoakland

Member
Joined 2010
Ryssen said:
Thanks for answer,I am to thinking of using a TT (Riaa)on the analog inputs.
Do you use the LM833 opamps or have you cahanged them?
Click to expand...

I read somewhere in this thread that the op amps are pretty decent (at least the ones on the DAC output) and one has to spend significantly more to improve the sound. All in all I guess that with the DSP most of the tonal changes due to high end cables, op amps etc are just a simple EQ of the house curve /input curve, because the slight filter action of these physical components is very easy to model in Najda. Of course if an op amp is having nonlinear distortion, this cannot be corrected - even though sometime nonlinear (tube) distortion is supposedly very pleasant. Najda cannot model is nonlinear distortion - it might be nice to add nonlinear distortion digitally, but I have only little experience with foobar distortion plugins. therefore I do not have any opinions on the positive effects of distortion (apart from guitar amps), but I guess an audiophile system should have low distortion and the distortion should be dialed in carefully with e.g. a tube buffer.
 
Ryssen

Ryssen

Member
Joined 2003
You must log in or register to reply here.
Top Bottom