DSP Xover project (part 2) - Page 67 - diyAudio
Go Back   Home > Forums > Source & Line > Digital Line Level

Digital Line Level DACs, Digital Crossovers, Equalizers, etc.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 17th March 2013, 05:27 PM   #661
chaparK is offline chaparK  Luxembourg
diyAudio Member
 
Join Date: Apr 2010
Location: Luxembourg
That's very interesting! Now we're getting into the topic!

First, *the* question How did you generate FIR coefficients for simulating a LR24? Would you mind sharing the coef file?

Can you explain the 128-sample delay on ch3?

Last question: could you measure the outcome?

[Edit] Another question: how do the 2 channels sum together?

Last edited by chaparK; 17th March 2013 at 05:41 PM.
  Reply With Quote
Old 17th March 2013, 05:53 PM   #662
paalj is offline paalj  Norway
diyAudio Member
 
Join Date: Nov 2005
Location: Norway
Quote:
Originally Posted by chaparK View Post
That's very interesting! Now we're getting into the topic!

First, *the* question How did you generate FIR coefficients for simulating a LR24? Would you mind sharing the coef file?

Can you explain the 128-sample delay on ch3?

Last question: could you measure the outcome?

[Edit] Another question: how do the 2 channels sum together?
1) I use Acourate to generate the filters. I will send the files to you.
2) Prosessing delay seems to be double for mid compared to tweeter. mid has 512 taps and tweeter only half.
3) See attachment. I had to set the mic up and did not hit the same spot, so the result is not as good as it could have been.

best
Paal
Attached Images
File Type: png Najda fir test.png (34.6 KB, 403 views)
  Reply With Quote
Old 17th March 2013, 06:09 PM   #663
JLOP is offline JLOP  France
diyAudio Member
 
Join Date: Sep 2011
Location: Luxembourg
Nice Paalj
Not as good : you mean eg dips at 90 And 220 ? Did you try to correct them or is it just a choice not to

SpeedySteve
Great system you have, congrats (;-)
Did you try a set up to Check the Najda DAC And output stages ( stereo filter feeding your analog filters) part compared to your former DaC ? I would be glad to know about the sound caracteristics.

BR
Jean-Louis
  Reply With Quote
Old 17th March 2013, 06:29 PM   #664
paalj is offline paalj  Norway
diyAudio Member
 
Join Date: Nov 2005
Location: Norway
Quote:
Originally Posted by JLOP View Post
Nice Paalj
Not as good : you mean eg dips at 90 And 220 ? Did you try to correct them or is it just a choice not to


BR
Jean-Louis
Hi Jean-Louis,
Because I did the first measurement yesterday and removed the mic I did not hit the same measurement spot today as I used yesterday. Therefore this "ragged" response. Anyway, I did no attempt to correct the 90 dip. The driver correction stops at 100Hz. Note that this is just for testing. It is not a good idea to try measuring down to 100Hz indoor but outside is 50cm of snow

best,
Paal
  Reply With Quote
Old 17th March 2013, 06:52 PM   #665
JLOP is offline JLOP  France
diyAudio Member
 
Join Date: Sep 2011
Location: Luxembourg
Snow is very good sound absorber ! You may build an igloo too (;-) as an ephemerous anechoic chamber !

Ok keep us posted ...
  Reply With Quote
Old 17th March 2013, 07:00 PM   #666
paalj is offline paalj  Norway
diyAudio Member
 
Join Date: Nov 2005
Location: Norway
Quote:
Originally Posted by chaparK View Post
That's very interesting! Now we're getting into the topic!


[Edit] Another question: how do the 2 channels sum together?
Did not see this one.
Regardig the sum, see attachment.
Attached Images
File Type: jpg el sum.jpg (268.7 KB, 403 views)
  Reply With Quote
Old 17th March 2013, 07:46 PM   #667
diyAudio Member
 
Speedysteve7's Avatar
 
Join Date: Dec 2009
Location: Hants/Berkshire/Surrey
Quote:
Originally Posted by JLOP View Post
Nice Paalj
SpeedySteve
Great system you have, congrats (;-)
Did you try a set up to Check the Najda DAC And output stages ( stereo filter feeding your analog filters) part compared to your former DaC ? I would be glad to know about the sound caracteristics.

BR
Jean-Louis

Thanks. I do love it. Taken a while to put it together - continually evolving.

Well having taken the caps off the S2's drivers the sound has blossomed.
The Vitavox S2 driver in the range about 1 KHz to when it runs out of puff and ramps off nicely at about 12KHz, on the right horn is an amazingly clear and fast transducer. Capable of such tonal quality too.
It needs to be fed the right quality signal. So much is going on in that freq range. It is also a driver that is doing very nice things lower down the dB scale of the slopes - hence only putting it on BW 6dB/oct - lets it play all it can.

I am at present gobsmacked by how good the Najda piece of kit sounds.
I am not missing the passives - i did not quite think I would be saying that!
Hats of to Nick.

I have only tweaked the gains a bit and chosen slopes that gave a reasonably flat simulation/calc. Not got the measuring mic out yet... That will be next weeks work. Then there is the correction side of things to try. Better PSU? Let alone other OP amps...

What sort of core load is normal. I am at 44 and 38% on 96KHz and doing 8 channels.

I've ordered a IR sensor, and have an All-In-One remote that should do JVC and NEC codes.

Can't stop listening and wondering.

Last edited by Speedysteve7; 17th March 2013 at 07:54 PM.
  Reply With Quote
Old 17th March 2013, 08:58 PM   #668
chaparK is offline chaparK  Luxembourg
diyAudio Member
 
Join Date: Apr 2010
Location: Luxembourg
I'm just coming back quickly (!) to the FIR filters that Paal has designed. Paal sent me his coefs - I hope he won't mind if I post a few screenshots here.

So I've loaded his coefficients on Ch1 and Ch2:

paal1.png

One filter is 512 samples long, the other one is only 256 samples.

The design procedure for FIR filters often yields to impulse responses that extend to infinity in the past ... and in the future, with (luckily) most of the energy being at the time 0 (the impulse response appears here in red). Time 0 is the time of now.

No computer, regardless of how powerful, can compute the output of a infinitely long filter. And no computer can guess what the samples of the future are going to be.
That's right, the theoretic model requires to know what the sound was before you were born.
It also requires to know in advance what the sound is going to be long after you have died.
That's why we need to approximate this infinitely long filter by windowing its impulse response - i.e. keeping only a certain number of coefs and ignore the remaining ones.

So first thing we do, we keep N points of the impulse response symmetrically about time 0. This way we have a filter that expands only approximately N/2 coefs in the past and N/2 coefs in the future.

There's still this problem that the filter needs to know samples of the future. To go around this, we 'shift' the impulse response in the past so that the future becomes the present
The amount of shift is usually half the length of the impulse response, so for a 512-coef filter, the amount of shift is 256 samples. In clear, by shifting the filter so that we can implement it, we effectively introduce a delay of half the length of the filter.

This is why Paal introduced a delay of 128 samples on the tweeter channel. Indeed, the medium filter, by design, induces a delay of 256 samples meanwhile the tweeter filter a delay of only 128 samples. Thus Paal has time-aligned the outputs of both channels.

So here's what the frequency response of both filters is looking like:

paal2.png

As Paal explained, they combine a FIR approximation of a 4th order LR to some EQing.

I asked Paal to show how the channels sum together. Here's what he showed:

paal3.png

I frowned quite a bit when I saw this. There's obviously something wrong with this crossover. Or something wrong with the graph.

The fact is, the Summing feature in the graph utility doesn't take the delay into account. So this plot shows what the sum would be if there had been no additional delay on the tweeter channel. (Rest reassured I will fix that asap).

So I cheated. I opened Paal's tweeter coef file and I inserted 128 'zeros' at the beginning of the file. These zeros act as a delay: indeed, during the first 128 samples, the modified filter ignores the incoming samples (it multiples them by the null coefs).

Finally, I reloaded the file and suppressed Paal's original delay. I had now the medium's 512-coef filter with a new tweeter 384-coef filter equivalent to Paal's original design.

paal4.png

Again, I checked how the channels were summing, and here's how it looks like:

paal5.png

Isn't it looking much better? If there was no EQing, the channels would sum flat.

Conclusion: I think this crossover has a good potential. I'm looking forward to read Paal's comments on his listening session !!

Last edited by chaparK; 17th March 2013 at 09:03 PM.
  Reply With Quote
Old 17th March 2013, 10:05 PM   #669
paalj is offline paalj  Norway
diyAudio Member
 
Join Date: Nov 2005
Location: Norway
Quote:
Originally Posted by Speedysteve7 View Post
Paal,
It's under setting / bass and treble / untick enable.
How exactly do you create the IIR XO filter files. I have not got upto speed on that yet.



.
Thanks Steve
I think the best way to make the IIR filters are to measure the raw response from each driver and import them into Najda under control

Quote:
Originally Posted by chaparK View Post
I'm just coming back quickly (!) to the FIR filters that Paal has designed. Paal sent me his coefs - I hope he won't mind if I post a few screenshots here.


Conclusion: I think this crossover has a good potential. I'm looking forward to read Paal's comments on his listening session !!
Hi Nick,
I don't mind at all

The XO might have good potential. I cannot say the same of the speaker I just trow some drivers together in an outdated box
I am using an Linkwitz Orion. That's an OB with huge requirement of EQ in low frequency range. Guess I have to wait for "Decimation" or similar

best,
Paal
  Reply With Quote
Old 18th March 2013, 06:47 PM   #670
diyAudio Member
 
Join Date: Jan 2010
I have been staring at switches for a long time at Digikey, and it is only slightly less grueling then reading this thread from start to finish. I would have swore this was several hundred pages long. Paid for my board today and was hoping for some input on parts. Switches, unlit or lighted? Power supply, case, etc. Any help would be appreciated.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
DSP Xover project chaparK Digital Line Level 141 3rd July 2011 10:16 AM
Help Please for simple active Xover project dcathro Analog Line Level 4 9th September 2010 05:47 AM
Violet DSP Evolution - an Open Baffle Project cuibono Multi-Way 211 18th May 2010 02:26 AM
Software digital DSP ... Xover/filters/EQ ?? JinMTVT Digital Source 2 8th October 2004 06:02 AM
DSP card & proper xover for heathkit 859A speaker cabinet x. onasis Multi-Way 6 22nd April 2003 07:39 PM


New To Site? Need Help?

All times are GMT. The time now is 01:41 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2