DSP Xover project (part 2) - Page 156 - 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 19th October 2013, 06:35 PM   #1551
diyAudio Member
 
Join Date: Jul 2004
Location: home
What is maximum amount of FIR taps/coefficients on 96 kHz for Najda?
For input processing taps maximum is 1023 per input/channel, for output processing Ch1,2,3,5,6,7 taps maximum is 1023, for Ch4,8,9,10 511 (decide by error messages what is get if I try load more taps) but what is total maximum and maximum for DSP Core0, what for Core1 on 96 kHz, on 48 kHz, on 192 kHz?

Last edited by kaameelis; 19th October 2013 at 06:40 PM.
  Reply With Quote
Old 20th October 2013, 07:26 AM   #1552
paalj is offline paalj  Norway
diyAudio Member
 
Join Date: Nov 2005
Location: Norway
Quote:
Originally Posted by chaparK View Post
Thanks all for your feedback.

The next release will feature:
- Enabling/Disabling of processing channels
- FIR/IIR mix
- Channel pairing
- Input processing optionally included in output graph
- Renaming of sources on LCD
- Possibly more

The remaining items will be included in a further release.

On another note, there's the expansion board. The original idea was to add 2 more analogue outputs for those who are into 2x5 ways systems.

My problem is that there's no economically viable solution for this. Indeed, a stereo volume chip that can do the same on 2 channels as the CS3318 on 8 channels, is about as expensive as the CS3318...

I'm thinking of extending the capabilities of the expansion board in order to justify the cost:
- 2 analogue outs with volume
- 2 analogue ins with adjustable input level
- 4 volume controls for external channels.

How does that sound?



Thanks for that, it's surely going to prove useful at some point.



We could do that, of course (provided there's a way to integrate this nicely in the app). In the meantime, you can use the custom biquads for that.



Thanks Bengt
Hi Nick,
I am too late to vote, but this sounds really great!
Evenings are getting darker now, so if you need betatesting I might be able to help ;-)

best,
Paal
  Reply With Quote
Old 20th October 2013, 08:53 AM   #1553
chaparK is offline chaparK  Luxembourg
diyAudio Member
 
Join Date: Apr 2010
Location: Luxembourg
Quote:
Originally Posted by kaameelis View Post
What is maximum amount of FIR taps/coefficients on 96 kHz for Najda?
For input processing taps maximum is 1023 per input/channel, for output processing Ch1,2,3,5,6,7 taps maximum is 1023, for Ch4,8,9,10 511 (decide by error messages what is get if I try load more taps) but what is total maximum and maximum for DSP Core0, what for Core1 on 96 kHz, on 48 kHz, on 192 kHz?
Hi Meelis,

The total number of coefficients is somewhere between 8500 and 9000 at 48 kHz - depending on the board settings.
FIR mode is mainly intended for 48 kHz sampling frequency. 96 kHz and 192 kHz sample rates are also available in FIR, but their purpose is more for experiments than actual use in a system because indeed the resolving power of the coefficients is dropping significantly.

Here's a quick rule of thumb for estimating the number of taps. The DSP is dual-core 250 MHz with 1 tap/clock period. So in 1 second, Najda can in theory process 2 x 250 x 10e6 = 500 x 10e6 taps.

Not all the processing power is available for taps processing because the DSP must also route the signals, manage the delay lines etc. At 48 kHz, the burden of the above tasks is roughly 15 %. So the actual number of taps per second is 0.85 x 500 x 10e6 = 425 x 10e6.

If the sample rate is 48 kHz, then the number of taps per sample period is (425 x 10e6) / (48 x 10e3) = 8854.

By using the above approach, you can compute the number of taps at the other sample rates.

Quote:
Originally Posted by paalj View Post
Hi Nick,
I am too late to vote, but this sounds really great!
Evenings are getting darker now, so if you need betatesting I might be able to help ;-)

best,
Paal
Hey Paal,

Long time no see! Thanks for offering to help with testing!

Nick
__________________
Najda DSP
  Reply With Quote
Old 20th October 2013, 12:43 PM   #1554
diyAudio Member
 
Join Date: Jul 2004
Location: home
Quote:
Originally Posted by chaparK View Post
FIR mode is mainly intended for 48 kHz sampling frequency. 96 kHz and 192 kHz sample rates are also available in FIR, but their purpose is more for experiments than actual use in a system because indeed the resolving power of the coefficients is dropping significantly.
Sad.
Did this mean that best/most effective way of using Najda is IIR mode and biquads as file?
If yes, then use of rePhase for creating taps is not the best choice?
Is existing any tool what can crate biquads and have Parametric Phase EQ as rePhase? Or is existing any way to convert rePhase result to biquads?
  Reply With Quote
Old 21st October 2013, 08:29 AM   #1555
ChrisPa is offline ChrisPa  United Kingdom
diyAudio Member
 
Join Date: Jan 2005
Location: Saddleworth
Quote:
Originally Posted by chaparK View Post
FIR mode is mainly intended for 48 kHz sampling frequency. 96 kHz and 192 kHz sample rates are also available in FIR, but their purpose is more for experiments than actual use in a system because indeed the resolving power of the coefficients is dropping significantly.
Being able to coordinate 2 najdas - using a 2nd najda for expansion - should immediately double its processing capability.

Without any detailed thought I'd guess the functional requirements would be:
- single input one one board fed to both boards
- single point of control for volume
(disabling/ignoring IR control on 'slave' board)
  Reply With Quote
Old 21st October 2013, 10:16 AM   #1556
chaparK is offline chaparK  Luxembourg
diyAudio Member
 
Join Date: Apr 2010
Location: Luxembourg
The problem is a bit more complex. Indeed, at higher sampling rates, not only the DSPs have less time for processing the samples, but also the resolution of the FIR filter is reduced if you keep the same number of taps.
Basically, in order to replicate, at 96 kHz, a filter you originally designed for 48 kHz, you need 4x the processing power. To replicate it at 192 kHz, you would need 16x the processing power.
Doubling the number of boards doubles the processing capability. Unfortunately it doubles also the cost. Is it really worth? What's your expectation by doubling the processing power (i.e. what would you like to implement that you can't currently achieve with a single unit?)
__________________
Najda DSP
  Reply With Quote
Old 22nd October 2013, 01:11 AM   #1557
ChrisPa is offline ChrisPa  United Kingdom
diyAudio Member
 
Join Date: Jan 2005
Location: Saddleworth
Please don't take any of the following comments too seriously or critically. I think Najda is a wonderful bit of design, technically well thought out and packaged providing a balanced design (in terms of features spec and price) and far better than the primary alternative - but I'm comparing with my ultimate 'wish list'

From my point of view I see three limiations with Najda.


24 bits/digital volume

I perceive some advantages in digital volume control, but I think it ideally should involve 32bit maths and a 32bit DAC. Actually I think diigtal attenuation within 24bits is also perfectly acceptable given the right gain stucture.

I'm not really intending this to be the time/place to discuss why, it's a slighlty different philosophy to the way najda is conceived and implemented. I don't think there are absolute rights and wrongs and I totally agree with your philosophy of getting the gain structure right and how the components you've chosen do that perfectly.

This can also be addressed by using external DACs with digital volume, but this involves an additional bit of engineering for me including syncing the digital volume of all the DAC channels.


Balanced outputs

Having gone to balanced I wouldn't go back to single ended, so if I don't go for external DACs I would want balanced outputs. I know because of najda's flexibility I can add balanced buffers but again it means I've got a task in order to build the system I want, and with limited free time at the moment it's delays me pushing the button.


44.1 -> 48k ASRC

I've spent some time owning and playing with an autonomous ASRC unit - Musiland SRC-10. To my ears, my least favourite configuration was going from 44.1 -> 48k; my most favourite configurations were sysnchronous upsampling - ie. 44.1 -> 88.2, 44.1 -> 176.4

I know there will be all sorts of reasons why the behaviour of one ASRC cannot be used to judge the actual performance of another ASRC implmentation, but I am most wary of 44.1->48 conversion. I am less wary of 96k conversion, therefore I would want to use it at 96k or have (say) an external 88.2k clock rather than 48k

Interestingly, the Hypex DLCP deliberately uses a clock frequequncy that is not a synchronous multiple of either 44.1 or 48k


On the other hand

The really positive bits:
- user interface
- DSP flexibility - a wonderful amount of flexibility and more being added all of the time
- performance limited by the processing you ask it to do rather than any other abitrary criteria
- user interface
- mixed FIR/IIR modes
- did I say user interface?
Thank you for all of this

In terms of my short list I'm comparing najda with Hypex DLCP. I want a system with 2 x 2way + multiple subs. If I need additional horsepower for the FIR then for a similar budget I can sensibly consider using 2 najdas (did I say how much I like the flexibility and the user interface). But in order to do that I need to make them behave as one.
  Reply With Quote
Old 22nd October 2013, 03:49 AM   #1558
diyAudio Member
 
Join Date: Apr 2007
Location: Australia
Quote:
Originally Posted by ChrisPa View Post
24 bits/digital volume
...
This can also be addressed by using external DACs with digital volume, but this involves an additional bit of engineering for me including syncing the digital volume of all the DAC channels.
The only digital volume control I've had is the one on a miniDSP 2x4 and in that case an analog volume control (stepped attenuator, post-dac) sounded better than the digital volume control.

Most of your volume control worries could be fixed by adding a Buffalo III DAC to your Nadja. It's an 8ch DAC with 32bit digital volume control that just happens to sound fantastic. If you use the IVY I/V stage, you get balanced (+unbal) output too, so another one of your problems is solved. I'm awaiting delivery on one for myself however my philosophical preference for analog volume control will likely see me pumping the output of the DAC back into the volume control chip of the Nadja and run unbalanced
  Reply With Quote
Old 22nd October 2013, 08:44 AM   #1559
diyAudio Member
 
i_should_coco's Avatar
 
Join Date: Dec 2003
Location: Surrey, UK
Has anyone tried any fancy pants op-amps such as AD797 or OPA627 in Nadja yet?
  Reply With Quote
Old 22nd October 2013, 10:58 AM   #1560
ChrisPa is offline ChrisPa  United Kingdom
diyAudio Member
 
Join Date: Jan 2005
Location: Saddleworth
Quote:
Originally Posted by MellowFellow View Post
The only digital volume control I've had is the one on a miniDSP 2x4 and in that case an analog volume control (stepped attenuator, post-dac) sounded better than the digital volume control.

Most of your volume control worries could be fixed by adding a Buffalo III DAC to your Nadja. It's an 8ch DAC with 32bit digital volume control that just happens to sound fantastic. If you use the IVY I/V stage, you get balanced (+unbal) output too, so another one of your problems is solved. I'm awaiting delivery on one for myself however my philosophical preference for analog volume control will likely see me pumping the output of the DAC back into the volume control chip of the Nadja and run unbalanced
So you haven't tried the digital volume control in the Sabre/Buffalo? You should. You may change your mind

My preamp is a Meridian G02 - fully balanced throughout, including the volume control chips - I've been trying to find out what chip is used within the G02, I'm sure I've seen a reference in hte past but can't find it now. To these ears there's a slight veil that's lifted when the DAC feeds directly to my power amps (nCore 400) rather than through the preamp. Also, not all analogue volume controls are the same - your stepped attenuator is one of the purest implementations.

I'm using a Sabre based DAC at the moment (Audiolab MDAC, balanced out) so I know that that's one of my options. However, I would never use it as 8 discrete channels - I'd be looking for no more than 2 channels per DAC, and therfore I need to externally synchorise volume between several DACs. I've been following AAK's modular DAC development with great interest http://www.diyaudio.com/forums/digit...ontroller.html but as I said, it's another development/implementation task (for me)
  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 05:23 AM.


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