or Connect
AVS › AVS Forum › Audio › DIY Speakers and Subs › Najda DSP board for loudspeaker applications
New Posts  All Forums:Forum Nav:

Najda DSP board for loudspeaker applications

post #1 of 43
Thread Starter 
Good day all,

My name is Nick and I'm an independent audio developer. I've just released a new DSP board designed with the DIY tweaker in mind.



This board has following characteristics:
- Analogue input
- SPDIF inputs (coax/optical)
- I2S input (suits nicely USB to I2S converter for example)
- 8 unbalanced analogue outs with analogue volume
- 10 SPDIF outs (coax, transformer isolated)
- 6 I2S outs
- Dual-core 250 MHz DSP
- IR receiver/keypad/LCD/leds
- Isolated command for relays
- Control application via USB
- Sampling frequency can be 48, 96 or 192 kHz

The application lets you load your own measurement, set filtering parameters in real-time and currently supports gains/delays/IIR and FIR filters.

Implemented IIR filters are:
- Butterworth 6/12/18/24/36/48 dB/oct
- Bessel 12/18/24/36/48 dB/oct
- Linkwitz-Riley 12/24/36/48 dB/oct
- Peaking and shelving parametric EQs

For FIR filtering, one must load raw coefficients.

More information is available on my website Najda DSP. I'm the one who developed the product so if you have a technical question, I should be able to come with an answer.

Warm regards,

Nick
post #2 of 43
Very interesting. I have a few questions:

1. How is analog volume controlled? Chip attenuation?
2. How many FIR taps per input channel? Can I implement FIR filters on the inputs and also IIR's on the outputs?
3. Which DSP chip is used?
4. Which ADC and DAC chips are used?
5. What are the voltage input and output ranges?
post #3 of 43
I would be interested if it had 8-10x XLR outs...can this be done? Also can it be supplied with a chasis/body?

I really like the USB touch, the RS232 on the DCX can be problematic with dropped connections.

Can it take a SPDIF input and output that signal untouched with no signal degradation to the all the SPDIF outs simultaneously?
post #4 of 43
Thread Starter 
Quote:
Originally Posted by coctostan View Post

Very interesting. I have a few questions:

Thanks. Precise questions deserve precise answers so here we go.
Quote:
1. How is analog volume controlled? Chip attenuation?

CS3318 with separate voltage regulators.
Quote:
2. How many FIR taps per input channel? Can I implement FIR filters on the inputs and also IIR's on the outputs?

Over 8500 taps at 48 kHz sample rate freely distributed among channels (i.e. resources that you don't use on one channel can be reported onto other channels). A decimation stage for increased resolution is in the pipeline (Najda is fully upgradeable).

A mix IIR/FIR mode has been requested by customers and will be made available soon.
Quote:
3. Which DSP chip is used?

DSPB56725
Quote:
4. Which ADC and DAC chips are used?

CS42428
Quote:
5. What are the voltage input and output ranges?

Analogue input: 1Vrms (fixed)
Analogue outs: 1 to 6 Vrms adjustable on each channel.

Nick
post #5 of 43
Thread Starter 
Quote:
Originally Posted by Decadent_Spectre View Post

I would be interested if it had 8-10x XLR outs...can this be done?

Unfortunately no. You can hook your own DACs to the SPDIF outs and get this way outputs on XLR - but the product as it is has only unbalanced analogue outs on RCA jacks.
Quote:
Also can it be supplied with a chasis/body?

No it's really a DIY product. The user has to source his own LCD, power supply, chassis etc and build a realization to his liking.
Quote:
I really like the USB touch, the RS232 on the DCX can be problematic with dropped connections.

Can it take a SPDIF input and output that signal untouched with no signal degradation to the all the SPDIF outs simultaneously?

Yes and no. SPDIF in passes through the onboard sample rate converter. Would that be an issue?
post #6 of 43
Quote:
Originally Posted by najda View Post

Thanks. Precise questions deserve precise answers so here we go.
CS3318 with separate voltage regulators.
Over 8500 taps at 48 kHz sample rate freely distributed among channels (i.e. resources that you don't use on one channel can be reported onto other channels). A decimation stage for increased resolution is in the pipeline (Najda is fully upgradeable).

A mix IIR/FIR mode has been requested by customers and will be made available soon.
DSPB56725
CS42428
Analogue input: 1Vrms (fixed)
Analogue outs: 1 to 6 Vrms adjustable on each channel.

Nick

Thanks for the quick reply and precise answers Nick.

The ability to utilize a combination of IIR and FIR filters would be very useful with only 8500 taps available. That combined with decimation would enable the use of IIR filters for crossovers/PEQ and the FIR filter for phase linearization and room correction. IMO this is the best compromise.

I've been looking at using a combination of a MiniDSP OpenDRC-DI feeding a MiniDSP 2x8 to achieve something similar. The OpenDRC has 12280 taps to use between 2 channels and the MiniDSP would handle the IIR filtering and volume control. It seems that if I could do a hybrid of IIR and FIR with the Nadja I could solve everything in one box.
post #7 of 43
Thread Starter 
Yes, the mix IIR/FIR has already been discussed with users on another forum (diyaudio) and will be made available with the next update.
post #8 of 43
I found your thread. I'll buy one if that is going to be available. I just downloaded the software to take a look. Have you considered a more powerful DSP version with around 20k taps available?
post #9 of 43
Thread Starter 
Quote:
Originally Posted by coctostan View Post

I found your thread. I'll buy one if that is going to be available. I just downloaded the software to take a look. Have you considered a more powerful DSP version with around 20k taps available?

Well, every design is a trade-off. The DSPB56725 is, as far as I know, the most powerful DSP among the low-cost.
If we were using a more powerful one, the unit cost would be significantly higher, and development tools would also be more expensive. We would end with a product in another price category.

On the other hand, not everybody wants to implement FIR filters - and that's fine because Najda is not a dedicated convolver, but rather a complete audio solution with analogue/digital ins/outs, a still rather powerful DSP in between and easy interfacing with keypad/LCD/IR remote.

I'm convinced that a few thousand taps are fine for setting up multi-ways loudspeakers. We need to find out how, and that's the object of maintaining a close contact with users.
post #10 of 43
Most interesting, I need to look more into this, I hope more info will be available
post #11 of 43
Now that I've found out FIR is a possibility, I might order one. I'm currently saving for a DEQXX.

Is it posssible to get to the SPI lines for the CS3318? I ask because I have developed my own system controller and use the CS3318 in my modified DCX I/O boards, so it would make implementation and control easier for me if I could directly tell the VC chip what I want it to do.
post #12 of 43
Thread Starter 
Quote:
Originally Posted by datranz View Post

Most interesting, I need to look more into this, I hope more info will be available

Thanks. There's actually plenty of info on the website www.waf-audio.com.
You can also download the application and give it a try. Some views and dialogues can be seen only when the board is connected though (VU-meters and hardware configuration dialogues).

Quote:
Originally Posted by A9X-308 View Post

Now that I've found out FIR is a possibility, I might order one. I'm currently saving for a DEQXX.

Is it posssible to get to the SPI lines for the CS3318? I ask because I have developed my own system controller and use the CS3318 in my modified DCX I/O boards, so it would make implementation and control easier for me if I could directly tell the VC chip what I want it to do.

There are 2 expansion ports with SPI and I2C.
The CS3318 on board is well integrated with the application: it does not only volume control and mutes, but also sets the output nominal level on each channel independently so that you can boost or cut in the analogue domain rather than in digital, in order to compensate for drivers sensitivities. There's a screenshot of the dialogue allowing this in the Software section of the website.
post #13 of 43
Quote:
Originally Posted by najda View Post

Well, every design is a trade-off. The DSPB56725 is, as far as I know, the most powerful DSP among the low-cost.
If we were using a more powerful one, the unit cost would be significantly higher, and development tools would also be more expensive. We would end with a product in another price category.

On the other hand, not everybody wants to implement FIR filters - and that's fine because Najda is not a dedicated convolver, but rather a complete audio solution with analogue/digital ins/outs, a still rather powerful DSP in between and easy interfacing with keypad/LCD/IR remote.

I'm convinced that a few thousand taps are fine for setting up multi-ways loudspeakers. We need to find out how, and that's the object of maintaining a close contact with users.

I understand. This should work well for FIR crossovers in the upper frequencies where they theoretically provide more benefit. A combination of IIR/FIR and decimation would likely give it equal sonic performance to a device with enough taps to do everything FIR but at a much lower price.

I could also use an OpenDRC-DI and feed the Najda. Since the OpenDRC has more FIR taps available, it could implement room correction while the Nadja would handle the FIR crossovers.

Quote:
Originally Posted by A9X-308 View Post

Now that I've found out FIR is a possibility, I might order one. I'm currently saving for a DEQXX.

Is it posssible to get to the SPI lines for the CS3318? I ask because I have developed my own system controller and use the CS3318 in my modified DCX I/O boards, so it would make implementation and control easier for me if I could directly tell the VC chip what I want it to do.

I think a combination of an OpenDRC-DI feeding a digital signal to the Nadja would give you more FIR horsepower than the DCX and more output channels to send to multi-way speakers. The Najda could do 3-ways with mixed subwoofers whereas the DEQX HDP-4 only does stereo 3-ways I believe. The OpenDRC already has more FIR taps than the DEQX and the Nadja will give you further ability with FIR filters. I also like the idea of splitting the crossover/EQ and the room correction. They are really two separate tasks. This more open solution would also allow for use of different FIR calculations whereas I believe the DEQX requires the use of its own algorithms.

The crazy thing is that you can probably do all of this for less than $1k USD vs the HDP-4's $5k.
post #14 of 43
post #15 of 43
I p,an on ordering one of these devises for a dedicated 2 channel setup that I have at my office. I would like to use this to actively crossover my DIY 3-ways and also to do the room correction and EQ'ing. I will likely get rid of my older Music Hall 15.2 integrated and run this directly into my amp.
post #16 of 43
Quote:
Originally Posted by coctostan View Post

the HDP-4's $5k.
The Express is $2.5k and doesn't require any work to implement.
post #17 of 43
Quote:
Originally Posted by A9X-308 View Post

The Express is $2.5k and doesn't require any work to implement.

I've seen $3500 US for the Express II but I haven't looked much. It also constrains you to a 3-way I believe. I'm sure it is exceptionally easy to pull off. For the non-DIYer it is probably an awesome option.
post #18 of 43
Quote:
Originally Posted by coctostan View Post

I've seen $3500 US for the Express II but I haven't looked much.
2k5 is what DEQX told me.
Quote:
Originally Posted by coctostan View Post

It also constrains you to a 3-way I believe.
I only need 3 way.
post #19 of 43
The Express II is $3,495. I should know...I am a dealer.

DEQX's algorithms are proprietary but you are not just paying for DSP power, you are also paying for 32 bit / 192 kHz DACs and a 3rd generation electronics platform that has had 10 years of optimization.
post #20 of 43
Quote:
Originally Posted by Nyal Mellor View Post

The Express II is $3,495. I should know...I am a dealer.
When I last spoke to Alan from DEQX, they sold direct and were 2k5, but that was for the Express 1. Speaking to a local dealer today, they are 3k5 retail for V2.
Quote:
Originally Posted by Nyal Mellor View Post

DEQX's algorithms are proprietary but you are not just paying for DSP power, you are also paying for 32 bit / 192 kHz DACs and a 3rd generation electronics platform that has had 10 years of optimization.
I know they are very good and easy to use; I have a friend with a 5 way system using 2 HDPs. I've been a fan of Kim's work since the Fairlight.

However the 32bit DAC stuff is spec rubbish. Show me a DAC that's linear for much more than 22 bits.
post #21 of 43
I plan on ordering a Najda for actively crossing over some 3-ways that I am almost finished with. I will also use the Najda for EQ'ing them as well. I am sure that their are many more uses for the Najda, I just don't know about them! I have choosen the Najda over the MiniDsp. I hope that it turns out really good!
post #22 of 43
Thread Starter 
V0.0.8 released, available from the website in the download section.

List of changes:

- Custom biquads (DF1 coefficients)
- Global EQ Bypass on each channel
- Small Apple remote support
- Auto-mute when SPDIF input is not PCM
- Individual IR commands for selecting each source
- Channel Status of SPDIF out 5 (channels 9&10) show now the sampling frequency.
- Bug fix for the I2S input.
post #23 of 43
This looks interesting, but limiting input to 1v rules this out for me. One of the selling features for the MiniDSP I'm currently using is 8v balanced inputs and outputs.

Onto another question..

Frequency response, how low does it go and how accurate are your filtering capabilities at 48khz around 20hz? 10hz? Where does it roll off?

Thanks for your time
post #24 of 43
So what does having a 1 volt input limit you to?
post #25 of 43
Quote:
Originally Posted by Martycool007 View Post

So what does having a 1 volt input limit you to?
Consumer grade gear, at -10dBV. Pro runs at +4dBV.
post #26 of 43
Thread Starter 
Quote:
Originally Posted by notnyt View Post

This looks interesting, but limiting input to 1v rules this out for me. One of the selling features for the MiniDSP I'm currently using is 8v balanced inputs and outputs.

Onto another question..

Frequency response, how low does it go and how accurate are your filtering capabilities at 48khz around 20hz? 10hz? Where does it roll off?

Thanks for your time

Hi,

This board is intended for domestic use. I don't know what you would connect to the analogue input. I can imagine a turntable or an occasional pocket player when pals on visit. In both cases a balanced line is probably not essential. Otherwise most of the sources would anyway be digital and connected to the SPDIF ins or I2S.

Regarding max level of the analogue ins, that's 1 Vrms or 2.8 V peak-to-peak or 2.2 dBu or 0 dBV. That's a pretty standard consumer level and it's fine if you're near clipping because you want to use all the dynamic range of the ADC. Otherwise if you need to adapt the input level, you'd probably rater attenuate than boost.

Filtering parameters go down to 10 Hz. With custom biquads you could go lower. All biquad coefficients are 48-bit, so you can trust the filters to be spot on even in the low freqs.

Best,

Nick
post #27 of 43
Quote:
Originally Posted by najda View Post

This board is intended for domestic use.
If you look about here you'll notice that many use pro-sound power amps running at +4dBV. DIY speaker builders are anything but average consumers.
post #28 of 43
Thread Starter 
Quote:
Originally Posted by Bill Fitzmaurice View Post

If you look about here you'll notice that many use pro-sound power amps running at +4dBV. DIY speaker builders are anything but average consumers.

Hi Bill,

That's perfectly fine. In the previous posts we were talking about input level.

If now we're getting to power amps, then we're talking about output level. Najda's max output level is adjustable, up to 6 Vrms (15.5 dBV). That should be good for driving any power amp.

Thanks

Nick

(PS. I'm a DIY speaker builder too wink.gif )
post #29 of 43
Quote:
Originally Posted by najda View Post

If now we're getting to power amps, then we're talking about output level. Najda's max output level is adjustable, up to 6 Vrms (15.5 dBV). That should be good for driving any power amp.
If that's the case it could solve a lot of problems for those trying to use pro-sound amps with AVR LFE sourcing. It would be even better if that output was balanced.
post #30 of 43
Thread Starter 
Software user manual posted today (in the download section).

Best

Nick
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: DIY Speakers and Subs
AVS › AVS Forum › Audio › DIY Speakers and Subs › Najda DSP board for loudspeaker applications