DIY active crossover in development - looking for inputs

[Sokel]

How do you mean? I haven't seen any signs of the DSP beeing unstable.

Anything that can go wrong will eventually do at a piece of software, either by itself or by a glorious BSOD for example.

Don't forget that the main x-over job is levels.
So, fail-safes everywhere, there's not too much relays you can use and there's not too much coding when reverting to defaults so everything is muted.

 

[DannerD3H]

Anything that can go wrong will eventually do at a piece of software, either by itself or by a glorious BSOD for example.

That is true.. But I guess any piece of electronics will die eventually?

Don't forget that the main x-over job is levels.
So, fail-safes everywhere, there's not too much relays you can use and there's not too much coding when reverting to defaults so everything is muted.

So you would like to see a design that if someting goes haywire in the DSP it will just auto mute?

The current plan was indeed to include mute relays on the DAC's. They will not unmute until all supplies are up. They also will not unmute before they get a valid signal from the DSP (which means the DSP has to be running its program properly). If DSP is bricked or dead the output of the MP pins will hopefully just be grounded?
Otherwhise I have no clue how to protect it any better against DSP failure?

Additionally the mute relays can be controlled form a buton on the front panel.

That is at least the plan.

 

[Sokel]

That is true.. But I guess any piece of electronics will die eventually?


So you would like to see a design that if someting goes haywire in the DSP it will just auto mute?

The current plan was indeed to include mute relays on the DAC's. They will not unmute until all supplies are up. They also will not unmute before they get a valid signal from the DSP (which means the DSP has to be running its program properly). If DSP is bricked or dead the output of the MP pins will hopefully just be grounded?
Otherwhise I have no clue how to protect it any better against DSP failure?

Additionally the mute relays can be controlled form a buton on the front panel.

That is at least the plan.

Seems like a sound plan to me.
Add some caution when using it and everything will hopefully be ok.

 

[DannerD3H]

Seems like a sound plan to me.

Awesome!

 

[Keith_W]

Those are some very interesting links, thanks! And sure, once the basic electronics and audio quality is in place I need to dig deeper in the DSP technical part of my filters.

Have a look at this thread, and welcome to your new hobby. It's an endless rabbit hole.

There are two DSP "flavours", IIR and FIR. It is much easier to design filters for linear-phase FIR for many reasons - it is immune to the vagaries of hardware, immune to feedback and instability, filters can be designed without having to worry about phase cancellation, and so on. Designing IIR filters is more difficult since everything is minimum-phase. There is "easy" IIR, where you rely on software like Dirac and REW to automatically assign PEQ's for you, and there is "difficult" IIR where you write all the coefficients yourself. If you have the skill and determination to make a DSP unit like this from scratch, you might enjoy the challenge.

And surely a set of proper DAC boards must provide an audible improvemen over these 3$ China DACs?

Weren't you talking about snake oil?

Yeah, "proper" DAC boards might give you a measurable improvement over $3 China DAC's. But it is very unlikely that the improvement will be audible.

 

[DannerD3H]

Have a look at this thread, and welcome to your new hobby. It's an endless rabbit hole

Was actually allready reading through that thread. Great stuff!

If you have the skill and determination to make a DSP unit like this from scratch, you might enjoy the challenge.

This isn't the first time I use a DSP for speaker design. Just haven't used FIR before as I have the older Fusion amps that doesn't support it.
I usually do measurements with REW and then simulate and verify in Vituixcad. This works great Vituixcad can give you all the biquads to load directly in sigmastudio. Doesn't get much easier than this in my eyes. And I do enjoy the simulation part as well Vcad can also du FIR. As mentioned, just havent tried it yet.

Weren't you talking about snake oil?

Yeah, "proper" DAC boards might give you a measurable improvement over $3 China DAC's. But it is very unlikely that the improvement will be audible.

Haha true... You got me there
As mentioned in the first posts, I am already very happy with the sound quality. I am very amazed by what they can do!
But I still feel like I want to try and put a opamp based output stage on the PCM5102 and also compare it to a AKM version (in blind test of course) to hear for myself what the difference is - if any. Also to learn
Another main reason for modularity - To test differnet hardware - To test what I can hear and what I cannot. In the future I want to be able to also make a all-in-one PCB with DSP and DAC for a small plateamp style build. But first I wanna know whats good enough!

 

[alc]

Congratulation to DannerD3H for taking on an ambitious project!

I am Al Clark, designer of the Danville dspNexus.

I would like to add a few comments since the dspNexus has been mentioned several times in this thread.

Facts:

The dspNexus is a modular based design that currently uses AKM AK4499EX DACs, AK5578 ADC and SHARC ADSP-21569 with DDR3 DRAM.
It supports fairly long FIRs at 192k sampling and also 100s of IIR biquads. The most popular version has 8 balanced output channels. It also has S/PDIF in & out, and Bluetooth apt-HD. It also comes with an Audio Weaver license which is more powerful that SigmaStudio and does not require Analog Devices CCES for SHARC based designs.

Opinion:

In my view, you start with good ingredients. I think the AKM converters are the best. The AK4499EX is the best DAC I have worked with. They are also about 10x the cost of AK4493SEQ which are very good DACs. I have built a lot of different DAC and ADC circuits from many manufacturers.

I think that to design good digital audio electronics, it starts with good analog design skills. It is more that just copying the manufacturer's reference design on to your own PCB.

There are good reasons to use IIRs and others to use FIRs. You want to avoid high Q IIRs and preringing in FIRs. It's nice to have both in your toolbox.

Offer:

If any of you would like to have a discussion (or sales pitch) about the dspNexus, we have the normal contact us form on our website and I would be happy to set up a video call.
I talk a lot better than I write.

Al Clark
Danville Signal

 

[DannerD3H]

Congratulation to DannerD3H for taking on an ambitious project

Thanks! You seem to have done pretty much the same as what I am going for. Its cool to see that these products actually does start to appear!

Audio Weaver license which is more powerful that SigmaStudio

Can you give an example of something audio weaver can do that SS cannot?

Analog Devices CCES for SHARC based designs

I was of the impression that SS and SS+ was able to handle SHARC processors as well. Did I get that wrong?

In my view, you start with good ingredients. I think the AKM converters are the best. The AK4499EX is the best DAC I have worked with. They are also about 10x the cost of AK4493SEQ

Thanks for the inputs here! Yeah the AK4499EX is out of my price range. I think I will be more than happy with the performance of the AK4493SEQ

It is more that just copying the manufacturer's reference design on to your own PCB.

Absolutely.. They are sometimes even down right bad.. Maybe not for AKM, but in general..

There are good reasons to use IIRs and others to use FIRs. You want to avoid high Q IIRs and preringing in FIRs.

I certainly have a learning curve ahead of me regarding proper use of DSP's. Until know I hvae just made filters without much thought of things like preringing and so on. But that will come eventually I will also be using my speakers for HT use, so tooo crazy bass FIR filters may have too much delay for me to use (Unles Im completely off here).

If any of you would like to have a discussion (or sales pitch) about the dspNexus

Thanks, but I'm not here to buy a finished product, that ship has sailed for me... I'm here to make my own!
I dont have 3000$ lying around anyway.. So far this has cost me around 150$ for what is currently in the DSP. That does of course not include all the stuff that created magic smoke along the way or the amps.. But it allready performs better than I imagined it ever would :O
The new chassis alone will add another 200$ alone though :O .. In the end I will probably land close to the 1000$ mark for a finished DSP.. But spread out over a couple of years, it doesnt feel as bad xD

 

[MCH]

I am glad to see your project here and I love it! In fact, I am planning to do something related, (but USB input and leaving the dsp to a computer upstream).
I didn't read all the thread, only the description , so maybe my comments have been addressed but there you go:

- don't be afraid of making your own dacs: nowadays there are cheap and good enough multichannel chips with additional features, like spdif outputs (in synch to the analog ones). Integrated output stages that save you some $ in opamps if you do multichannel... You might end up much better than with off the shelf modules in performance and cost.

- don't know how you are dealing with clocks, but you could make different inputs and outputs to be clock synchronous or even have master clock input. Could become handy for measurements etc.

- with raspberry pi in general and the new 5 in particular.... I don't know, but for a diy project, unless boot time is of the foremost importance.... I would go the rpi way, even the boot time with the 5 is really short. I also hate having to wait but the advantages: support, computing power, expansion possibilities... are unbeatable

Good luck and keep us posted!

PS: ah, forgot to mention, I also use usb-pd in all my projects now, don't know how you are dealing with the inrush current, but my experience is that USB chargers behave very differently, most increase the voltage with a nice slope, others are more aggressive

 

[DannerD3H]

I am glad to see your project here and I love it! In fact, I am planning to do something related, (but USB input and leaving the dsp to a computer upstream).

Thanks!
So basically you wnat a multichannel USB DAC?. Thats also interesting, but I want something standalone without the need of a PC. But thats two different things

don't be afraid of making your own dacs

I'm not

with raspberry pi in general

Any sort of microcontroller is secondary for me, mainly to add IR remote.. Sp thats very low priority at the moment. For sure will happen some day, and maybe will be used with display and other nice features at that point, but for now and the next couple of years its barely an idea.. xD

I also use usb-pd in all my projects now

While it is really nice and I certainly will be using it more in other porjects, I am kinda leaning towards making it running on mains instead. Just to avoid external power adapters and simplify/streamline the interface of my DSP and the two power amps that will accompany it.
But I didn't do much against the inrush at startup.. The USB power was going straight through a filter after the PD IC.

 

[DannerD3H]

Regarding clock...

don't know how you are dealing with clocks, but you could make different inputs and outputs to be clock synchronous or even have master clock input. Could become handy for measurements etc

This is also one of my biggest challenges (in my head at least). Not sure how I want to do it...
Right now there is a very simple and probably not super great clock circuit on the DSP board, which feeds the ADC and DAC's. While it works it still has to be changed in the next version. I am just not sure if I want to just do a better integrated circuit in the DSP board or make a seperate clock module that connects to the main board via U.Fl or something like this:
View attachment 471039
I like the idea of this just plugging into the DSP core.

1. First of all I can experiemnt with different clocks, or even make an adapter PCB to allow for a panel mount clock input (not sure I wil personally ever need that though).
2. If I use exotic clocks it will be very easy to salvage it and reuse it with a different DSP module.

However, I cannot help but think its a little downgrade from implementing it directly on the board with proper impedance traces and so on from the start instead of going through the pins here which much have a higher impedance than the 50ohm used everywhere else, including the UFL cables.
Or am I overthinking the impact of this connection? Just seems a bit counter productive to take a expensive clock and the put a bad interface between it and the DSP and ADC/DAC's..

 

[DannerD3H]

Prototype chassis done and looking good!


Bottom, rear, top and front plates will be aluminum when I know exactly where I want all the holes.

For now the rear plate is prepared to move my current prototype over to the new chassis.

 

[ThatSoundsGood]

Amazing project. Let me know if you need a beta tester!

 

[Bjorn]

Good luck with the project!

If you have the time, knowledge and interest of going commercial there's presently a clear opening for that. No one is delivering a great DSP today for a decent price. My company is still looking.

 

[Keith_W]

If you have the time, knowledge and interest of going commercial there's presently a clear opening for that. No one is delivering a great DSP today for a decent price. My company is still looking.

Out of curiosity, what do you mean by "great DSP" and "decent price"? I am chatting to someone about bringing a DSP product to market and he might appreciate the feedback.

 

[Davey]

Out of curiosity, what do you mean by "great DSP" and "decent price"? I am chatting to someone about bringing a DSP product to market and he might appreciate the feedback.

Yeah, it's a curious statement. There are a number of great DSP products at a decent price already available. Maybe my definition of "great" and "decent" are not the same as Bjorn's?
Anyways, this DIY project looks pretty cool.

 

[Bjorn]

Out of curiosity, what do you mean by "great DSP" and "decent price"? I am chatting to someone about bringing a DSP product to market and he might appreciate the feedback.

I don't think it's fair to DannerD3H to bring that topic here and lead the discussion off topic. He is presenting his DIY product.

 

[Ktacos]

Out of curiosity, what do you mean by "great DSP" and "decent price"? I am chatting to someone about bringing a DSP product to market and he might appreciate the feedback.

Basically the market needs something like a minidsp flex but for half or less than half the price of the flex unit.

 

[Keith_W]

Basically the market needs something like a minidsp flex but for half or less than half the price of the flex unit.

Interesting perspective. I would have said that the MiniDSP Flex has "good, but not great" DSP and IMO it is competitively priced.

 

[Sokel]

Basically the market needs something like a minidsp flex but for half or less than half the price of the flex unit.

The opposite, market needs a decent processor even at double the flex money.
And I'm not talking about SINAD.