Let's create an 'open source' hardware project.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Just like the Jehovah's haven't converted me by knocking on my door hundreds of times, I fear that hundreds of posts in ZMB's thread isn't going to make many people actually want to pursue open source. Perhaps something more tangible can get people excited. :D

We've got plenty of experience on this forum in analog/digital audio design, clock design, DSP, and who knows what else. Who says we can't pool a few of our talents together, use a few heads, bang a few heads and create some sort of open, collaborative, "big" yet fun hardware project with a DIY audio theme?

My idea of an "open source" hardware project is something like what the ADI Blackfin DSP geeks are up to. They've taken a cheap DSP chip, created a "BF533 STAMP" board with lots of ram/flash/peripherals and a range of add-on boards, and ported ucLinux and the GCC compiler toolchain to the DSP. http://blackfin.uclinux.org/ has all the details on that.

Drawing inspiration from that but thinking "audio! audio!", what's a project we could undertake? Here's a couple of ideas:

- a DSP board, a Behringer DEQ type of thing, which could be used for active speaker crossovers, room equalization, and who knows what else.

- A digitally controlled preamp, receiver, or something in between.

Thoughts? ideas? comments?
 
BrianDonegan said:
Aren't all the project (or most of them) posted here open source? I think we may be confusing "open source" with "do-it-all"?

If you publish your hardware and software designs - PCB's, HDL, uC/uP source code, etc... and do so under a public domain or perhaps GNU/BSD license, it's "open source" IMO. And that counts as lots of things.

I guess what I'm suggesting is a "perhaps too complicated for 1 guy in his spare time" type of collaborative project - with all the aspects of the design made public.
 
Sounds good, PATA would interesting to implement as you could then use formatted compact flash cards instead of hard disks if you wanted for noise reasons, and still maintain compatibility with HDD. Although, today's fluid bearing 7200 RPM drives are pretty much inaudible from a few feet away depending on the model.
 
Interms of a DAC, it would be great to incorporate balanced analog outputs and a modern DAC chip like the the TI/BB 1792/4 or the Wolfson 8740. The Chinese Zhalou 2.5 had interesting feature of placing the core DAC chip on a daughter card that allowed interchangeble DAC modules. Designing a base set of capabilities that are augmentable with a daughter card for the DAC chip and I/V stage would be an interesting concept. A design platform approach might have a broader acceptance in our community. There is also a lot of knowledge over at www.head-fi.org, and the Pitcan Island guys (former www.diyaudio.com folks)over at www.diyhifi.org.

Designing DAC's seems to be a specialized art. It is surprising how few DIY DAC designs are not built around more modern DAC chips. Especially given how long 1792/4 and 8740's have been in production and well reviewed commercial implementations around these chips have been.

Something to consider.

-David
 
chris719 said:
Maybe some kind of digital filter where one could choose from many filter profiles and just upload the code to the DSP? Although digital filtering might be better suited to FPGAs.
DSP's are specifically designed for digital filtering - a MAC instruction is 99% of the time used to process an IIR or FIR filter tap.

And DSPs tend to be muchmuchmuch cheaper than FPGAs - you'll find FPGAs doing filtering at very high sample rates (>10MHz) while DSPs are used almost universally at audio rates.
 
How about a squeezebox type controller which can read flac directly from a hard drive? This eliminates the need for a pc server, or even a network harddrive, and perhaps a convolver on the output (perhaps this requires too much processing). Suppose this would be similar to a number of the comercial harddisc players.

I liked the 'Digital audio over IDE' idea, but guess that could be held up by the same issues as last time i.e. lack of an IDE driver (same for the first idea).

I think a lot of the projects here could be classed as open source, but no reason not to have another - there arent many if any DSP based projects.

Ross
 
DSP => floating point calculations => filters.
So instead of implementing boring effects like "Hall, Rock, Jazz"-programs, lets implement "Rotel, NAD, Thule, Tube" etc...

There would be quite a big deal of measuring and getting the transfer functions right, but wouldnt it be cool :devilr:?
 
gmarsh said:
My idea of an "open source" hardware project is something like what the ADI Blackfin DSP geeks are up to. They've taken a cheap DSP chip, created a "BF533 STAMP" board with lots of ram/flash/peripherals and a range of add-on boards, and ported ucLinux and the GCC compiler toolchain to the DSP. http://blackfin.uclinux.org/ has all the details on that.

Drawing inspiration from that but thinking "audio! audio!", what's a project we could undertake? Here's a couple of ideas:

- a DSP board, a Behringer DEQ type of thing, which could be used for active speaker crossovers, room equalization, and who knows what else.

- A digitally controlled preamp, receiver, or something in between.

Thoughts? ideas? comments?

Instead of gathering a bunch of harebrained suggestions about what the project should be, perhaps a more prudent first step would be to identify who is willing and able to do different parts of such a project and let those who will likely be doing all the work decide what they want to do. Unless you first line up the talent, nothing will get done and this will be just another pointless exercise like the IDE project and ZMB’s hopeless thread.
 
jbokelman said:
Instead of gathering a bunch of harebrained suggestions about what the project should be, perhaps a more prudent first step would be to identify who is willing and able to do different parts of such a project and let those who will likely be doing all the work decide what they want to do. Unless you first line up the talent, nothing will get done and this will be just another pointless exercise like the IDE project and ZMB’s hopeless thread.

You can't round up people, then push a project on them they're not interested in and expect them to work on it for free. All hands need to be passionate about the work they do or you're going to get work that's crap, unfinished or both. So, projects first.

If this is a pointless exercise, *shrug*

As far as the IDE project, I got about 75% there - I had a proof of concept running on my Blackfin BF533 evaluation board using a homebuilt IDE interface. But I ran into a few things:

- The interest in the project was practically zero.
- My own interest wavered and I found myself spending my time on other projects.
- Making Linux run (as opposed to VisualDSP++ code) would have required a great amount of effort at the time.
- But the nail in the coffin - CDROM drives are loud! I had digital audio extraction running on a few different CD/DVD drives, and once you start overreading 4 times for cdparanoia quality extraction you need the CDROM set to at least 24x mode to make it work realtime. I started wondering at that point whether the source hardware should be something quiet like a compactflash or SD card, even a hard drive.

But it was no waste of effort. The IDE interface ended up integrated into a Blackfin STAMP add-on board and a Linux driver exists now, and I've used the same interface and code with a Compactflash card in another project.
 
BrianDonegan said:


I've participated in the above referenced Wolfson 8740 thread, but the design is still untested. To reinforce the original point, this is still the only 8740 project that I'm aware of to date.

There is a shockingly limited number of DIY DAC's that have been built using modern DAC chips. A really good balanced DAC seems like an essential building block to a project of this type having traction.

There seems to be a lot of nuisance in building a quality DAC. There are so many nice looking Chinese DAC's, but their performance seems to come up seriously short when A-B compared to entry level US commercial models like the Lavry Black, Benchmark DAC-1, etc. for under $1KUSD retail list.

There has been a lot of success building clones of commercial Amps and Pre-amps, but a strong DAC seems to elude the DIY community.

Just my $.02 worth, but the DAC seems a more critical starting point than a DSP.

-David
 
gmarsh said:
You can't round up people, then push a project on them they're not interested in and expect them to work on it for free. All hands need to be passionate about the work they do or you're going to get work that's crap, unfinished or both. So, projects first.

Exactly, that’s why I suggested those who were going to do the work should decide on what they wanted to do. Maybe we should start a registry (wiki?) where interested parties could list their skills, interests, and availability. So, for instance, if I wanted to pursue a particular project and didn’t have the skills or time to do it all, I could search the registry for individuals who could supply the talent I lacked.
 
...or just post your ideas and what you want to achieve, and what you have, and others will post comments and advice. It could be like a threaded conversation among people interested in the topic giving what help and advice they can based on thier knowledge. The best part is that others who have limited knowledge in some areas would be able to read all the input and possibly learn something new. We could call it an online forum... wait ;)


:clown: Sorry, just had to. :clown:

Now, back to the constructive posts....:smash:
 
BrianDonegan said:
...or just post your ideas and what you want to achieve, and what you have, and others will post comments and advice. It could be like a threaded conversation among people interested in the topic giving what help and advice they can based on thier knowledge. The best part is that others who have limited knowledge in some areas would be able to read all the input and possibly learn something new. We could call it an online forum

Yes, you can do it that way but look how far it’s gotten GM and ZMB. With the current system, anyone with expertise and an interest in collaborating in particular areas has to read nearly every post to learn where he might lend a hand. Likewise, anyone starting project has to hope the right people see his post before it disappears into the void. That’s not very efficient.
 
What I would like to see is a clone of the DEQX digital crossover system. In time, and it will be a long time, I am going to do this myself. I think I even posted this idea somewhere in this forum before, although, I know a lot more now than I did then. Basically, I'd like to take the digital output (in some lossless format) from the source (which could be computer, SqueezeBox, CD player, etc), send it through the digital crossover to separate it into 2-,4-,6- channel, whatever is desired really, and then send each of those channels to a separate DAC channel, and then obviously to an amplifier, and on to the speakers. The essential component here is the modularity of the digital crossover, and that's where the DEQX clone comes in. This "clone" should be able to do room correction, low-pass, hi-pass, whatever you want in software, and should be easily programmable by a user, with a straightforward, user-friendly GUI (like an audio cockpit!). Every other component - upstream or downstream - would then be up to the user to decide how to setup (i.e. how many DAC's, amplifiers, speakers, etc.). This project would be mostly about software, because it would be digital in-digital out (of course, it could be transformer-coupled digital out, etc). Once the chips are chosen and PCB is made, the "open-source" part of it would take over. Anyone could add new algorithms, fix bugs, etc. This is my idea of a great community project, and I know the shared experience and intelligence of the DIY community is fully capable of making such a beast.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.