Open Hardware DSP Platform

Hi everyone,

A while back I had started playing with some digital signal processing stuff, had been looking at FreeDSP and a similar project in Elektor, neither offered what I wanted and the MiniDSP seemed the simpler option to get working with initially.

I did start playing around with designs for a more powerful/complex DSP board about a year ago (raspberry pi hat type board: https://github.com/pwzj/ohdsp), never finished it in that form though. And then I also picked up on the FreeDSP V2 thread. Again what was being designed (at the time) wasn't quite what I was after and I shifted my focus to designing a set of modular boards I could play with (but not bothering to link in a Pi or any other small board computer).

Anyway I have real boards I am building up and stuck some design files and a website up: Open Hardware DSP Platform

This is not designed to be a rip off or clone of any other work but as I have completed PCB designs and actually have real hardware I am building up now it would seem a waste not to make it available for others.

All hardware designs are under the TAPR open hardware license. PCBs/Schematics are in KiCad. I don't sell PCBs or hardware or anything for this, the website is information only at the moment.


Cheers.
 
Congrats to the project and for using KiCAD. :D

You may want to overlook the pcb images on your website, which are quite "huge" pngs. (600kb)

I'll attach the shrunken down files.
 

Attachments

  • 2453515_orig.png
    2453515_orig.png
    80.6 KB · Views: 1,366
  • 5832289_orig.png
    5832289_orig.png
    97.9 KB · Views: 1,361
  • 758901_orig.png
    758901_orig.png
    64.7 KB · Views: 1,331
  • 8978959_orig.png
    8978959_orig.png
    68 KB · Views: 1,326
  • 4654162_orig.png
    4654162_orig.png
    59.9 KB · Views: 1,322
Last edited:
Super project and congratulations !!

I am also following the freedsp V2 thread which is also very interesting. I love the idea of you sharing and opening all your designs for us to learn and discuss. Sincerely, thank you.

On a more operational note, I have done some smd soldering in the past with different results, to the point that I am considering building a diy oven to make things easier. My eye are not what they use to be, even with a magnifier. How is your experience with the ADAU1452 and your boards ? Any recommendation for manufacturing ? (site...)

cheers
Pepe
 
An SMD toaster oven may be a good investment if you fancy doing a lot of boards. I have been building up 8 of these boards (3 DAC, 3 DSP, 1 ADC, 1 PSU) and there are a lot of SMD parts...

Soldering the ADAU1452 was ok, I put a large (3mm) via in the centre pad, fixed the chip around the sides and soldered up through the centre, 2 boards are fine although the 3rd looks like it may not have flowed as well as it should have, but that was just me being unsure having not used this method before.

I have boards up and running now. Feeding the ADAU1452 from a SPDIF output and into the ADAU1966. I have connected a headphone amplifier to the positive outputs of channel 1 and 2 on the DAC (only passive filters). The ADAU1966 sounds very good. Very very good in fact, given its a 16 channel dac chip for less than £8. It is also seems happy driving long spindly unbalanced phono cables with no buffering.

I have not tested everything yet but this is a good start! And a good sounding start at that as well.

Here is a (bad) video of my test setup running on a bench supply, on some stupidly small table.... I do now have it running off of the real supply as well:

https://goo.gl/photos/qqrevPUD3Sfw4fRK8

All I need now is a decent workshop area I could setup in...
 

ojg

Member
Paid Member
2003-05-15 11:35 pm
Norway
Hi ohdsp,

I really like your boards and I really like the "competition" between you and freeDSP, finally there is some momentum on the DIY DSP front.

I am curious about your soldering of the ADAU1452, are you saying that you were able to solder this using just a soldering iron? And not stencil, paste and toaster-oven? If so, you should consider making a video on how you did that, that would be very interesting.

Br,
OJG
 
Yes I did solder it by hand with a fairly large soldering iron bit, although on a cheap temperature controlled station. I can have a look at making a video on how I soldered the chip, but this may not happen for a couple of weeks as the parts I have would need baking. That is where a toaster oven would come in handy! Ideally the parts and boards should go through a preheat stage. I figured parts have been soldered by hand for many years without issue. If it works using my crude method then any form of reflow (air gun) or oven methods should work well if you stick within the temperature and time specified in the datasheet. If you can get free samples of the parts from Analog then the risk isn't so high.


Also on Pepe's question about manufacturers I used DirtyPCB for all my boards in 1 batch.

The 4 layer board is designed for a 0.2mm prepreg, 1.2mm core, 0.2mm prepreg stackup (core could be any thickness though). Boards are not impedance controlled but traces to IDC connectors are designed to have a ballpark figure to match with 26AWG ribbon cable driven in signal-ground-signal-ground config (but not 28AWG as I could not match that). The traces are quite thin (0.135mm/5.3mil) but a lot of manufacturers now support down to 0.127mm/5mil as standard. I would recommend DirtyPCB for the 4 layer board.

The 2 layer boards can be from anywhere, really does not matter. Traces are 0.3mm/11.8mil or thicker and the PCB thickness is 1.6mm as normal.

The boards I have are generally very good. Some silkscreen and a couple of holes not quite lined up 100%, also they sent 12 4-layer boards but 2 of them have the via's not seated in the board properly (copper sitting proud) which ruins the ADAU1452 footprint, but I did get what I ordered in usable boards.

There are mistakes on Release 1.0 boards but they do work, I will work on 1.1 boards with all the mistakes fixed and some minor tweaks to things like connector hole sizes.

Cheers.
 
Thanks for the comment DieterK. I have thought about selling completed, or part completed, boards if enough interest builds in the project. I did not see the point in trying to do this without any interest.

Ideally I would have waited for a polished/finished system before posting, but as I was making it all free/open it didn't matter much to me. Long term I have more ideas, some I have already started working on, which might make this system a little more desirable. If not I'll just end up with some crazy DSP/volume control system driving my homebuilt speakers, and it would still cost less than equivalent passive commercial speakers+amplifiers, which is fine by me :)
 
If you have a DSP and DAC board you can cheat and connect to the 0V and positive output of each channel on the DAC PCB. Leave each negative output unconnected. (or if you want to have the phase inverted connect to negative and leave positive unconnected). The output level may be a little high for some amplifier inputs.

You would need to add a dc blocking capacitor as the positive and negative outputs sit at around 2.5V. A 100uF aluminium electrolytic (Panasonic FC or other good cap) and 100nF polyester film in parallel would be fine.

A diagram will probably make it clearer; Have attached a sketch with added channel markings and two channels connected to phono/rca sockets.

DAC-ADAU1966-brd-anon.png

Cheers
 

carlos515

Member
2016-01-08 9:49 pm
Thanks ,
I have ordered all the boards and a hot air rework station . This is an interesting project. Will the freeusbi and CYC68013A usb eeprom board work to program your dsp board? Also will I have to disconect the dsp control board from the power supply when programming ?
I have the option of a 1A wall wart or 3.3A wall wart. What would the draw be for , 1 adc , 1 dac and 1 dsp board. I would be using 2 input and 6 output. 3 way crossover.

Thanks,
Carlos
 
The freeusbi option should work. I have not tested it with this DSP board but it should just emulate the official Analog Devices EVAL-ADUSB2EBUZ.

Also a 1A transformer would be fine; my speaker crossover box I have been testing seems to draw around 170 to 200mA from a 9Vac transformer. My transformer is rated 10VA at 9Vac on the secondary, so around 1.1A.

Edit: I should also point out that the PCM4202 ADC seems to have weak output drivers; I accidentally set my DSP and the PCM4202 to be outputs on BCLK, LRCLK and SDATA and now my PCM4202 is spewing out junk on the I2S interface. Just make sure you have the DSP setup properly before connecting the cable to the ADC!

Cheers.
 
Last edited:
3 channels in ?

Hi ohdsp,
I have my boards. All 50 lol . I was wondering if the ADAU 1452 can handle 3 channels in. I would use 2 ADC4202-se boards. I was thinking of my center channel .But i have not looked at the Sigma software. Can the software also handle 3 distinct channels to cross over ? The center channel is a 2way while my main channels will be 3 ways.

Thanks,
Carlos
 
Hi Carlos,

Good news with the PCBs, at least you won't run out :D

The ADAU1452 can support up to 48 channels in and 48 out through a mix of I2S and TDM interfaces.

The PCM4202 is I2S only though. But you could connect up 4 of these for 8 channels in.

Internally you can pretty much wire up the filters as you wish. Under SigmaStudio there is a routing matrix where you can tell an input to go to the Core, or through an ASRC (really useful for different clock domains).

With inputs routed to the Core, and outputs from the Core, you can "drop" input/output connectors on the main schematic. And then just connect up whatever filters/equalizers or other effects are needed in between.

A good play around with SigmaStudio will help your understanding; not sure if there are any tutorials online.

Cheers.