Hi all,
I have a few spare DSP boards left from a past project, they would nicely fit now a new Xover project.
I'm thinking of writing a computer application + DSP firmware that would allow the user to set crossover filters, parametric EQ, delay lines and gains, in order to resell these boards that i'm not likely to use.
The purpose of this thread is to poll the interest of potential buyers and collect requirements before starting to work on it.
What i'd like to know is:
- Is there anyone interested in such a board?
- What kind of processing should i make available, apart from Xover (Buuteworth/Linkwitz-Riley/Bessel up to 48dB/oct), PEQ, shelving filters, delays and gains?
- What form should the project take? Me doing the whole stuff in a closed and controlled manner, or releasing an open-source framework?
- Whatever comes to your mind.
Here are some details about the board itself:
- The DSP is a Freescale DSP56321 clocked at 275 MHz. It's capable of 275 MMACS, has a coprocessor and boasts plenty of RAM, enough to make a 4 second delay line at CD sampling frequency. The DSP is 24-bit fix-point with 56-bit accumulators. It's been a classic architecture for years in DSP for audio processing, and i guess that many of you have some custom code for it.
275 MMACS is A LOT! Enough for steep crossover filters, countless PEQ or FIR filtering.
- There's a USB connection allowing transfers of data to/from PC up to 8Mbit/s (USB High-Speed). The controller chip is manufactured by FTDI.
- There's a codec onboard with 2 analog inputs and 6 analog outputs. The chip is by Analog Devices ref. AD1838A. ADCs and DACs have both over 100 dB dynamic range, inputs and outputs are balanced or unbalanced. There are plenty of clock options, with 2 quartz onboard switcheable from the DSP.
- There's an expansion connector for separate I2S in and out, SPI and GPIO. Everything is there if you want to drive your own ADCs/DACs/SPDIF tranceivers.
- There's an other port with 5 outputs and 3 inputs for leds and pushbuttons.
- JTAG is also onboard if you wish to write your own code.
- The board requires 5V and +/-12V. The rest of the power supply is included on board with quality TI regulators. I can supply the matching SMPS.
- Finally, it's a high-quality board with 6 layers, vast ground planes and branded chips.
Best!
chaparK
I have a few spare DSP boards left from a past project, they would nicely fit now a new Xover project.
I'm thinking of writing a computer application + DSP firmware that would allow the user to set crossover filters, parametric EQ, delay lines and gains, in order to resell these boards that i'm not likely to use.
The purpose of this thread is to poll the interest of potential buyers and collect requirements before starting to work on it.
What i'd like to know is:
- Is there anyone interested in such a board?
- What kind of processing should i make available, apart from Xover (Buuteworth/Linkwitz-Riley/Bessel up to 48dB/oct), PEQ, shelving filters, delays and gains?
- What form should the project take? Me doing the whole stuff in a closed and controlled manner, or releasing an open-source framework?
- Whatever comes to your mind.
Here are some details about the board itself:
An externally hosted image should be here but it was not working when we last tested it.
- The DSP is a Freescale DSP56321 clocked at 275 MHz. It's capable of 275 MMACS, has a coprocessor and boasts plenty of RAM, enough to make a 4 second delay line at CD sampling frequency. The DSP is 24-bit fix-point with 56-bit accumulators. It's been a classic architecture for years in DSP for audio processing, and i guess that many of you have some custom code for it.
275 MMACS is A LOT! Enough for steep crossover filters, countless PEQ or FIR filtering.
- There's a USB connection allowing transfers of data to/from PC up to 8Mbit/s (USB High-Speed). The controller chip is manufactured by FTDI.
- There's a codec onboard with 2 analog inputs and 6 analog outputs. The chip is by Analog Devices ref. AD1838A. ADCs and DACs have both over 100 dB dynamic range, inputs and outputs are balanced or unbalanced. There are plenty of clock options, with 2 quartz onboard switcheable from the DSP.
- There's an expansion connector for separate I2S in and out, SPI and GPIO. Everything is there if you want to drive your own ADCs/DACs/SPDIF tranceivers.
- There's an other port with 5 outputs and 3 inputs for leds and pushbuttons.
- JTAG is also onboard if you wish to write your own code.
- The board requires 5V and +/-12V. The rest of the power supply is included on board with quality TI regulators. I can supply the matching SMPS.
- Finally, it's a high-quality board with 6 layers, vast ground planes and branded chips.
Best!
chaparK
Last edited:
looks like a very interesting board! You might want to get one of the moderators to move the thread to the "swap meet" section though as you are talking about sales. Would be nice to have a 3 or 4 way DSP crossover with linux opensource software for setting up 😉 Nice to have all balanced connections!
col.
col.
Thanks col for your comment, the boards are pretty nice indeed.
I just checked the data relative to the codec: dynamic range is 105 dB for the ADCs, and 108 dB for the DACs (A-weighted). Pretty usable.
The pic shows the prototype. Production boards are red and have screw terminals for the power supply leads.
So far i'm not selling anything. Just gathering suggestions and interest. It's not a topic for selling anything, but for discussing what would be cool to have in the software 🙂
yeah the board looks great! Iv'e been watching the miniDSP stuff a bit and would have probably invested long ago if it wasn't for the fact that there is no balanced connectivity and the plugin software runs on windows. What about making the setup software using something that is multi-platform like Java? Also, for pro audio use it would be better if it was 4 way.
col.
col.
Ok so let's start the list of features:
- 4-way crossovers
- Multi-platform
I have to say that i had in mind to make a Windows app as well. But you're right: the app should be multi-platform. Don't like Java though. Maybe Python? I know there's a Python wrapper around the FTDI API for USB com. Haven't tested it though.
I previously posted the following wish-list in another thread (so not all relevant to your hardware):
How many boards do you have, and at what cost?
Bohdan, the author of SoundEasy loudspeaker design software, wants to partner up with someone who can provide the DSP hardware for his latest DSP utilities. See SoundEasy product page.
I'm very much interested! Good luck with this project.
How many boards do you have, and at what cost?
Bohdan, the author of SoundEasy loudspeaker design software, wants to partner up with someone who can provide the DSP hardware for his latest DSP utilities. See SoundEasy product page.
I'm very much interested! Good luck with this project.
very interesting indeed. do you have a rough idea of what it will sell for?
ive just bought a miniDSP + MiniDigi board to play with, and so far im fairly impressed, though its not without drawbacks. the first being (as col says) only windows interface - luckily i do have a windows box to program it, but ideally i'd want something i can run from a mac (which is my audio workstation). the interface is very pretty, but very 'laggy' when running live to the board (no big deal, set it up offline then sync) also, if you are going to make a pretty graphical interface, then it ought to be editable graphically - not just from edit boxes. that said, give me raw functionality over pretty interface any day.
the ability to edit the parameters of say, parametric EQ's, live - so if you are eq'ing a system, you dont have to edit the setup file, download the board, recheck your freq response, edit params, download, etc. its important (imho) to be able to make changes in real time. this is something you can do with the miniDSP
the miniDSP 'plugins' system is a bit weird. I'd rather just pay more money upfront and have all the options of all the plugins available as settings, rather than be limited to whichever particular optionsets they decide they will include in a given plugin. eg the ability to choose whether to use a graphic eq or a peq at a given point in the signal chain.
one bit of *essential* functionality missing from the miniDSP is output limiters.
anyway, count me in as interested (if the price isnt too high)
I second Col in saying Hooray for Balanced ins and outs...
cheers,
ben
ive just bought a miniDSP + MiniDigi board to play with, and so far im fairly impressed, though its not without drawbacks. the first being (as col says) only windows interface - luckily i do have a windows box to program it, but ideally i'd want something i can run from a mac (which is my audio workstation). the interface is very pretty, but very 'laggy' when running live to the board (no big deal, set it up offline then sync) also, if you are going to make a pretty graphical interface, then it ought to be editable graphically - not just from edit boxes. that said, give me raw functionality over pretty interface any day.
the ability to edit the parameters of say, parametric EQ's, live - so if you are eq'ing a system, you dont have to edit the setup file, download the board, recheck your freq response, edit params, download, etc. its important (imho) to be able to make changes in real time. this is something you can do with the miniDSP
the miniDSP 'plugins' system is a bit weird. I'd rather just pay more money upfront and have all the options of all the plugins available as settings, rather than be limited to whichever particular optionsets they decide they will include in a given plugin. eg the ability to choose whether to use a graphic eq or a peq at a given point in the signal chain.
one bit of *essential* functionality missing from the miniDSP is output limiters.
anyway, count me in as interested (if the price isnt too high)
I second Col in saying Hooray for Balanced ins and outs...
cheers,
ben
Python even better! There are a lot of people that can program in python and if the software is opensource you would hopefully get programmers contributing.
col.
No, it is just the factors that were foremost in my mind at the time, not in order of preference. I had been investigating the requirements for a DSP-based active crossover for my Meridian D6000 speakers which I bought minus the DSP board. I have actually purchased the DCN23 from GroundSound. I have yet to build the amplifiers, though.
I am very much pumped about DSP audio since I started enquiring about them.
Thanks Shaun for your inputs.
In your list above, i've put between brackets the things that i believe do not apply to the current project. Indeed, for example Analogue Volume is a nice feature, but the board is already there and as such it's too late. Please correct if i misunderstood a requirement.
Also if you feel like, please check again your list and remove what doesn't apply to the existing hardware before i incorporate the requirements in the list i've started with Col 🙂
Cheers,
chaparK
Thanks for your post Ben, it's cool to have you guys giving directions.
I believe i understand your comment about 'laggy' interfaces: you turn a knob and the effect takes place one second later. I hate that, and want real-time control: you change a seeting and the effect takes place immediately. I think it's specially important for prototyping filters, where one wants to concentrate and listen to the signal while sweeping a parameter.
Your second point about editing parameters live: do you mean that the processor would run standalone, then you would plug in the USB socket and edit the parameters without the sound interrupting?
Finally, output limiters. I was just wondering why you would use these in a home sound setup. As far as i know, limiters in public shows are used to protect the speakers and to make sure the DJ doesn't go beyond a certain limit. Why would you use limiters? (just asking 🙂 )
Hi chaparK
No problem; I stated that not all of of the list I posted was applicable to your existing hardware. It was just a general wish-list.
So that will leave:
1. high-pass filters
2. low-pass filters
3. shelving filters
4. Notch filters
5. Attenuators
6. Arbitrary slopes for all filters
7. Time delays
9. Equalisation, with selectable preset/programmed filters
15. Digital or dot matrix display of system status (volume levels, inputs and muted channels).
16. Control outputs to turn on/off/mute power amplifiers.
This is still an impressive list.
Some of the functions, such as digital room correction (now removed from the list), can be done off-board. In fact, I would imagine that all design processing would be done in a computer application then uploaded to the device (do you have a name for it yet?).
PS. I see you've removed dithering. I reckon this a fairly low-level requirement that would be already incorporated in the DSP processing? Please enlighten me if I'm wrong.
No problem; I stated that not all of of the list I posted was applicable to your existing hardware. It was just a general wish-list.
So that will leave:
1. high-pass filters
2. low-pass filters
3. shelving filters
4. Notch filters
5. Attenuators
6. Arbitrary slopes for all filters
7. Time delays
9. Equalisation, with selectable preset/programmed filters
15. Digital or dot matrix display of system status (volume levels, inputs and muted channels).
16. Control outputs to turn on/off/mute power amplifiers.
This is still an impressive list.
Some of the functions, such as digital room correction (now removed from the list), can be done off-board. In fact, I would imagine that all design processing would be done in a computer application then uploaded to the device (do you have a name for it yet?).
PS. I see you've removed dithering. I reckon this a fairly low-level requirement that would be already incorporated in the DSP processing? Please enlighten me if I'm wrong.
Last edited:
As i said, these boards are left over form a past project. I have about 50 of them sitting in a box, and i'd be glad to see them being used in a nice audio project because they deserve it. I'm not going to run a new production, and i'm not willing to make any sort of profit.
I hope moderators aren't going to be upset if we briefly talk about $$$, otherwise the topic's lacking of pragmatism. I'm happy to resell the boards for what they cost, which is about USD 130 each. The processor alone costs USD 46.50 on a budgetary price on Freescale's website. The power supply would be priced USD 20, coming with the proper cable with Molex connector as well as a perforated steel cover against EMI emissions.
At this price, i'd make available the application + firmware including most of your realistic requirements, then hopefully hand it over in an open-source project to the 'community' so that it can keep growing and include more options.
What do you think about it?
chaparK
List of features updated:
- 4-way crossovers
- Multi-platform (Python?)
- high-pass filters
- low-pass filters
- shelving filters
- Notch filters
- Attenuators
- Arbitrary slopes for all filters
- Time delays
- Equalisation, with selectable preset/programmed filters
- Digital or dot matrix display of system status (volume levels, inputs and muted channels).
- Control outputs to turn on/off/mute power amplifiers.
- 4-way crossovers
- Multi-platform (Python?)
- high-pass filters
- low-pass filters
- shelving filters
- Notch filters
- Attenuators
- Arbitrary slopes for all filters
- Time delays
- Equalisation, with selectable preset/programmed filters
- Digital or dot matrix display of system status (volume levels, inputs and muted channels).
- Control outputs to turn on/off/mute power amplifiers.
The price is good. I do, however, think that the limited available quantity will in turn limit the likelihood of open-source activity to follow. But that is neither here nor there, as long as core functionalities are available.
Do you have the features list complete yet? Edit: heh, I spoke too soon.
Do you have the features list complete yet? Edit: heh, I spoke too soon.
Last edited:
Shaun, i would use dithering if i'm changing the resolution of a digital stream. For example, i'm recording and mixing 24-bit audio tracks, and want to export the master to a CD. In this case, i need to round 24-bit words into 16-bit words, then apply some dithering before writing the tracks onto the CD.
I'm not an expert about dithering, but i wouldn't add any to this project.
Understood. I guess the 56-bit accumulator also obviates any resolution/trunkation error concerns...?
I'm also very very interested!
My primary interest is in a stereo 3 way crossover that can do:
- Low latency (possibly conflicting requirement with FIR?)
- Crossover and EQ with IIR filters, FIR filter available to correct phase.
- Output channel limiters would be nice.
- Optional SPDIF input/output - I see this would be user addable via I2S.
- Interface software that allows importing of measurements (impulse response or .frd), then visualizing/optimizing filters. Output level meters would be nice too.
- Presets (full system) selectable by buttons
- Digital volume control via rotary encoder or analog pot.
And of course open source firmware software would be a HUGE benefit.
Some questions:
- Is USB audio streaming possible?
- Are free compilers available for the firmware?
- Would you consider doing another run of boards? Or open sourcing the PCB design? I'd be getting in on the first sale, but if this was an ongoing product more people would be interested in contributing to the firmware.
Consider me ready to drop cash 😀
My primary interest is in a stereo 3 way crossover that can do:
- Low latency (possibly conflicting requirement with FIR?)
- Crossover and EQ with IIR filters, FIR filter available to correct phase.
- Output channel limiters would be nice.
- Optional SPDIF input/output - I see this would be user addable via I2S.
- Interface software that allows importing of measurements (impulse response or .frd), then visualizing/optimizing filters. Output level meters would be nice too.
- Presets (full system) selectable by buttons
- Digital volume control via rotary encoder or analog pot.
And of course open source firmware software would be a HUGE benefit.
Some questions:
- Is USB audio streaming possible?
- Are free compilers available for the firmware?
- Would you consider doing another run of boards? Or open sourcing the PCB design? I'd be getting in on the first sale, but if this was an ongoing product more people would be interested in contributing to the firmware.
Consider me ready to drop cash 😀
Basically, the idea is to perform all operations with double precision (2 x 24 bits into 56-bit accumulators).Then you don't need to worry much about the noise induced by the algorithms.
In the final stage, you're truncating 48-bit words into 24-bit ones. Considering that the theoretical dynamic range of 24-bit words is 6 x 24 = 144 dB, and that the DACs have 108 dB dynamic range, there's no need to add dithering as it would be lost in the DAC noise. (Dithering applies only to the lower bits).
- Status
- Not open for further replies.