DMX512 Multiplexer PCB

[Luke]

Does anyone have or know where I can get a schematic (with componant overlay) or PCB for a DMX512 Multiplexer. It's a long shot, but worth it.

 

Regards,

 

Luke.

[Tomo]

Do you mean an analogue-to-DMX or a DMX-to-analogue converter box?

 

Or something else entirely?

[Luke]

I need the PCB layout of a DMX Multiplexer for inside of a Lighting Console that I want to build. I guess this isn't available?

[Brian]

Does anyone have or know where I can get a schematic (with componant overlay) or PCB for a DMX512 Multiplexer. It's a long shot, but worth it.

 

Ready built PCB from Soundlight

 

Order code 8112A-EP or 8116A-EP

[Luke]

Ohh.. lovely. I was ideally looking for a free/shareware version, in a sense. By which I mean an open source PCB for free use throughout projects. I've done googles - perhaps I'm looking for the wrong thing.

[Tomo]

As far as I know, open source ones don't exist.

Mostly because there are very few people interested in such a device.

 

Anyway - they would actually cost you quite a lot to produce - they require a channel of ADC chip for every input channel (~£1 ea), suitable microcontroller/FPGA to do the conversions (£10-BigBucks), not mentioning etching a PCB.

 

Etching one-off PCBs is fairly expensive, and the programmers for most microcontrollers aren't cheap either.

 

If you don't have any of the equipment, you can expect it to cost well over £200 just to get the equipment to start.

[Luke]

I have, or know people that I can borrow from, all the equipment that would be needed to etch a PCB successfully and program microcontrollers and PIC chips and stuff.

 

I guess I could open a console up and try and re-design the DMX board?

[Modge]

It's not actaully that hard to get an FPGA \ pic to produce DMX - it doesn't need any error checking or sillyness.

 

As for redesigning - if you have a good idea of what the rest of the board is feeding it then it could be possible I guess. There's a £12 pic from microchip, 20MHz lots (probably not enougn for every fader, but lots) of A/D's on board. You could maybe have as many of them as you need to have enougn A/D's linked togther and have one of them output DMX.

[Tomo]

It's not actaully that hard to get an FPGA \ pic to produce DMX - it doesn't need any error checking or sillyness.

 

As for redesigning - if you have a good idea of what the rest of the board is feeding it then it could be possible I guess. There's a £12 pic from microchip, 20MHz lots (probably not enougn for every fader, but lots) of A/D's on board. You could maybe have as many of them as you need to have enougn A/D's linked togther and have one of them output DMX.

<{POST_SNAPBACK}>

Most of the PICs actually only have one ADC on board - however, you can connect it to several different inputs to read several different faders in turn.

 

Whether that's fast enough for you will depend on how fast you intend to squirt the DMX and how fast the particular PICs ADC runs at.

 

The design I have been working on uses a chip of Static RAM to contain the universe of DMX data, and there is a single PIC who's entire life is spent reading this RAM and squirting out the DMX signal.

 

Another processor writes the DMX values required into this chip (to prevent data collisions a flag is set to indicate that the RAM is busy whenever the transmitting PIC is reading data). So if the desk itself gets very loaded, the DMX update frequency does not change (much - there may be an occasional long stop bit during bus contention), which should prevent any problems with fixtures getting confused or think they have lost the signal.

 

In your case you will probably be best using an I2C or SPI bus (or other serial protocol) with the ADCs on it - in theory this could be easily expandable using addressed ADCs.

Choosing free-running ADCs which report the last-good conversion will mean that the within-desk serial link latency issues can be ignored, and a single PIC can do the whole lot very easily.

 

Implementing this is left as an exercise for the student. :blink:

[Dave]

Anyway - they would actually cost you quite a lot to produce - they require a channel of ADC chip for every input channel (~£1 ea), suitable microcontroller/FPGA to do the conversions (£10-BigBucks), not mentioning etching a PCB.

<{POST_SNAPBACK}>

As Tomo later hinted at, you don't actually need an ADC per channel. The standard way is to use an analogue multiplexer ahead of the ADC and "scan" through the channels sequentially.

 

Incidentally, all the demultiplexers I've ever seen use this technique in reverse - DAC->analogue demultiplexer->sample and hold (ie capacitor and buffer).

[dbuckley]

For the student who likes the work done for him, have a look at Midibox at http://www.ucapps.de, where one can get designs for a PIC controller board, and analog and digital I/O boards. It is true that the project as a whole is aimed at MIDI rather than DMX512, but adding an RS485 interface is a trivial bit of veroboarding, and the software is structured to permit changes.

 

Who knows, with a bit of work this could enable a whole new category of home made lighting boards :-)

[jfitzpat]

It's not actaully that hard to get an FPGA \ pic to produce DMX - it doesn't need any error checking or sillyness.

 

As for redesigning - if you have a good idea of what the rest of the board is feeding it then it could be possible I guess. There's a £12 pic from microchip, 20MHz lots (probably not enougn for every fader, but lots) of A/D's on board. You could maybe have as many of them as you need to have enougn A/D's linked togther and have one of them output DMX.

<{POST_SNAPBACK}>

Most of the PICs actually only have one ADC on board - however, you can connect it to several different inputs to read several different faders in turn.

 

Whether that's fast enough for you will depend on how fast you intend to squirt the DMX and how fast the particular PICs ADC runs at.

 

The design I have been working on uses a chip of Static RAM to contain the universe of DMX data, and there is a single PIC who's entire life is spent reading this RAM and squirting out the DMX signal.

 

Another processor writes the DMX values required into this chip (to prevent data collisions a flag is set to indicate that the RAM is busy whenever the transmitting PIC is reading data). So if the desk itself gets very loaded, the DMX update frequency does not change (much - there may be an occasional long stop bit during bus contention), which should prevent any problems with fixtures getting confused or think they have lost the signal.

 

In your case you will probably be best using an I2C or SPI bus (or other serial protocol) with the ADCs on it - in theory this could be easily expandable using addressed ADCs.

Choosing free-running ADCs which report the last-good conversion will mean that the within-desk serial link latency issues can be ignored, and a single PIC can do the whole lot very easily.

 

Implementing this is left as an exercise for the student.

<{POST_SNAPBACK}>

 

As someone mentioned above, a straight analog to DMX mux could be done with an analog mux, an ADC, and a Xilinx or Altera gate array. With plenty of low cost ADCs conversion speed would not be a problem. In production quantities, this would be a pretty low cost solution. For a one off, the learning curve and possible cost of the dev tools would probably be too high.

 

FWIW, in your own design, if you are going PIC->ram->PIC, say to get from USB to repetively streaming DMX, you might want to consider Silicon Labs, Atmel, etc. You could probably find a one chip solution, maybe even 1:2 or 1:4. Because of competition, the dev tool buy in for many of these families is pretty low.

 

The biggest hurdles are probably voltage and packaging. Aside from the dev tool costs, one reason that Microchip has remained so popular with built-it-yourself types is you can still get through-hole, or at least socketable packaging.

 

-jjf