Thought I'd throw another design out for anyone who is interested. We'd talked about this a few months ago, but it took longer than I expected to have boards made.
Featuring:
- DMX in & through on XLR-5 connectors
- Totally isolated DMX receiver stage
- Optional: PIC16F88 processor
- Optional: 9 position DIP switch
- Screw terminals for +12V power supply
- RJ-45 jack + DMX transmitter for driving chains of pixels
- DMX output on 3-pin header for making other connections
- Serial output on a pinheader for debugging. If you use this, install a 300R resistor in series with your Tx line. Otherwise, there's a chance the PIC pin could be shorted to ground. If you're talking to a local computer, generate your serial stream 'inverted' and you'll be fine not using an RS-232 transmitter. Don't do this for more than a few feet of shielded cable, though.
The rationale behind this design is thus:
If you're using a standard RS-485 receiver, you're not supposed to chain more than 32 devices together. Too many devices will overload the transmitter & possibly corrupt the datastream.
Yes, there are receivers out there which present a much smaller load (1/8?) to the system. You could put 256 devices on a single line if you wanted. They also cost about 10x more than the bog-standard SN75176.
So in theory, 32 of these drivers could be connected to your master controller. And each driver could control 32 downstream devices...
The other interesting part of this design is the included processor & programming header.
If you wanted to, you could make a stand-alone light show. No computer required.
Or, program the processor to accept DMX input, then trigger specific effects on the downstream chain.
As always, the design is totally open. Schematic & bill of materials are posted here:
http://www.response-box.com/rgblights/2007offset.shtml
If you make something cool or noteworthy, be sure to share with the rest of us.
No, I don't have any code for this design at the moment.
JEC
Very interesting (as always) JEC... Being a network geek I would have used the term DMX Router instead of Offset Machine, 😉 but I get the point of it...
I have not figured out what I'd do with the PIC but boy it is interesting...
thanks
Tech Christmas wrote:Very interesting (as always) JEC... Being a network geek I would have used the term DMX Router instead of Offset Machine, 😉 but I get the point of it...
Thanks!
The original idea came from a guy in the UK who bought a few dozen pixels. He does theatrical lighting and frequently tours from place to place.
His idea was to hard-code a string of pixels to the addresses 1, 4, 7... etc.
Then, use the PIC & dip switches to set the start address for the entire chain. The PIC would receive an entire DMX packet, then shuffle and re-transmit only what's needed for the downstream pixels. Thus 'offsetting' 96 channels of data from where they were in the datastream to somewhere else.
The end result is that the pixels don't need to re-addressed for each show. He can just set a 'system address' which plays nicely with the other lighting equipment in the space.
I think the concept may be similar to 'NAT' in the networking world.
Being able to bypass the processor completely, or program a self-contained light show on-chip, is sort of a free-of-charge side benefit. Your programming time excepted, of course.
It's worth noting that the received data doesn't need to be DMX. It's set up for regular RS-485 (differential) reception. So long as you get the polarity correct on the transmitting end, the isolated receiver will spit out regular serial data to the PIC.
So if you wanted to send control data at 19,200, the processor could happily receive it, then dump DMX out the back door.
As always, the software is what would make this sort of project really shine.
It's probably even possible to make the system accept P.Short's Renard datastream. Many, many possibilities.
JEC wrote:Thanks!So if you wanted to send control data at 19,200, the processor could happily receive it, then dump DMX out the back door.
Eww, JEC you'll have everyone excited!!!
JEC wrote:
Optional: PIC16F88 processor
Optional: 9 position DIP switchNo, I don't have any code for this design at the moment.
JEC
If the PIC is optional and not even programmed at this point how do you tell offset one that it is to start at DMX address #1 thru 96 and unit 2 that it is to start at #97 thru 120 and sent out the back as 1 thru 24
confused..err joel
jrock64 wrote:If the PIC is optional and not even programmed at this point how do you tell offset one that it is to start at DMX address #1 thru 96 and unit 2 that it is to start at #97 thru 120 and sent out the back as 1 thru 24confused..err joel
I have stacks of pseudocode rattling around in my mind, but haven't taken the time to actually implement it.
I should have mentioned above that while I discussed the design with the UK customer, he didn't require the functionality immediately.
It's probably a day's worth of work to get the code stable, but I just haven't had the time.
So as it sits *this moment*, the system is more of a driver board + single-channel optosplitter. Great for expanding your available device count, but everything in line needs to have its own (world-visible) DMX address.
Tech Christmas wrote:Very interesting (as always) JEC... Being a network geek I would have used the term DMX Router instead of Offset Machine, 😉 but I get the point of it...I have not figured out what I'd do with the PIC but boy it is interesting...
thanks
Ahh another Network geek... What in particular? I liked the term DMX router, its kind of cute.. However we might need to decide if its actually a router, a switch or a bridge? What sort of networks do you do btw?
Let's not get off topic.
Have any of you programmed this into a "Self-Contained Show Controller" mode? Care to share some of your process, and comments on your success/difficulties?
I have a couple other question, more or less directed at JEC:
- The Offset Machine would replace the Data/Power Injector in that mode?
- Could such a system start running it's programming upon power-up, or would some sort of "start" signal be required?
- Do I understand correctly that one of these gizmos can drive up to 32 of your pixels?
- Could you daisy chain multiple Offset Machines and have all of them controlled by one of them running in this mode?
- 12V powers the board _and_ is delivered to the pixels through the RJ-45? I assume I'd still need to "double power" the string if it got larger than 20?
JEC wrote:I have stacks of pseudocode rattling around in my mind, but haven't taken the time to actually implement it.I should have mentioned above that while I discussed the design with the UK customer, he didn't require the functionality immediately.
It's probably a day's worth of work to get the code stable, but I just haven't had the time.
So as it sits *this moment*, the system is more of a driver board + single-channel optosplitter. Great for expanding your available device count, but everything in line needs to have its own (world-visible) DMX address.
Couple of updates here...
1. I re-spun the board, using an 18F4550 instead of the 16F88. More memory, etc. Am quite happy with the results and will post some photos shortly.
2. I have basic DMX-Routing code finished and stable.
The board has a 9-position DIP switch used to set the system's start address. The processor then receives and transmits DMX packets based on the DIP switch address.
That is, if you've set the address to 10, the system will receive an entire incoming packet. Then, the first nine channels are deleted and a truncated packet is re-transmitted. So incoming channel 10 is outgoing channel 1...
