DMX splitter circuit

[david.elsbury]

Hey guys

Does anyone have a circuit diagram (PCB layout a bonus) of a DMX splitter that they can share? Looking to build a couple for myself.

 

Thanks

David

[musht]

Isolated on input, not between outputs:

 

http://users.skynet.be/kristofnys/files/bo.../dmxbooster.pdf

 

If somone could remember the Pulsar DMX protection setup it would be a handy addition to the above circuit

[Dmills]

The following is a quick and dirty, "get you out of trouble", non isolated, PASSIVE trick.

 

You can build a passive split for DMX and have everything see the correct impedance values as follows:

Take a small diecast box and fit one input and two output connectors (either 3 or 5 pin depending on need).

Wire all the pin ones together.

Connect a 56 ohm resistor between each output pin two and the input pin two and repeat for pin 3 (4 resistors total).

 

With both outputs correctly terminated, each output will see a signal 6db down on the input, but as RS485 specifies that the receivers work down to 100mV and the driver outputs 5V that still leaves a healthy margin.

 

You can optionally add a bicolour led and push button to verify signal present at one or both outputs, and a polarity reverse switch can be handy for dealing with those 'not quite DMX' moments.

 

Optionally polyswitch auto reset fuses and transorbs could be added to give some protection from short circuits.

 

I commend this device to your toolkits.

 

Regards, Dan.

[martinw]

http://users.skynet.be/kristofnys/files/bo.../dmxbooster.pdf

The receiver in this circuit is of a very poor design, and whilst it might work, it would degrade the signal such that other devices connected to this part of the network might not receive correctly. The receiver should use the same 75176 chip that the transmitter circuits use, connected between the cable and the optoisolator, powered by a DC-DC convertor.

 

For ESD protection, 12V bi-directional transient suppressors (sold under the brands Transzorb or Transil) should be connected between data+ and GND and between data - and GND.

 

Martin

[dbuckley]

For ESD protection, 12V bi-directional transient suppressors (sold under the brands Transzorb or Transil) should be connected between data+ and GND and between data - and GND.

 

Thats what I thought too, until I read this application note on the Maxim website that says

 

Maxim has invested a substantial effort in developing ICs with internal ESD protection. Starting with RS-232 and RS-485 interface ICs, these protected devices now include several analog switches and the MAX681_ family of switch debouncers. All withstand the application of IEC 1000-4-2 ESD events directly to their I/O pins. Maxim believes this is the best way to control ESD in a system. It is robust, readily available, requires no external real estate, and costs less than most alternatives.

[martinw]

Maxim has invested a substantial effort in developing ICs with internal ESD protection. Starting with RS-232 and RS-485 interface ICs, these protected devices now include several analog switches and the MAX681_ family of switch debouncers. All withstand the application of IEC 1000-4-2 ESD events directly to their I/O pins. Maxim believes this is the best way to control ESD in a system. It is robust, readily available, requires no external real estate, and costs less than most alternatives.

 

Be careful if you decide to use specific Maxim parts, as they have a bad reputation for availability, and you might find that having completed a design, you suddenly discover that the parts you selected are now on 16 weeks lead time, for a minimum order of several thousand. The best approach would be to design a board using a 75176 and Transzorbs, but find a nice Maxim transceiver with ESD protection and slew rate limiting which is pin-compatible with the 75176. That way, you have other options should the Maxim part go AWOL.

 

Martin

[peternewman]

A bit of a bump to this. For a couple of projects I could do with building some small, simple two or three way DMX splitters, mainly for various systems that send the DMX down a proprietary cable system, e.g. RJ45, where it would be nice to not have to loop out the end, and some where looping just isn't possible. Anyway I've had a look around on the net (for some reason this didn't come up on my first Google) and found the Kristof circuit mentioned above, as well as Ujjal's (Word file in a Zip) (from here). Anyway Ujjal's looks a lot simpler. Presumably if I also removed the 120 Ohm resistor on the input I could link out of it too if needed, giving the link and two buffered outputs? I'm not really bothered about protection or anything, as theoretically the units would just be looped anyway, the splitter is primarily for ease. My main aims are small, cheap and simple.

 

So any comments/improvements on Ujjal's, any other alternative circuits?

[Jivemaster]

Beyond the passive cct, I'd be looking at a bought unit. DJDicky on MDJF is a dealer importing things, he had a 4 way splitter for £60 odd recently all isolated, Link later!

 

Most active splits need power supplying.

[peternewman]

Are you talking about Dmills passive one? Presumably while I could daisy chain a few of these, using too many wouldn't be advisable? I'm somewhat hoping to avoid purchasing traditional ones, for a few reasons, firstly they are quite big and bulky, I'm ideally hoping to install these in the end of the fixtures if there is space, or if not in a small project box or similar just nearby, I'm therefore hoping for something that won't look like an eyesore on a truss or stage. They are also expensive, ideally I'd like a splitter per fixture so I can loop them traditionally if needed, whereas at £60 each this seems a bit pricey, especially given I can purchase the necessary components for about £5-7.50. Also as mentioned the isolation isn't too much of an issue.

 

Being a bit facetious for a second, surely by the nature of the term "active" implies needing power.

 

I should perhaps have noted previously, I've got 12 or 24V DC available at the fixture, so can quite easy power a splitter if needed.

 

So any comments on the alternative splitter circuit I listed, presumably its input stage is better than the Krystoff one? Can I just substitute the input stages?

[boatman]

Ujjal's circuit is OK with one exception. There should be a 120R source termination on each output. This is standard practice for all RS485 networks and prevents back reflections at the source end should the destination end be left unterminated.

 

Yes, you can leave out the 120R termination on the input as long as you provide an external loop-through XLR where a termination can be fitted. If you want galvanic isolation the simplest solution is to use the IL485. Although it seems a bit expensive, if you add up the cost of opto-isolators and DC-DC converters plus the extra PCB space it's cheap.

 

If anyone is interested I'll knock up a layout and put up the Gerber files somewhere, so you can get your own PCBs made.

[Dmills]

Are you talking about Dmills passive one? Presumably while I could daisy chain a few of these, using too many wouldn't be advisable?

 

Indeed, one seems to be fine, a second passive split also seems to be ok, but I am not really happy and a third passive split starts making me really nervous (but still works).

As a quick get out of trouble for the toolbox, highly recommended, as a large multi-way DMX distro, forget it.

 

Regards, Dan.

[peternewman]

Ujjal's circuit is OK with one exception. There should be a 120R source termination on each output. This is standard practice for all RS485 networks and prevents back reflections at the source end should the destination end be left unterminated.

Presumably given the receive of the 75176 isn't connected to anything, back reflections can't actually be an issue? Also I doubt I will, given it could raise many other potential issues, such as having to terminate unused outputs, but could you leave the 120R source terminator off and provide two legs, so the transmitter is at the source of a Y?

 

If anyone is interested I'll knock up a layout and put up the Gerber files somewhere, so you can get your own PCBs made.

I may be building a few of these, 10 max probably, any suggestions for places to get PCBs made up (although I guess we should split it into a new topic), I'd rather not build them all on stripboard.

 

Anyway I've drawn up two circuits with the comments from everyone included, would people mind taking a look. A few questions, should the DMX signal ground be commoned to the 0V ground or kept separate (looking at Kristof's I imagine the former)? I assume the capacitors are for smoothing, I just copied Ujjal's design, presumably the values will be fine? Although I'm confused why you need one per chip, surely just one directly across positive and negative would be fine?

 

Is the splitter circuit with buffers correct? Their use was a complete guess although I can't think of many other ways you could connect them up.

[Modge]

For runs that small pcbpool are all good. We've used them successfully for prototyping.

http://www.pcb-pool.com/ppuk/info.html

[boatman]

Ujjal's circuit is OK with one exception. There should be a 120R source termination on each output. This is standard practice for all RS485 networks and prevents back reflections at the source end should the destination end be left unterminated.

Presumably given the receive of the 75176 isn't connected to anything, back reflections can't actually be an issue? Also I doubt I will, given it could raise many other potential issues, such as having to terminate unused outputs, but could you leave the 120R source terminator off and provide two legs, so the transmitter is at the source of a Y?

 

Yes that would be OK, but you would have to be sure that both ends of the Y network are terminated.

 

If anyone is interested I'll knock up a layout and put up the Gerber files somewhere, so you can get your own PCBs made.

I may be building a few of these, 10 max probably, any suggestions for places to get PCBs made up (although I guess we should split it into a new topic), I'd rather not build them all on stripboard.

 

Anyway I've drawn up two circuits with the comments from everyone included, would people mind taking a look. A few questions, should the DMX signal ground be commoned to the 0V ground or kept separate (looking at Kristof's I imagine the former)? I assume the capacitors are for smoothing, I just copied Ujjal's design, presumably the values will be fine? Although I'm confused why you need one per chip, surely just one directly across positive and negative would be fine?

 

Is the splitter circuit with buffers correct? Their use was a complete guess although I can't think of many other ways you could connect them up.

 

It looks OK except for a couple of minor points. You shouldn't use a 7407 as it's too slow, but a higher speed device such as 74HC08 or 74HC32 with the two inputs of one gate paralleled. Also don't use separate gates as a single one will easily drive four 75176 output chips.

 

 

As promised I have laid out a PCB. Here's a link to the schematic here. There are several things of which you should be aware.

 

1. I have used the IL485 in the schematic, but any of these can be substituted with a standard MAX485/75176 and a couple of zero ohm links.

 

2. All of the components are surface mount but the caps and resistors are 0805 so easy to solder.

 

3. The layout is completely untested.

 

If you are interested in making some boards, send me a PM with an email address and I'll send back a zip file with all the Gerbers, a drill file and a parts list.

 

Edited: for SPaG.

[church]

You should use an Opto isolator on the input this provides protection. I have built a number of splitters and they are straightforward to design and build. I always use opto isolators to protect other equipment should there be a major equipment failure somewhere. Saves taking out everything on the universe.