DMX distribution idea

[Chris15]

Originally Posted by Tex

I'm assuming that your patch panel is comprised of XLR connectors? Mine is 1/4" stereo phone plugs that do not allow DMX to pass through.

Originally Posted by epimetheus

but a 1/4" TRS has all the same connections as an XLR, unless your patchbay has common shields or something like that.

Originally Posted by JD

Be aware that most "pulled" audio cable (what's in the conduit) is 64 ohm. (Often Belden 8761) DMX will work as mic cable, but mic cable should not be used for DMX. Pulling DMX would be a good idea as you can splice new to old and then use the old cable to pull the new one through. (be careful!) Whenever splitting the run, use an active opto-splitter. Each output line should then be terminated. As for using stereo phone jacks for DMX, sounds like a good way to static spike the line and damage something. With XLR cables, pin #1 is designed to make contact first, which drains off any static charge. Phone Jacks hit tip first, so any stray voltage will spike the actual data line.

Originally Posted by epimetheus

Good point about the tip making contact first. I had thought of that. If you were to try to run DMX through a 1/4" patchbay, I certainly wouldn't try to make a connection with the system powered up.

Ok, considering a number of things;
To say mic line is 64 ohm is likely wishful thinking. Until digital came along, most manufacturers might list an impedance on the data sheet, but the actual impedance is probably best described as whatever came off the machine...
AES grade audio cable is in general suitable for DMX usage (110R nominal Z).
Patch bays - I don't recall seeing a 6.5mm patch bay in some time that used standard plugs. They normally use longframe plugs that have tips that are smaller and so don't suffer the shorting on insertion syndrome as much. Agreed that in general, patching whilst non live is good. The reality is that you only need 2 conductors for DMX, it's a differential balanced line and industrial control systems run RS485 (which is the underlying transport of DMX-512A) for hundreds of metres without shielding...

Oh and common grounds on anything digital are bad news. AES audio suffers badly when attempted to run on common grounded multicores...

[epimetheus]

Originally Posted by Chris15

Ok, considering a number of things;
To say mic line is 64 ohm is likely wishful thinking. Until digital came along, most manufacturers might list an impedance on the data sheet, but the actual impedance is probably best described as whatever came off the machine...
AES grade audio cable is in general suitable for DMX usage (110R nominal Z).
Patch bays - I don't recall seeing a 6.5mm patch bay in some time that used standard plugs. They normally use longframe plugs that have tips that are smaller and so don't suffer the shorting on insertion syndrome as much. Agreed that in general, patching whilst non live is good. The reality is that you only need 2 conductors for DMX, it's a differential balanced line and industrial control systems run RS485 (which is the underlying transport of DMX-512A) for hundreds of metres without shielding...

Oh and common grounds on anything digital are bad news. AES audio suffers badly when attempted to run on common grounded multicores...

The 3rd (5th) wire (the bare wire) is needed for DMX or any other RS-485 signal. This wire provides the ground reference for the balanced signal, without it the signal plus and minus can pick up a voltage offset that will fry an input that isn't properly isolated. This is more of a concern for longer runs than it is for short runs.

BTW, industrial applications do run RS-485/422 for hundreds of meters, but not with shielding/ground wire. Industrial RS-485/422 installations consist of either +, -, and G or RX+, RX-, TX+, TX-, and G. The former, a 2-wire 485 system, is only half-duplex, i.e. one way communication at a time. It is bidirectional, but not at the same time. The latter, a 4-wire 422 system, is full duplex, i.e. bidirectional at the same time, since transmit and receive do not share the differential pair as in 485.

[JD]

Originally Posted by epimetheus

The 3rd (5th) wire (the bare wire) is needed for DMX or any other RS-485 signal. This wire provides the ground reference for the balanced signal, without it the signal plus and minus can pick up a voltage offset that will fry an input that isn't properly isolated. This is more of a concern for longer runs than it is for short runs.

Correct!
In practice, the RS485 transceiver chip needs ground. It can only compare the + and - inputs if those signals are between its own supply rails (0 and 5 volts) Outside of that range, the output will be random regardless of what ratio is between the data lines. The "ground" line, or actually common, connects the electrical commons (0 volt) on the circuit boards in all of the units. This is also why opto-couplers are so good. The "ground" or pin 1 on the input does not have to be the same as the ground on the output (thus the "opto" part), but the proper reference is maintained on both input and output legs.

Proper operation is best achieved when the + and - data lines are between 0.75 volts and 4.25 volts with reference to ground. This is due to the fact that the transistor junctions on the chip require a junction drop voltage. If the + has a higher voltage then the -, the chip will output a "1". If the - has a higher voltage then the +, the chip will output a "0".
Some examples:
"+" = 3.5 volts "-" = 3.3 volts, output = 1
"+" = 2.2 volts "-" = 3.3 volts, output = 0
"+" = 0.1 volts "-" = 0.3 volts, output = arbitrary
"+" = 4.8 volts "-" = 4.9 volts, output = arbitrary
"+" = 4.5 volts "-" = 0.5 volts, output = arbitrary
"+" = 4 volts "-" = 1 volts, output = 1
In the third, fourth, and fifth cases, the voltages are too close to the chip supply rails for the chip to operate properly.

[Knave]

Originally Posted by BillESC

Just a quick note about running digital signals through installed house audio cables.

We were contracted to provide lighting and effects for a dance recital in the brand new 4 plus million dollar auditorium at West Point High School, West Point, VA.

To make things easy I brought a half dozen NSI 4 channel dimmer packs to plug the effects into and plugged the microplex into an audio channel behind the procenium wall. As soon as I powered up the packs, all kinds of things started to blink, wink, strobe, etc.

After trouble shooting the situation, I discovered all the the house audio were run with just two conductors. Nice job audio contractor.

I've had similar experiences. It was the shared ground on the patch panel in our case, but it was disappointing at any rate.

If it works and is reliable enough to run your show, awesome!
Don't count on it when you're planning a show, though.

A spool of Cat5 and some time soldering is far cheaper than a show that's missing critical effects/fixtures, or worse a fog machine that goes off at the wrong time, a mover that can't be controlled...

(If I had a money tree I'd just keep a wireless DMX system in my toolkit for scenarios like this. I bet they aren't too expensive to rent, though.)

[Les]

Originally Posted by Knave

A spool of Cat5 and some time soldering is far cheaper than a show that's missing critical effects/fixtures, or worse a fog machine that goes off at the wrong time, a mover that can't be controlled...

Very well said.

[FMEng]

If one of those mic jacks doesn't get properly patched, and the audio console gets turned on, 48V phantom power could be applied to a DMX connection on the equipment. If that happens, there will be smoke, but it won't be coming from the fog machine's normal place.

This is especially dangerous if the patchbay is normalled to the console, because connections are made when no patch cords are inserted.

[Chris15]

I rescind my earlier comments about the optional nature of ground for DMX. A truly balanced system ought to be able to be run at any common mode voltage, without a common reference, with the signal being conveyed purely by the difference of the signals relative to one another. It seems RS 485 is not truly balanced...

[JD]

Originally Posted by Chris15

It seems RS 485 is not truly balanced...

You are correct! Example, a balanced microphone line. Doesn't matter if there is an offset voltage. In fact, that is exactly what the phantom power is. DMX is not a true balanced system as DC can throw it off. In fact, one of the more common chips (SN75176) ( http://focus.ti.com/lit/ds/symlink/sn75176b.pdf ) input toggles at 2.3 volts with reference to ground. in other words if "-" is 2.4v and "+" is 3.5 volts, the output will still be arbitrary! To build a true balanced system you would encounter problems when the stream hits a long run of "0"'s or "1"'s in that DC decoupling would cause drift. It could be done, but the electronics involved would kill the cost.

[Chris15]

Originally Posted by JD

To build a true balanced system you would encounter problems when the stream hits a long run of "0"'s or "1"'s in that DC decoupling would cause drift. It could be done, but the electronics involved would kill the cost.

I tend to disagree... You just need to change the encoding method. Look at the way AES audio uses Biphase Mark Coding - to produce a] clock recovery via the data stream, b] zero DC component - ground IS optional on AES.and c] polarity neutrality, you can transpose hot and cold without issues...

[JD]

Originally Posted by Chris15

I tend to disagree... You just need to change the encoding method.

Then it would no longer be the DMX standard. We are stuck with what we have.