Broken core, removing terminators and a dimmer that didn't do as i

[pritch]

I've just cleared a DMX fault - we've got five dimmers, the fifth one didn't respond to DMX with the terminator plugged in, but worked fine with the terminator removed. The problem only became apparent after I'd installed a long cable run from our control position to the dimmers. If I had the desk sitting right next to the dimmers with a shorter length of cable, all was OK.

 

I tracked it down to a broken core in the cable between the 4th and 5th dimmers (I think it was data +, but I could be wrong - it was definitely a data core, though). Quick bit of soldering iron wielding, and we were sorted.

 

My question is... why? Why did it work without the terminator? And why did it make a difference without the long cable run between desk and dimmer?

 

I know I don't need to know why this happened, but I'm just intrigued.

[boatman]

First of all, get a copy of this application note.

 

At the end you will see an eye-height diagram which shows the difference between a good signal and a bad one. The cable attenuates the signal to a certain degree and so a longer cable means less signal at the receiver. Removing the terminator will cause the signal level at the end of the cable to increase and that's why your dimmers worked again. Unfortunately, removing the terminator also causes reflections, which will probably not be a problem at the end of a cable run but will almost certainly cause problems to fixtures in the middle of the cable run. This is because the forward and reflected signals interfere and cause aliasing.

[JimWebber]

I cannot give you all the whys and wherefores, but basically The protocol DMX Uses, RS 422 (?) RS 423 (?) relies on differential signals.....

 

Ie) If Data +ve is high, and Data -ve is low, then the reciever sees a 1. If they are the other way round, then it will see a Zero.

 

Now, in normal operation, the driver (LX desk) is "randomly" sending + and - 5 (?) Volts down each leg of the line.

 

 

Now, assume there is no terminator in place:

 

If one of these lines becomes open circuit, it "floats" at around zero volts. However, there is usually enough of a differential between the two lines for the receiver to recognise it's ones and zeros.....

 

If there is a terminator (Basically resistor) across the lines, then as there is no current flow, (As one leg is disconnected) there is no voltage drop across the resistor, therefore, both legs are always at the same potential. - Therefore the receiver cannot distinguish between a 1 and a 0.....

 

I do not know if this helps, I am sure others will be able to improve and expand on this description!

 

Jim

[pritch]

Cheers, both. I think I've got the vague idea now.

 

Looking back, I'm glad I went through checking everything, rather than just taking our stage manager's suggestion of "Well, if it works without the terminator, just leave it out!"

[Tomo]

The standard is called "RS485" or "EIA/TIA 485" depending on how old your document is.

 

There are quite a few useful documents about that standard - Maxim have some good app notes with pretty 'scope pictures, but it's all way beyond what you really need to know.

 

However, the 'one legged' fault is very common - if removing the terminator makes it work, it's basically proof of a cable fault on a data line.

Removing the terminator will often get you through that performance, but it's definitely worth finding the duff cable and fixing it - if one core has gone, the other is usually on its way out, not to mention you're very likely to get noise causing flicker or moving light jitters as it's can't do the normal noise-rejection stuff.

 

One book that's highly recommended is Adam Benette's "Recommended Practice for DMX512 (2nd edition)".