The 5 pin DMX Cable Religion

[andy_s]

Some of the Strand and ETC dimmers are running on Ethernet straight into the processor,

Liam

 

 

this is what our ETC dimmers do - although interstingly they show a permanent error message, as they go looking for DMX even though they get the data elsewhere. this doesn't affect the functioning of the system.

[Liam FitzG]

I am afraid that I can't offer any info, I'm fairly much a Strand man. I haven't a lot of faith in ETC dimmers, but that is just my opinion biased on experience. I don't want to start an argument on how good ETC dimmers are. But it might be worth starting a new thread to see if anyone else running on Ethernet has the same issue. You may even catch the attention of an ETC rep.

Of course if they are running, solidly, then the "let sleeping dogs..." rule comes in.

 

Liam

[Dmills]

Sigh!

Ok,here is the situation as I understand it....

 

Electrically DMX is RS485 physical layer (5V differential signalling, 120ohm transmission line impedance), this impedance value is what separates mic and DMX cable electrically and it is determined by the ratio of capacitance to inductance of the

cable. This is purely a dynamic effect and comes into play because the cables are electrically long compared to the wavelengths being transmitted. It is NOT a cable resistance effect.

 

It is worth noting that AES-EBU digital audio cable is 110 Ohm impedance which is sufficiently close to work in almost any DMX application. If you make mic cables out of this stuff they can double as 3 pin "DMX" lines which is useful.

 

The original standard called for two data links with a single overall screen, one carrying the DMX protocol and one for a poorly defined "Second data link", this second link was never well defined and even the fact that it was required to be RS485 was not exactly obvious from the original standard.

Due to this lack of detail, the second data link pair was used for all sorts of random things, Pulsar for example put the power supply to the desk down it!

The practical effect of this is that it is almost impossible to actually use the second data link in a practical installation.

 

Now the standard requires 5 pin connectors (even if only 3 are used) and I view this as a good thing in that it forces separation of the DMX (RS485 cables) and mic cables (Back in the day AES digital mic cable was unheard of). This would certainly be a better situation then the mess we live with now, with a well defined 5 pin standard and TWO 3 pin standards!

 

In my view having an extra two pins on the connector is less of a pain then having two different 3 pin standards!

At least when presented with a pile of 5 pin kit you can be reasonably sure that plugging it all together, adding a terminator and setting the addresses will give a system that at least should work! Try saying that with a random pile of 3 pin gear!

 

Now DMX is arguably not a very good protocol for controlling moving lights (remember it was designed for dimmers) and something better would be nice, but it has the virtue of being simple to implement and easy to fault find. It also seems to interoperate fairly well most of the time (Connector silliness apart), my one beef is that a simple checksum would have been nice.

 

Just my thoughts.

 

Regards, Dan.

[thinkoutside]

Thanks everyone, especially Dan and Andy. All your inputs were really helpful.

 

Dan would you say it is 100% safe then to use 110 Ohm cable or should I keep looking for 120 Ohm cables.

 

Here are some 110 Ohm cables I found that comes out to be about $1 a foot, what do you all think?

 

http://eventhorizon-srv.com/ecommerce/os/c...roducts_id=1577

 

BTW: I like the Ethernet idea. The cable is very cheap and you could have switches on each of the light bars feeding each of the lights. Then you would only need one cable going from the controllers to each switch. WiFi could then also be used for each of the lights and we could all just forget about cables all together. I can't see any draw backs. I wonder what the engineers are thinking about this one?

[Madmac]

It may be worth explaining in detail what goes on in hardware terms when DMX information is transferred between devices over metallic cables using the RS-485 specification (it should be known now as the TIA/EIA-485-A, the EIA has replaced the RS which stood for recommended standard as it is now an ANSI standard and is referenced there by other standards organisations).The following is a simplified view of impedance mismatches and termination..

 

RS485 involves the use of two wires - the first one is a data + wire, and the second a data - wire, to protect these two lines from electrical interference they are usually enclosed in a conductive sheath connected to the third pin on the DMX connector. The voltage used on the line is nominally +5 Volts. To transmit a ‘1’, the + wire is taken to +5 volts, and the - wire to 0 volts, to transmit a ‘0’, the + wire goes to 0 volts, and the - wire to +5 volts. The two wires over which RS-485 is transmitted should always be twisted. This means, that if there are disturbances on the line, this will affect both lines simultaneously, and more or less by the same amount - the result is that the voltage on both lines will fluctuate together so that the difference in voltage between the lines remains the same - hence rejecting the noise on the line.

 

This is quite different to the more common RS-232 interface used on personal computers, where one wire is always kept at 0 volts. A ‘1’ is transmitted by putting between +6 and +12 volts on the line, and a ‘0’ is transmitted by putting a voltage between +3 and -12 volts onto the line. If interference appears on the line all it has to do is add or subtract to the current signal until it crosses the +3 to +6V threshold and the data will be corrupt.

 

This results in a hardware standard which can connect devices over distances hundreds of times further than would be possible when using RS-232. It also increases the maximum data rate between devices, for example using RS-232 it is normally about 19200 baud (faster communication up to 115K is possible, but the distances over which data can be transmitted is reduced significantly. At 19,200 baud you should not attempt to send data further than about 30 metres and even that, although it works, is more than specified by the standard) while DMX is able to send data at 250K baud over distances of hundreds of meters without problems, the maximum data rate recommended for RS485 is 10M baud but only over 12 metres of cable.

 

The way the data is encoded onto the cable is the same in both RS232 and DMX. Every byte transmitted has one start bit (data low) added, which is used to warn the receiver that the next character is starting, eight data bits (this conveys up to 256 different levels) and two stop bits (data high) added - used to tell the receiver that this is the end of the character. Each bit is 4 microseconds long, this time and hence the time for the complete byte is fixed at 44uSecs. All other timings, gap between bytes, length of break, end of break to start byte can vary as well as the number of channels in the DMX packet.

 

When the DMX transmitter outputs on to the cable three things happen to the signal as it propagates along taking about 5nSec per metre (5, 1000 millionth of a second). Firstly as the levels change from low to high or high to low the transmitter has to charge up or discharge the capacitance of the cable taking time which will vary depending on the length. Secondly the nice clean rectangular pulses start to spread out, this is due to the different frequencies making up the pulse travelling at different speeds. Thirdly every time the signal passes thru an impedance mismatch a reflection is generated. An impedance mismatch will occur for any joints in the cable, connectors, wrong type of cable or the end of the cable without a termination resistor.

 

The signal as received by a fixture is decoded by a UART (Universal Asynchronous Receiver Transmitter). This needs to synchronise with the incoming data and work out what bits are high or low. The usual technique is to look for the start bit going low and trigger a timer, the circuitry then looks 2uSecs later and checks the signal is still low, this confirms the start of the byte plus its stop bits. To work out the following bits it looks every 4uSecs to see if the signal is high or low and stores the result. On the last 2 bits it checks these are high if not generates what is called a framing error. (Some more expensive UARTs sample twice or three times during the 4uSecs bit time to confirm the data and reject noise. Some low cost UARTs do not have framing error detection to allow the corrupt byte to be ignored.)

 

From the above it can be seen that data can be corrupted by the signal at the instant the UART is checking it being the wrong value. As noted above each connection in the signal cable will generate a reflection, for it to corrupt it must occur at a fixture during the UART sampling window. In a simple example of a fixture connected to an unterminated length of cable this will need to be, at a propagation speed of 5metre/nSec, around 400 metres long. Remember that in a real system fixtures are placed at random points along the cable length, reflections will also reflect again making it virtually impossible to predict problems. This is further complicated by all controllers generating the rest of the DMX timings with slight differences. So one desk can work on a system and another will have problems.

 

From the above you can see that on a small system the chances of the delay being enough to cause problems is small. But you can be certain that the more connectors and longer the cable the closer you get to having a problem. If you want to take the risk of a fixture receiving a corrupt signal at the point you are doing a reveal etc and blowing a show…. that is the difference between a professional and a cowboy.

 

Using Microphone cable adds another detail to the story.

 

 

Madmac

[Oovis]

ESTA has carried out research and have found CAT5e cable perfectly capable of carrying DMX data.

 

We've had a CAT5e infrastucture installed which coped admirably with DMX and ethernet data.

[Ekij]

Dan would you say it is 100% safe then to use 110 Ohm cable or should I keep looking for 120 Ohm cables.

It's not 100% safe. It's perhaps 99.9% safe, it's very close to the required impedance so the length at which you can expect to experience problems is going to be 'long'.

There is no magic point where things change from "working mostly" to "working guaranteed" it is a sliding scale, the more things you do right the better the connection will be and the less frequent the problems will be. 110 Ohm is much better than random Mic cable, but not quite as good as 120 Ohm cable. You have to make the choice as to whether that is good enough for you. In a fixed installation you can afford to skimp a little more than on a mobile one as the enviroment will be mostly unchanged. In a mobile scenario you could end up one day not too far from a radio station where one of your cables happens to be a resonant frequency of the radio transmission and the whole just thing stops working. In this scenario as the punters come through the doors and you're still no closer to getting a single lamp on, you really start to think that that extra few pounds for proper screened cable would have been worth it. :blink:

 

BTW: I like the Ethernet idea. The cable is very cheap and you could have switches on each of the light bars feeding each of the lights. Then you would only need one cable going from the controllers to each switch. WiFi could then also be used for each of the lights and we could all just forget about cables all together. I can't see any draw backs. I wonder what the engineers are thinking about this one?

 

Wireless is always a bit flaky, it's very popular these days and works well for many people but it's not as reliable or tolerant to interference as a good old fashioned hardwired input (when was the last time your land-line phone didn't work compared to your mobile)? It's simple and easy to use, I would consider it for an amateur show but I wouldn't use it for a large professional installation.

Edit: WiFi that is, I would consider DMX over hardwired Ethernet

[maeterlinck]

Surely, as with mostly all things, this depends on it's usage?!? How is it used, what is it used for, who uses it, who sees the output?? If the target audience is (as someone suggested) a club then a blimp, or even a full freakout may not be a problem, however, if it's at the National Theatre then it could be deamed slightly more of a problem.

 

Correct me if I'm wrong, but so far no one has asked Adam (thinkoutside) where this cable will be used and how! I think we all know that using 120ohm cable with a low resistance is best. However, one needs to weight this against the usage. I've done many gigs where running the DMX to the stage through the audio multi saved time and money, of course having proper DMX cable would have been better but that was a better solution is that context. I'm currently running in a DMX run at the uni I work at. We could buy DMX cable at £3/m or we could try the FST cable we have. If it works then thats fine and the student users won't notice any difference.

 

To draw an audio comparison the bell wire that so many places sell as speaker cable will work, however, you'd all prefer the real stuff.

[Dmills]

Surely, as with mostly all things, this depends on it's usage?!? How is it used, what is it used for, who uses it, who sees the output?? If the target audience is (as someone suggested) a club then a blimp, or even a full freakout may not be a problem, however, if it's at the National Theatre then it could be deamed slightly more of a problem.

 

Indeed! But I would caution that a full freakout when it involves sending the DMX controlled smoke machine to full power can be a problem even in a club (BTDT). This can also be really hard on discharge lamps!

 

Correct me if I'm wrong, but so far no one has asked Adam (thinkoutside) where this cable will be used and how! I think we all know that using 120ohm cable with a low resistance is best. However, one needs to weight this against the usage. I've done many gigs where running the DMX to the stage through the audio multi saved time and money, of course having proper DMX cable would have been better but that was a better solution is that context. I'm currently running in a DMX run at the uni I work at. We could buy DMX cable at £3/m or we could try the FST cable we have. If it works then thats fine and the student users won't notice any difference.

 

Personally I would get a reel of install grade AES cable for this, it is very nearly as cheap as FST and is almost sure to work reliably (unlike the FST). Even a bit of Cat 5 FTP would be better and that is dirt cheap.

I am buying a whole pile of multicore for a theatre install within the next few months (with lots of other clobber if the funding comes off) and we are standardising on a mixture of screened cat5E and AES cables for basically everything that is not power or speaker.

We are doing this as it means that we can basically stuff anything down anything and it will work, DMX, audio, AES, Phones, intercom.... stuff it all down the snake (or the cat 5)!

 

To draw an audio comparison the bell wire that so many places sell as speaker cable will work, however, you'd all prefer the real stuff.

 

An awful lot of installed speaker cable is massively OVER specified, the real way to do this (at least for an install) is to do the sums!

 

Prepare a link budget for your data links and check that it leaves sufficient link margin, that way you KNOW it will work and will probably save on overspeced wire. The same for speaker feeds, prepare a power budget and decide what effective source impedance you will accept at the speaker terminals (Hint in an active system this can be quite high as the voice coil effective series resistance is in series with the cable, systems with passive crossovers get a little more tricky), then work back to what cable size you need.

 

Sure it is a little work with a spreadsheet but if it saves a pile of copper (and saves my back from pulling it in) then in my opinion it is worth it.

 

Obviously for temporary cables issues of robustness tend to trump everything and 2.5mm 4 core H07 is quite reasonable for a 10M cable feeding a microwedge at a few hundred watts!

 

Of course all of this does depend on having someone who understands lumped constant (and some transmission line) circuit theory.

 

Regards, Dan (Seriously pissed off with some of the "Design Professionals" on this building job!).

[james jag]

if you look at the back of most avolites desk you can see that the five pin sockets are wired with 2 DMX universes they share the same ground and pin 4 and 5 are the other twisted pair, on my desk I have 4 universes, socket one is wired with universes 1 and 3, and socket two is wired with universes 2 and 4

[dbuckley]

I think now might be the time to explain some stuff about wire.

 

Most folks who do theatre are fairly comfortable with basic electrics. You assume that a cable with a 15A plug on the end will plug into a dimmer and you can put a couple of cans on the end, and all will be well. They also "know" (but perhaps don't know why) that whipping off those 15A plugs and putting a 63A C17 on each end and running 12 cans off it is likely to result in the cable catching fire. Some would also know that if that 15A cable wasnt a 50 footer but a mile long then you wouldn't put a couple of cans on the end as there wouldn't be enough voltage across the cans to make then light up like their normal brilliant selves.

 

Some folks even remember their schoolboy physics, and know that this all relates to the resistance of the conductors in the wire, and that W=IR2 ("eye are squared") and Ohm's Law explains away the cable catching fire and the dim parcans. And for allmost cases you will commonly come across, this simple stuff explains all you need to know. This is true for DC and most low frequency AC systems.

 

As you get further away from DC and low frequency AC, then other properties that a cable posesses become important. For the stuff we come across, the most important is characteristic impedence measured in ohms. Note that old fashioned resistance, which varies by the length of the cable remains important.

 

Lets start with some black magic. Take a random piece of cable, plug it into a test instrument called a TDR, and it tells you how long the cable is. How does that work?????

 

This isn't black magic at all really, it's how wires work in the AC domain, and this trick is called Time Domain Reflectometry. Basically a short pulse is sent down the cable, and (and heres the non-intuitive bit) the open far end of the cable acts like a mirror and bounces that pulse right back. The TDR just measures the time from the pulse it sent out to the one it got back, divides it by the velocity characteristic of the cable (thats another cable property that is unimportant for that 15A dimmer cable) and thats how long the wire is. Simple, yes?

 

So why on earth would the open end of a wire act as a mirror.....?

 

The reason is that the open end of a wire is an impedence discontinuity, and every time you get a change in impedence in a wire, you get some energy reflected back to the source. So going from "wire" to "not wire" is obviously a fairly radical impedence discontinutiy, but every change generates some energy kickback. Ideally the wire needs to be "terminated" at each end with a resistor (or some posher circuit element) whose impedence matches the characteristic impedence of the wire between the two endpoints. It is these terminators that prevent reflections from the wire ends.

 

Most systems have lots of these discontinuties, and they survive. For example, most folks run their DMX512 lines unterminated, and the system works, even though now you know why the terminators are a good idea. But every different cable, and every connector in that cable(!) and every tight bend in that cable(!!) causes an impedence discontinutiy causing waveform distortion through reflections. If the waveform becomes distorted enough then the UART chips in the fixtures that receive the bit pattern get confused.

 

Now DMX512 only runs at 250kbit/sec which isnt a very high frequency, and the cable lengths are not too long, so most of the time it all works OK.

 

So, thats the end of a short, abbreviated introcution to cable properties in the AC domain, and I leave you with just these words of advice: Use cable of the right characteristic impedence throught your network, and terminate correctly!

 

Oh yeah, almost forgot - cable resistance, and why DMX512 is limited to 31 devices: because there is real resistance at the end of the cable, and the driving device has to be able to source enough juice and the cable resistence has to be low enough that at all points along the cable the differential voltage is high enough to be sensed by our old friend the UART. Each RS485 receiver chip ,on the bus both add capacitance and DC resistance across the differential pair, and if you add too many devices then eventually the added capacitance will blunt the waveform edges enough and the resistance load the line enough that the receiver chips dont see anything.

[bruce]

Cracking article, David...

 

 

 

Having been away for a few days, and then read this whole thread, it seems that some people are missing one very important fact. DMX is a standard. It's the way in which things are done. If we can't all agree to adopt that standard, then there is no guarantee that things will continue to work as intended.

 

So if you want to use 3pin XLR connectors, or a different pin-out from everyone else, or some old cable that happened to be lying in your workshop, then go ahead. I don't care. Do what you want. but don't call it DMX, cos it isn't. It doesn't meet the standard. It might be close to the standard. It might vaguely resemble the standard. But since it doesn't meet the standard, it isn't DMX. It is "something that is not DMX".

 

Is that clear?

 

I don't really care whether it "usually works", or whether it "works for you". That's not what standards are about. Standards are sets of specifications that the industry has agreed to abide by. They are the only way we can have any hope of equipment form different vendors interworking.

 

 

[/rant]

 

ps - another relevant definiton of "standards" is "flags that we wave before going into battle.." :blink:

[pritch]

Lets start with some black magic. Take a random piece of cable, plug it into a test instrument called a TDR, and it tells you how long the cable is.

 

Every time I've had dealings with them in the day job (telecoms), the figure has been different to what it should have been, although they do tend to give you the distance to the nearest JCB! :blink:

 

In cable TV systems the range of frequencies travelling over fairly poor (ie cheap) cable is such that the upper frequences are more attenuated than the lower, so they need "tilt equalisers" at intervals along the distribution network to reduce the low frequencies amplitude to match amplitude of the upper frequencies.

 

Indeed. In fact, on most CATV networks, the tilt is set at a given set of values at the distribution tap plate so that by the time it's reached the end of an average length drop cable into a subscriber's home, it's nice and flat. I was frequently amazed at just how well that system worked!

 

Again, great stuff David and Madmac. Really interesting read.

[Dmills]

So if you want to use 3pin XLR connectors, or a different pin-out from everyone else, or some old cable that happened to be lying in your workshop, then go ahead. I don't care. Do what you want. but don't call it DMX, cos it isn't. It doesn't meet the standard. It might be close to the standard. It might vaguely resemble the standard. But since it doesn't meet the standard, it isn't DMX. It is "something that is not DMX".

 

Is that clear?

 

I don't really care whether it "usually works", or whether it "works for you". That's not what standards are about. Standards are sets of specifications that the industry has agreed to abide by. They are the only way we can have any hope of equipment form different vendors interworking.

[/rant]

 

Amen to that! This is my major beef with the 3 pin connectors, it wouldn't even be so bad if they had picked one polarity and stuck to it, but there are TWO "standards" for 3 pin "DMX"!

 

Of course there are some proposed standards that I have severe doubts about from the point of view of interoperability between different manufacturers, ACM springs to mind.

 

There is some leeway in how an individual USER chooses to use gear that complies with the standard, but there should be no doubt from the manufacturers perspective about what is required to COMPLY with the standard.

For example, if I as an end user chose to use mic cable or AES cable or cat 5 or wet string that is my choice but if the gear claims to have a DMX-512 (1990) interface that says that it complies with the standard, not that it sort of works most of the time with some DMX sources (Grumble)....

 

In fact, even if I (as an end user) chose to ignore recommended practises, the fact that the kit is known to play by the rules specified in the standard can allow me to make an informed assessment of what I can get away with.

For example, given I know that all the gear on the line has correctly specified '485 receivers, I may decide that splitting a line with a Y with 56 ohm resistors in the legs will almost certainly work, this would fail if some cheapwad had used an opto isolator directly across the line instead of doing it right.

Thus even if the end user does things which are incorrect, knowing the gear complies with the standard allows for a certain confidence in what will happen.

 

dbuckley: Note that the capacitance of the cable is does not cause HF attenuation in a transmission line system (like DMX) as it just forms part of the line impedance. This is strictly an issue for voltage transfer systems (like audio), not for power transfer systems like any transmission line situation. The transmission line losses are a combination of resistive, raditation and dielectric losses, which makes sense as a capacitor dissipates no power (in the ideal case).

 

Regards, Dan.

 

Note a transmission line is (for these purposes) any cable which is long compared to the wavelength of the highest frequency carried, for DMX which (sort of) looks like a squarewave at 250Khz this is about 2.5Mhz (assuming 10th harmonic is sufficient), say a bit under 100M allowing for the cable velocity factor or around 0.6.

 

Edit: Fix a formatting typo.

[dbuckley]

dbuckley: Note that the capacitance of the cable is does not cause HF attenuation in a transmission line system (like DMX) as it just forms part of the line impedance. This is strictly an issue for voltage transfer systems (like audio), not for power transfer systems like any transmission line situation.

 

Does it not? Well, you learn something every day. Better go back and delete that section then. I have no formal education in these matters (just a thick kid with a few O levels to his name), I make it up as I go along :blink: I was desperately and deliberately trying to avoid terms like "transmission line" to not scare people away...