DMX speed

[dynodio]

Hello everybody,

I've bought a Vellman interface and I'm not happy with it. It's very slow
I think I'm gonna buy one from Enttec.
In fact, the DMX world is quiet new for me.
What is the maximum speed I can have with the DMX protocol?
Is it possible to drive spot light with a precision of 0.01 second?
Is it possible to set 1/100 of second in FreeSyler?
My project is to trigger/synchronise a camera with a fast light sequence.
I have 5 chauvet LED rain 36 to do my test.

Thanks
Dyno

[remco_k]

Hello everybody,

I've bought a Vellman interface and I'm not happy with it. It's very slow
I think I'm gonna buy one from Enttec.

Unfortunately for you, that is no news to us. Do a search on this forum for the VM116 / K8062 (they are the same) and you'll see why... (a lot of posts and investigation from me, having the same problem).
The one and only solution is to dump the Velleman and buy something else, like one of the Enttec interfaces.
If you want to be sure that it all works okay, buy the Enttec USB DMX Pro.

In fact, the DMX world is quiet new for me.
What is the maximum speed I can have with the DMX protocol?

DMX runs at about 250 kbit/s. Every 24 msec all 512 DMX values are sent out by the interface. (that is: if you don't have a Velleman interface. )
FreeStyler has an output refresh rate of 50 msec by default, that should be changed to 20 msec to get faster output.

Is it possible to drive spot light with a precision of 0.01 second?

0.01 second is 10 msec.
So, no. You cant do that. Its a limitation of the DMX protocol itself.The theoretical minimum time a light can go on or off (or something else) is 24 msec. With FreeStyler in between, with his own output refresh rate, the theoretical minimum response time is (24+20) 44 msec.

Is it possible to set 1/100 of second in FreeSyler?

For the same obvious reason, no. Would be useless.

My project is to trigger/synchronise a camera with a fast light sequence.
I have 5 chauvet LED rain 36 to do my test.

... which could also be to slow to respond fast enough.

With so few fixtures, you could try set the num output channels to 96. Then the DMX rountrip for all 96 channels would be around 4 times faster, 96 channels are sent out every 5 msec. But with the Velleman interface being so slow, I don't think I will be that fast. Also, FreeStyler does'nt sent new values that fast (minimum 20ms output refresh rate).
One tip to get the Velleman interface as fast as possible: set all unused channels to value 0. Don't waste any space between channels.
The Led rain 36 is 6 channels?
Then place them directly after eachother:
Led Rain 1: 1...6
Led Rain 2: 7...12
Led Rain 3: 13...18
... Leave no unused channels in between.
Still the Velleman would be way to slow, but maybe you are helped a bit with this.

[dynodio]

Hi,

thanks very much!
Yeah, I already read some post about the Velleman interface.
It sounds like the DMX direction is not good for my project
I'm going see if a MIDI interface could be faster.

Cheers
Dyno

[remco_k]

Hi,

thanks very much!
Yeah, I already read some post about the Velleman interface.
It sounds like the DMX direction is not good for my project
I'm going see if a MIDI interface could be faster.

Cheers
Dyno

Midi is about more than 10 times slower than DMX (31.25 Kbits/sec ), but with lesser channels (16) and lesser values (7 bits) it might respond faster.
Still I don't think its fast enough for what you want.

[roadrunner84]

Is it possible to drive spot light with a precision of 0.01 second?

Yes... it is. However, you cannot use DMX-512 for it, with a speed of 4microsecs (us) per bit, the required time for transmission is:
Break: 88us
MAB: 8us
SC: 44us
Byte: 44us
DMX-512 uses a fixed number of 512 Bytes, however, any fixtue accepts a lower number of bytes to be sent. Counting from the other side:
0.01sec = 10ms = 10.000us
10.000 - 88 - 8 - 44 = 9860us
9860 / 44us = 224 (and a tiny bit)
so whenever you limit the number of bytes in a frame to 224 or less, you can send a full frame every 0.01sec!
Let's call this DMX-224
In your setup, you use 5 x LEDrain36 (that be the plain white ones) which use just two channels (a.k.a. bytes or packets), so you'd need 10 bytes - way less than 224 - to drive each one independendly. When you set them all to channel 1 you'd need even less (2) bytes! Just be sure these fixtures don't choke up on this speed.

[remco_k]

As I noted earlier, sending lesser DMX channels makes the DMX refresh rate faster.

The formula is:
[(88)+(12)+(44)+(CHL*44)+(CHL*MTBF)+(MTBP)] = microsecs (source - also contains a good explanation of how DMX works)
MTBF and MTBP are customizable. For cheap fixtures could be a problem.
Some (stupid) fixtures don't understand a DMX protocol that is not sending 512 channels and getting confused.
I believe I've read somewhere that the minimum of DMX channels that needs to be send is 8. I can't find it now... So it might not be true.

But lets assume DMX-512 has a refresh rate of 44 Hz. (=it takes 22,7 msec to send all 512 DMX values)
Lets "forget"the MTBP, Break and MAB for easy calculation.
If we are going to send out half number of channels (DMX-256) the refresh rate doubles (88 Hz , it now takes 11,36 msec to send 256 DMX values).
DMX-224 (as stated above) takes 9,93 msec to send.
BUT that is not enough. If you want to set a value it can happen that the current DMX 'position' that you wanted to change just had passed. So it can take twice the time (2x 9,93 msec = 19,86 msec) to get the value to the fixture.
So the number of channels need to be devided by 2 to get a response time of 10 msec.

Next there is all kinds of latencies in the controlling software that needs to be added and take in consideration. FreeStyler has an built in refresh rate. Fastest is 20 msec (!). The driver and USB controllers all have latencies which I don't know. They also needs to be added to the total latency.
So at minimum the time a value gets to a fixture is 20 msec + DMX refresh time (say 10 msec when using 112 channels).

But in short: theoreticly you can get the DMX cycle so fast that you should be able to control lights with a resolution of 10 msec.
In practice you'll need a computer and software that have latencies also. Thats were it goes wrong.
But above all: as long as using the Velleman VM-116 interface, this idea will never work this fast.

[remco_k]

...I believe I've read somewhere that the minimum of DMX channels that needs to be send is 8. I can't find it now... So it might not be true.
...

Found the following on the DMX512 -Wikipedia site in the timing section.

A packet is required to have at least 24 channels.

Ah! Thank you. I kept on thinking about this 8 the whole day. Would have had a sleepness night if you did'nt post this.
So its 24 - stored in memory for the next time. Now I don't know were my 8 came from...

[remco_k]

Well the MAB break is 8 μs minimum.

Take it as being that and do not worry anymore and have a wonderful Christmas

U2.
And it seems it is going to be a white one.