DMX fading

[Steven]

The good news: I just controlled some RGB lights for the first time through a DMX controller from the sequence editor. However the fades on the DMX channel do not seem as smooth as LOR channels.

I am using a DMX RGB controller from HolidayCoro, which they say: "These controllers also feature a full 24-bit, 256 levels of fading per color - the maximum DMX supports – so that you have the best possible fading." I have one of these 3-channel controllers hooked up to a Rainbow Flood (bare board, no enclosure yet).

I first used a USB-485B as Universe 1. Then I switched to an Enttec Pro, but I couldn't notice any difference.

Next, I thought the problem was because I was using an ordinary fade from 100% to 0%, so I changed every other fade (so I could compare) to a DMX fade from 255 to 0. This made a small improvement, but it still doesn't fade smoothly.

I don't remember this problem with the CMB-16D and the Rainbow Floods, but I didn't get a chance to pull those out of the storage shed to compare. It's possible that I'm only noticing this because the lights are now indoors on my workbench. Perhaps having them outdoors in the display with other lights nearby make this problem not noticeable.

[JonB256]

I think you are seeing the non-linearity of LEDs more than a problem with DMX fading. If that is the 12vdc controller, connect a 12vdc halogen lamp to it and see how that fades.

From 255 (full on) down to 240 and sometimes down to 210, most LEDs do not seem to dim much. At the bottom end, they go nearly off around a value of 20. (that's about a 10% deadband at both top and bottom)

So, if you want LEDs to actually start to dim like incandescents, then you need some sort of dimming table to re-map the values to the 80% of the usable lighting.

[bretk]

The shimmer is definitely slower on dmx vs lor networks, but not a function of the 3 channel jobs, more a lor thing as neither enttec pro or other dmx devices make a difference.

[bretk]

I think the fading issue is similar, hopefully a correctable (software) issue. Maybe someone with additional rgb dmx experience can chime in here.

[Steven]

JonB256 wrote:

I think you are seeing the non-linearity of LEDs more than a problem with DMX fading.

From 255 (full on) down to 240 and sometimes down to 210, most LEDs do not seem to dim much.

That's not what I mean. I didn't explain what I mean by "smooth."

What I mean is that it looks like it dims in steps, as if there were only 8 levels instead of 256. It kind of "jerks."

bretk wrote:

The shimmer is definitely slower on dmx vs lor networks, but not a function of the 3 channel jobs, more a lor thing as neither enttec pro or other dmx devices make a difference.

I think the fading issue is similar, hopefully a correctable (software) issue. Maybe someone with additional rgb dmx experience can chime in here.

I think that may be it. It's probably because the sequence editor (or show player) has to send each DMX dim level separately, and it's too busy doing everything else to have time to send 40 dim commands in 1 second. On the other hand, the LOR controller handles the entire fade sequence itself, so it can make it smooth.

I just looked at the documentation for the Enttec DMX Pro, and it doesn't have the ability to generate a "DMX fade" itself, so it still has to be done by separate dim events, generated by the show player and sent through the CommListener.

I think my best bet is to leave the channels that should "pulse" to the music on LOR channels, and use DMX channels for lights that fade slowly.

[JonB256]

Steven, how fast are these fades? If a fade is less than one second long, then it WILL be flashy with most LEDs because the controller only gets a command 30 to 40 times per second. A one second fade means that, at best, the fade from 100% to Zero is in 30 steps. An LED string, depending on construction, can actually show that. An incandescent string will not because it has to cool down to dim.

Any fade longer than 2 seconds (60 to 80 steps) should appear smooth.

[WilliamS]

JonB256 wrote:

Steven, how fast are these fades? If a fade is less than one second long, then it WILL be flashy with most LEDs because the controller only gets a command 30 to 40 times per second. A one second fade means that, at best, the fade from 100% to Zero is in 30 steps. An LED string, depending on construction, can actually show that. An incandescent string will not because it has to cool down to dim.

Any fade longer than 2 seconds (60 to 80 steps) should appear smooth.

This is correct, DMX runs normally at 40 packets per second in terms or refresh depending on overall network congestion could be slower or faster but thats about the median. So if you have a fade of 100% in one second of DMX you only get 40% of the visible fade. With LED's it may look worse as the first 10-20% sometimes has little to no visual effect, and depending on the LED and power the last 10% may also be already off. Leaving you with 70% of the fade leaving you 28 fades total that you can see but it will be in the middle of that second.

If your fading from full to zero in that time frame, test at what % you can visibly see as the fade starts in terms of % with the control panel, and where the lights turn off, then adjust your intensitys for that fade to go with those numbers. This will give you the full 40 steps to fade out, will look much smoother.

[dmoore]

Steven wrote:

The good news:  I just controlled some RGB lights for the first time through a DMX controller from the sequence editor.  However the fades on the DMX channel do not seem as smooth as LOR channels.

This video maybe of help - it shows a variety of controllers (LOR, Lynx, HolidayCoro, Misc, Pixels, etc) all fading to the same signal (from an IDMX):

I think you'll find that there are a LOT of factors that affect fading - bit levels, firmware design, LED vs Incandescent, software, etc. I'd always recommend testing any combination of software/hardware/lights/controllers/etc prior to use in a production display to make sure they work as you expect they should work.

[dmoore]

Steven wrote:
I just looked at the documentation for the Enttec DMX Pro, and it doesn't have the ability to generate a "DMX fade" itself, so it still has to be done by separate dim events, generated by the show player and sent through the CommListener.

I think my best bet is to leave the channels that should "pulse" to the music on LOR channels, and use DMX channels for lights that fade slowly.


The LOR and DMX protocols (above RS-485) are completely different in the wway they operate. Either one isn't best or better than the other overall - each has pros and cons. For a better understanding of what exactly the differences are, see my presentation on the subject here:

http://www.holidaycoro.com/kb_results.asp?ID=21

The issue at hand with respect to the fades, shimmers and twinkles is mainly software. LOR devices using the LOR protocol will generate these effects at the controller, not in the software or dongle. DMX devices create the effects at the software/dongle level, so technically, you can create any type of effect, not limited to just what the controller is programmed for. The issue you are experiencing is because the software doesn't properly handle generation of an "effect" correctly - for the most part, a software issue.

[dmoore]

bretk wrote:

The shimmer is definitely slower on dmx vs lor networks, but not a function of the 3 channel jobs, more a lor thing as neither enttec pro or other dmx devices make a difference.

It's really a function of the software driving it, not of "LOR" or "DMX". All DMX devices should respond exactly the same, be that an LOR running in DMX mode, a "3 channel job" or a $1000 DMX controller. The Enttec Pro is but a method of getting that data from the sequencing software onto the RS485 "wire" - an LOR running in RAW mode would be exactly the same on output.

[bretk]

David,

I understand, it just seems that it. Is the s3 that is addressing the dmx system is the problem, it doesn't seem to shimmer at the same rate for lor controllers including the idmx that it does in outputting dmx thru a dmx dongle, be it enttec pro, or lynx or enttec open.

[dmoore]

bretk wrote:

David,

I understand, it just seems that it. Is the s3 that is addressing the dmx system is the problem, it doesn't seem to shimmer at the same rate for lor controllers including the idmx that it does in outputting dmx thru a dmx dongle, be it enttec pro, or lynx or enttec open.

This is because shimmers, fades and twinkles (or combos of those) are produced at the controller level when using the LOR protocol. When using DMX, the effect is produced at the computer/dongle (active). Again, there are pros and cons to each protocol and neither is perfect for every circumstance.

[marsh28]

Shimmer and Twinkle look fine if you have a light with built in strobe capabilities. This is one of the advantages of the Rainbow Extreme. Haven't noticed any issues with fading when testing the regular 3 channel Rainbow floods. I'm using an Enttec Open.

[Steven]

dmoore wrote:

When using DMX, the effect is produced at the computer/dongle (active).

Even when using an active dongle like the Enttec Pro, the LOR software itself has to send every "Off" and "On" command that make up a shimmer. Unless you have a fast CPU, the computer cannot do this as fast as the LOR controllers (including the iDMX).

[84berlinetta]

Steven,

I too am diving into RGB for the first time this year, and I have also witnessed the same issues with performing fades on the RGB LEDs while testing with the Sequence Editor. I have my RGB controllers configured to turn the lights OFF by default. When I run a fade sequence, it appears I am getting dropouts in the fade (the lights momentarily turning OFF). After a bit more testing, I have discovered this anomaly also occurs when I set up a loop in an animation sequence and simply turn all LEDs ON (white color) for the duration of the loop... I see occassional drop-outs of the lights. Is this what you were seeing?

I am beginning to wonder if the PC is dropping packets (due to Windows delays?). I have a feeling the DMX protocol is set up in such a fasion that the packets must have instructions for each address on every communication, and if not, the channel returns to its default state. Can anyone confirm that?

I am also wondering about my network configuration for show time operation...

In several places I have seen people talking about running the DMX network on pins 1&2 and the LOR network on pins 4&5 of what seems to be the same cable. I would like to do this so I can daisy-chain all my elements (LOR and DMX) together, if possible. Is this OK to do, or are pins 1&2 used by the LOR controllers for anything that would cause a conflict? The last thing I want to do is cause contention on the bus and damage a USB to 485 dongle...

Thanks in advance!

Mark