Help! My Chauvet COLORStrip Fixtures Won’t Hold Still!
Somewhat recently, lighting company Chauvet introduced a product call the ‘COLORStrip Mini.’ It’s a neat little LED lighting fixture, and very reasonably priced. A local staging company buys them by the pallet, because they work so well in their particular application.
Unfortunately, DJ and lighting forums are lately filled with posts discussing ‘flicker’ when these fixtures are used in RGB mode, when driven by some lighting consoles and in some situations. The flicker is not the regular stepping seen when fading LED fixtures up and down over a 256-step scale. Rather, the fixtures will randomly flash different colors, and sometimes turn off entirely when DMX is applied. The fault occurs both when the incoming data is stable and when it changes.
The staging company was irked that they’d invested so much money into a product that would randomly fail. The lights worked OK when connected to a Hog. But they failed miserably with a GrandMA. Etc.
So they packed up a couple of the fixtures, a mini lighting console, and came to visit our shop.
An hour later, we came up with a rock-solid solution. Here’s how it works:
The Numbers Behind DMX-512
At its core, DMX data is a serial bitstream running at 250 KBaud (250,000 bits per second). Yes, per the USITT standard there are some special sections, such as the ‘break’ and the ‘mark after break,’ but at the most basic level, it’s just serial data. As with all such protocols, there are very specific timing requirements. Here’s a diagram, courtesy Microchip’s AN1076 application note:
Thus, if you want to generate DMX data as quickly as possible and still adhere to the official spec, you’re looking at (for the data only) a start bit (4 uS), 8 data bits (4 uS * 8 = 32 uS) and two stop bits (4 uS * 2 = 8 uS) = 44 uS per byte of data. Of course, a full packet contains 512 data bytes plus the start byte. So 44 uS + (44 uS * 512) = 22,572 microseconds total.
Adding in the time for the ‘break’ section of the signal, and a DMX datastream can refresh 512 channels of data at almost exactly 44 Hz.
To test (and ultimately solve) the problem with the COLORStrip fixtures, we used an ENTTEC Pro DMX-USB interface, a DecaBox Protocol Converter, and some clever bits of assembly language.
Some people on the forums reported that they could ‘fix’ the COLORStrips by dropping their console’s refresh rate from 44 Hz to 30 Hz. This seems reasonable – some small microprocessors may be so busy doing their other chores that they just can’t keep up with full-frame data.
We wrote a firmware profile for the DecaBox which accepts, buffers and re-transmits DMX data. The input routine has been field- and lab-tested to easily accept 44 Hz packets. The output routine generates USITT standard DMX, but with a twist: since it’s written in assembly language, we have extremely granular control over the timing of the output waveform. With a few lines of code, we can surgically adjust the signal in 12.5 nanosecond timeslices.
That’s really, really precise.
So for the first batch of testing, we set the DecaBox to re-transmit at 30 frames per second. Essentially we were stretching the ‘mark before break’ section of the waveform, which makes the resulting datastream less dense. Fewer packets per second are being dumped down the wire.
This helped reduce the flickering visible in the fixtures, but not completely. Certainly not enough change to be pleased with.
Slowing down the frame rate to 15 Hz didn’t result in any noticeable difference either.
Then, we supposed that the problem related to the inter-byte time of the DMX waveform. Using an oscilloscope, we verified that the DMX output bit width was precisely 4.00 microseconds wide. The DecaBox’s output stream, along with that of most lighting consoles, doesn’t insert any delay between bytes. It’s two stop bits followed immediately by the ‘break’ signifying the beginning of the next byte. Perhaps, we wondered, the Chauvet brain doesn’t run so quickly?
A few moments later, we’d added a short pause between bytes in the output stream. About 40 uS worth, per byte.
40 millionths of a second. But in embedded hardware designs like this, it’s an eternity.
And now, the Chauvet COLORStrips run perfectly. Sadly, there’s no possible way that a regular lighting console will give enough control over the output waveform to fix this. Unless they release a firmware update that also changes the time between bytes. At the very most, there might be an adjustment for ‘fast’ (44 Hz) and slow (~20 Hz) output data. As we’ve proven, for the Chauvet controller, the time between bytes is much more crucial than the time between packets.
Take a look at this demo video. You’ll see the fixture receiving regular DMX data, both static and changing. The flashing and glitches are awful – certainly it’s not something you could show a paying client as part of a light show.
And then, you’ll see the same system running, but with the DecaBox added in-line.
Video clip has no sound.
Need one? They’re in stock and ready to ship. Grab yours from the online store today.
NOTE: The flickering exhibited by these misbehaving fixtures is different than what you’d see if an unterminated DMX line or flaky cable is ghosting or causing errors. Before you buy one of these boxes, double-check your system’s signal integrity. Naturally, we also offer a 30-day no-questions-asked return policy.
NOTE #2: We’ve recently received the following email…
We have several Pro Color Strips from Mega Lite that flicker/jump whenever connected to anything but a standard DMX fader board or when not in stand-alone mode. Once hooked up to any sort of console or PC based rig, they are basically unusable. Hopefully your product will do the trick.
Thanks for your time today, John. I just wanted to make sure you knew we are satisfied customers, but here’s what happened if you’re interested:
Dave and I remembered that during our initial troubleshooting session a couple of weeks ago we had switched one of the old color strips out with
one of our newer ones that wasn’t giving us problems, and all of our glitchy color strips were slaved to it. When we were talking to you we had
the decabox ahead of that new master in the signal path. We placed the box after it when we realized this, slaved old with old, and it’s all working
-Aaron S, Crosspoint Church, Texas
This box saved our stage design. Our recent stage design has 8 chauvet colorstrips used to light up coroplast panels. It looked great except the led bars would flicker sporadically.
I ordered the box plugged it in and bam it fixed it. It was as easy as plugging it at the beginning of the dmx line. Super easy to use, very pleased.
-Brandon Reed, Media Director, Grace Church