Level 1% is very bright on Chauvet COLORdash Par Hex 7

[Stevens R. Miller]

I wrote about a similar problem in another thread, but that focused on possible issues with the control console. This time, I have eliminated the console from the equation.

I'm working with a number (19, actually) of Chauvet COLORdash Par-Hex 7 instruments. Very nice lights with a wide range of colors. The director wants a slow fade to black in some scenes, and an immediate black-out in others. The immediate black-out works great. But, when I fade slowly, the lights gradually fade down (which is what I want), but suddenly drop from a farily bright level to black at the low end (which I do not want).

I swapped the ColorSource 40 console for my computer and an OpenDMX. This way, I could be sure that I was sending an exact value of 1 to an instrument. At that level, it was surprisingly bright. going to 0, of course, turned it off. At the backpanel, one can select any of three "dimming curves," but none of these seems to change the amount of light emitted at level 1 (it appears they don't change the level-to-intensity function but, instead, introduce a lag when the level changes, to simulate the persistence of incandescent bulbs).

The instruction manual describes a 16-bit mode, which I have never used. Is that likely to give me access to lower levels of light? At it is, when doing a slow fade, it seems like the lights fade to a low level, then it's like someone just flips a switch and off they go.

If the 16-bit mode would help, I'm willing to pursue it, but I'm not sure the ColorSource 40 can handle it. Also, each instrument would use up 13 addresses to get at that mode. That would consume 247 in total, and those 19 pars are not my whole universe.

As often happens, I'm stuck at a point where I'm accessing a lot more theory than I have practice.

Can anyone advise me?

 

[Michael K]

Have you tried the dimming modes in the lights menu? (Detailed on page 15 of the manual) These will adjust the how the brightness changes from linear (probably default, and how computers like to think) to an exponential curve, which is how we perceive brightness.

 

[Stevens R. Miller]

Have you tried the dimming modes in the lights menu? (Detailed on page 15 of the manual) These will adjust the how the brightness changes from linear (probably default, and how computers like to think) to an exponential curve, which is how we perceive brightness.

Yes, I have. Here's what I said about that, above:

At the backpanel, one can select any of three "dimming curves," but none of these seems to change the amount of light emitted at level 1 (it appears they don't change the level-to-intensity function but, instead, introduce a lag when the level changes, to simulate the persistence of incandescent bulbs).

We tried all four modes. The amount of light is the same at level 1 in each mode. What does change is how rapidly the light will change intensity. In the "slowest" mode, " it fades out fairly smoothly, but this precludes a sudden blackout, and also still drops out unexpectedly early if one is doing a really l-o-n-g fade to black.

It would be nice if there truly were a dimming curve, rather than a response rate, setting. That would probably do what I want, and I was certainly optimistic when I saw "dimming curve" in the manual. But it doesn't really change the level-to-intensity curve of the instrument, only the rate at which it slews from one intensity to another.

 

[notoriousRBG]

As you are discovering, low-end dimming smoothness is one of the biggest differences between cheap and expensive LED equipment.

One of the reasons things like ETC's D40s are so expensive is how well they can dim down to zero.

16 bit mode may well help. Otherwise, better equipment is the solution.

 

[robmerow]

If you can get away with less light output, you can always gaff tape some neutral density gel on the front. Your brights will be dimmer, but so too will your low end before it drops to zero.

 

[MNicolai]

You need to flip your profiles over to 16-bit.

# of addresses is only a problem if it would max out your universe. The limitation on the Colorsource 80 control channels. Doesn't matter if your channels are 3/ea or 25/ea until you reach that limit of 512 addresses on your universe.

 

[Stevens R. Miller]

As you are discovering, low-end dimming smoothness is one of the biggest differences between cheap and expensive LED equipment.

Gracious, are you saying the COLORdash Par Hex 7 at $400 each is cheap? My crummy Coidak lights have better low-end dimming at $50 per unit.

 

[Stevens R. Miller]

You need to flip your profiles over to 16-bit.

# of addresses is only a problem if it would max out your universe. The limitation on the Colorsource 80 control channels. Doesn't matter if your channels are 3/ea or 25/ea until you reach that limit of 512 addresses on your universe.

Yes, but we have a lot of other instruments, so we are already running out of addresses in our single universe. If 16-bit mode solves the problem, we may use it on some, but not all lights. That will probably keep us below a total of 512 addresses.

Might get to experiment with it a bit this weekend. If so, will post back here.

 

[Michael K]

At the backpanel, one can select any of three "dimming curves," but none of these seems to change the amount of light emitted at level 1 (it appears they don't change the level-to-intensity function but, instead, introduce a lag when the level changes, to simulate the persistence of incandescent bulbs).

OK, I missed this in my first read through.

Just to confirm, you're talking about DMX value 1, not console 1% right? If so, then yeah, 16 bit is the only thing that will help that and keep full intensity. If a bit dimmer overall is fine, then an ND filter gel will get around that final step. If it's still too sharp a drop on a fade in 16 bit mode, you should be able to set an actual curve on the console.

Gracious, are you saying the COLORdash Par Hex 7 at $400 each is cheap? My crummy Coidak lights have better low-end dimming at $50 per unit.

Yup, the really nice ones are 4 to 5 times as much (Chauvet Vari-white, ETC X7/Lustr)

 

[JD]

Maybe 16 bit is the answer, but don't count on it. Getting technical, here is the problem:
LED's use pulse-width modulation to vary brightness, not voltage level. So, as you bring the dimmer down, the pulse width (which happens at a very high frequency) gets smaller and smaller. At some point, it has to go from some type of waveform to no waveform. That is the jump you are seeing. You are actually at the mercy of whatever designer programed the microprocessor in the fixture and how accurately pulse width information is fed back to the microprocessor. Generally, an expensive fixture had a bigger budget, and frankly, cared more about that final transition. The thinner those final pulses can be at DMX 1, the closer to black the fixture can be run.
So, try one on 16 bit and see what happens. The results may be good, but I wouldn't hold my breath.

 

[danTt]

Yup, the really nice ones are 4 to 5 times as much (Chauvet Vari-white, ETC X7/Lustr)

If you're paying $1600-2000 for a D40 you're making some dealer very happy....
And if you're paying $1600-2000 for anything Chauvet you're making future you very unhappy...

 

[Chris Pflieger]

If you're paying $1600-2000 for a D40 you're making some dealer very happy....
And if you're paying $1600-2000 for anything Chauvet you're making future you very unhappy...

Not sure you meant it that way, but Chauvet Rogue washes are around $1600 and they make a lot of people very happy...


And speaking of... The difference between 0% and 1% is dramatic on those fixtures. They do fade smooth since it's in 16 bit mode, but if I want sustained low light, I have to use the DMX direct setting in my software to get sub 1% control.

 

[Stevens R. Miller]

The difference between 0% and 1% is dramatic on those fixtures. They do fade smooth since it's in 16 bit mode, but if I want sustained low light, I have to use the DMX direct setting in my software to get sub 1% control.

As far as I can tell, the CS40 console won't let me call for an explicit DMX value in the 0-255 range. It only operates in percentages. However, when I used my computer, I was able to set an actual value of exactly one as the DMX value sent to the light. Yes, it's bright, which is what got me started on this line of inquiry.

I will try 16-bit mode and see what that does for me.

 

[JD]

% gives you 100 steps between black and white. DMX gives you 255 steps. 16 bit DMX gives you over 64,000 steps. One would think that should fix it, but it still depends on exactly how the PWM is being handled by the Micro as I said in post #10
It simply depends on the tolerance and thought that went into designing the microprocessor output. Having a waveform with a pulse width of one-sixty-four thousandth of the waveform width is pretty tricky.

 

[Stevens R. Miller]

Having a waveform with a pulse width of one-sixty-four thousandth of the waveform width is pretty tricky.

We'll find out. LEDs have switching times measured in nano-seconds. So long as the modulator sets the pulse width as a monotonic function of the input level and is reasonably linear, 16-bit mode ought to at least be able to solve my problem. Whether or not it actually does will remain to be (heh, heh) seen.

 

[JD]

We'll find out. LEDs have switching times measured in nano-seconds. So long as the modulator sets the pulse width as a monotonic function of the input level and is reasonably linear, 16-bit mode ought to at least be able to solve my problem. Whether or not it actually does will remain to be (heh, heh) seen.

Keep in mind that if the PWM is at 200Khz, and we are talking about 1/64,000th of one of the 200Khz cycles, that pulse is the equivalent freq of 12,800,000Khz or 12.8Ghz. I suspect some of the bottom end of the 16 bit DXM is left for dead, or the PWM is running at a much lower frequency

 

[Stevens R. Miller]

Keep in mind that if the PWM is at 200Khz, and we are talking about 1/64,000th of one of the 200Khz cycles, that pulse is the equivalent freq of 12,800,000Khz or 12.8Ghz. I suspect some of the bottom end of the 16 bit DXM is left for dead, or the PWM is running at a much lower frequency

Where did you get 200kHz?

 

[JD]

Where did you get 200kHz?

Switchers run at very high frequencies so that the transformers do not need much of a core. LED drivers use a "buck" type power converter on the individual channels which consists of a driver, diode, MOSFET and a choke. It is much cheaper to run them high and use a very small choke. I was scoping out a LED tape driver about a month ago and that was the frequency they were using. When cost is not an option, larger cores and lower frequencies are used, but I have never run into one that wasn't at least a few Khz. There is also the photography problem. You want to run them high or you run the chance of having interference patterns on video. Now this is all about dimmable RGB or RGBAW fixtures. The direct replacement dimmable LEDs (think home depot) all bets are off. Some of them run near line frequencies which is why some have visible flicker.

 

[Stevens R. Miller]

Switchers run at very high frequencies so that the transformers do not need much of a core.

Could there be a difference between the switching frequency of the power supply, and that of the PWM? 200kHz sounds about right for the supply, but high for an LED PWM.

 

[JD]

Could there be a difference between the switching frequency of the power supply, and that of the PWM? 200kHz sounds about right for the supply, but high for an LED PWM.

All about the cost of copper and cores. I'm sure they chose a frequency that was friendly to the microprocessor clock frequency as it is usually the micro driving the buck regulator. Only one way to find out, open it up and scope it!