Smoothness of LED PARs (fades, not beam quality)

[delicolor]

Now I know that the actual beam of these LED PARs can be rather splodgy if they shine on something close like legs (drape or human), but I've noticed in shows by my local am-dram that they don't seem to fade in and out overly smoothly either. The last time I saw fades that lumpy was on a cyc (story below). It stands out compared to the regular FOH wash but I've no idea whether they use one desk or three as I've never been interested enough to go and have a look. (I say three because they have a pair of wigglies in their standard Panto rig and these will often demonstrate St. Vitus tendencies but I don't know if it is down to sloppy programming, sloppy operation or the poor guy fighting to get the best out of unsuitable equipment. I can commiserate with the latter as I used to enjoy instant lighting design on sometimes unfamiliar rigs for music and that can be seat of the pants stuff).

 

Back to the lumpy LEDs, is this likely to be a function of DMX, the head or both? I know that DMX supposedly has enough granularity at 8 bits for 256 levels and refresh up to 44Hz but I imagine that this is down to the processing welly of the controller, the smoothness of the head electronics and the method of control.

 

I don't know what make I am looking at, they look like regular Thomas stubbies with a LED lamp insert. I've never played with LED stuff outside of a decorative lighting installation but I am curious if they are up to the job of long, smooth, discrete almost inperceptable cross-fades or if they draw attention to themselves like many movers do with their much cruder dimming that changes the beam texture as well as the intensity.

 

(As I tend to ramble on about the bad old days when I post, I once went to Sheffield Library theatre in the early 90s when they had a large but low tech furse desk. (A 48 way Regent, or Delta as they renamed it). I was rather shocked to see a fade-up of the cyc was extremely lumpy and my first assumption was grit in the fader or a faulty opto-coupler on the channel. I was then astounded to be told that it wasn't a fault- it was a limitation of the desk multiplex, which CCT called FMX (to sound snazzy in the world of AMX & DMX, or more likely, D54). Unfortunately the refresh rate was rather low (although it was synchronous as the electronics was all discrete unless you bought the BBC Model B adapter which did exhibit variable response) so when you did a fade in the 3s -10s range you only saw a small number of actual levels on the output as a staircase. Your average T26 had enough thermal lag to smooth this out, but a batten with 100W GES lamps was like a sunset dimmer with half the studs shorted out. ;)

[timmeh2]

Hi

 

It all depends on the resolution of the PWM driver that's running the chips. Cheapy LED cans tend to be quite low, which won't give you very many steps for your 255 values of DMX. The same is also true for dimmers.

 

This is opposed to high-res drivers like (for example Martin Stagebar) which cost 25 times more!

 

All the best

Timmeh

[niclights]

The cheap, basic LED Pars are particularly bad over their entire dimming range in my experience. So in this case I would say it is the fault of the fixture.

 

More expensive units make use of two channels for 16-bit resolution (with the necessary electronics to make use of it of course) and this certainly helps. Consoles will still be limited by fader resolution in manual fades but timed fades do benefit. In this case you will only notice steppy artifacts at initial turn-on and the very low levels which are both still issues in the LED world.

[Barney]

I've often found that the cheaper LED units tend to have a poor dimmer curve. This means that, at low intensities, the 'step change' in output is very noticeable when going from say 1% to 2% or 2% to 3%.

 

Conversely, at high intensities, the step change between 80% and 90% is not noticeable!

 

For fixtures that I use often, I have created a specific dimmer curve on the desk, which counteracts these idiosyncrasies!

 

Barney

[delicolor]

Very informative answers, thanks. I hadn't appreciated that 16 bit was available on some LED-PARs, indeed I had forgotten that some movers were like that.

 

I can remember a Plasa show (at Olympia) where we had our Swiss distributor on the stand demonstrating his whizz-bang florry light dimmer which used conventional dimmable 4' tubes (the permanent heater type with earthing strip between the caps) but variable frequency, waveform shape and amplitude for the arc current. (I can't remember the details but it struck me as very clever that he had worked out how to do this)

 

The performance of the unit was absolutely brilliant- you could go from the tube glowing in view but making no impact at all on the cyc to full smoothly at any speed you wanted without any strobing or rolling bands. (I don't know how if would perform on a desk to a rock beat though, as it only had a rotary pot on the box).

 

Where it was tricky though was the native dimmer curve- it stampeded in to about 20% brightness within about 10 degrees rotation before becoming much flatter with practically no visible impact from 60%. It wouldn't need an S curve, more of a half bathtub.