LED conversions

[gyro_gearloose]

Following on from the renewed interest in my Miniscan LED conversion, Brian sent me a private message regarding converting a wash light to LED. I hope you don't mind Brian, but I've posted the entire message here along with my response as a way of exploring what is possible and sharing ideas and information. Given the large number of intelligent fixtures using discharge lamps and the increasing cost of those lamps (£120 for a 1200W MSR, for example) I thought it would be worth exploring the idea of converting these fixtures to LED. Anyway, here's Brians message:

 

Hi Mark,

 

have you had any thought on LED conversions for wash units?

 

I've got 4 Robe 250XT washes sit here. They are currently for sale but if they don't fetch a sensible amount then I may think about converting them to LED. I'd be able to lose quite a bit of weight: ballast, PFC capacitor, mains filter capacitor, ignitor and maybe even the DC supply transformer (swap with a SMPSU).

 

At the moment the optics are very simple - an MSD 250 lamp sits in a deep reflector, the light exits and passes through a 1" hole in the chassis and an IR filter. A set of CMY flags, a fixed colour wheel a dimmer and a frost complete the colour change module and a fresnel lens on the front brings it all together.

 

My problem would be getting the light form the LED through the first hole. I see ledengin do lenses for their 40W LEDs so this might be the easiest way to go.

 

The clever bit would be to sniff the DMX coming in and use it to drive an RBG led to do away with the colour flags and dimmer flag. But that might be a step too far.

 

Any thoughts?

 

Brian

 

Quite a few thoughts, actually :) It sounds like your lights are a lot like the old Futurelight design, which many other manufacturers have used (i.e. lamp/reflector as you've described, colour 'flags' instead of wheels, colour wheel, beam shaper/frost wheel, manual focus). You're in luck because I've got an old Futurelight MH 640 at home which has got some electronics issues, so I can use it to test a few ideas.

 

I would try and retain most of the original reflector first as it might do a better job of collecting the light from the LED than the LEDEngin optics will. I would try and site the LED in the same place as the discharge lamp, but build your LED holder such that you can move the LED backwards and forwards within the reflector so that you can get the best light output. You'll most likely need to cut some of the reflector away in order to fit the LED, but that won't matter as LEDs emit all of their light forwards, unlike a discharge lamp which will send some of it toward the back of the lamp.

 

I wouldn't bother trying to get rid of the dimmer or colour flags as I believe you'll get better dimming and colour changes with mechanical flags rather than doing it electronically. After all, people quite rightly complain about the quality of LED dimming particularly at low levels, but they seem to find mechanical dimming of discharge lamps to be more than acceptable. The only thing I'd add, if your LED driver supports it, is a switch which turns the LED off only when the dimmer is fully closed. This would be very easy to incorporate into existing fixtures. Much easier than try to determine the dimmer level by 'sniffing' the DMX.

 

Hope this helps. Do feel free to ask any more questions. You've piqued my interest, so I'll be fitting an old low Wattage Lamina LED to the MH640 that I've got just to see what will happen. I'll have to manually move the dimmer, colour flags, and frost wheel, but it will hopefully give me some idea of whats possible.

 

Mark

[dashroom66]

First post, being primarily an electronics designer (not saying engineer as I'm 17 with no formal qualifications:-) ) this seemed like a good place to dive in.

 

For the dimming, electronic dimming of an LED is very easily possible, just very hard to do right:

 

1 - A resolution of higher than 8 bit PWM is needed to make it smooth enough at the bottom end (even if only one DMX channel)

2 - When at the lowest possible value (greater than 0), the LEDs are usually visibly on when looking at the fixture (mainly for units like the robe washes and mac auras). The shutters on discharge units usually minimize this. Depends on if you think this is undesirable or not (I'm just getting in to lighting, tell me if I'm wrong).

 

Quite a lot of the electronic ballast units dim the source for the first half, and then use shutter from half way to blackout. Might the same principle be good for LEDs?

 

Another thought would be White LED + Colour Wheel / CMY vs RGBW led + Colour wheel.

I don't know if that would give satisfactory colours, but if it did it would free up space for other goodies (not familiar with the unit, but I'm sure that there would be stuff worth adding, probs not worth the effort though)

 

All of the above would require data though, for simple things like hybrid dimming you could just "intercept" the shutter data and make a custom controller for the shutter stepper and PWM/BAM driver for the LED.

Other stuff gets into bigger and bigger bodges (or redoing all the electronics from scratch - which is tedious and boring)

 

Also, if it might be used with cameras, CC drivers will be better than PWM/BAM/other obscure modulations as they are much better for ripple, and thus nicer for cameras.

 

George

[gyro_gearloose]

For the dimming, electronic dimming of an LED is very easily possible, just very hard to do right

Exactly. The problem of dimming has already been solved, in my opinion. Just use the existing mechanical dimming. If we wanted to design an LED fixture from scratch then we'd use electronic dimming, and iron out any issues in the firmware (i.e. snapping off instantly versus a more tungsten-like fade even when snapping off, or a shift in colour temperature as the fixture is dimmed). However, since we are simply trying to retrofit an LED to a discharge fixture it makes sense to use as much of the original fixture as possible.

 

I'm not its worth looking too far down the path of custom electronics and electronic dimming in this topic. The idea here to explore the idea of just fitting an LED to an existing fixture. If we start out by adding in complex custom electronics then it makes things more difficult than they need to be. It makes sense to keep things as simple as possible, and that means using the existing mechanical dimming.

[Don Allen]

http://i46.photobucket.com/albums/f138/allend12/Mountek150WRGBWFresnel_zps1b384914.jpg' alt='Mountek150WRGBWFresnel_zps1b384914.jpg'>

Not sure if I have worked out how to use Photobucket but this is a picture of a 150W RGBW led Fresnel. It has a puck with 8 led die, 2 of each colour, a condenser lens to concentrate the output onto the lens, only one optimal setting in the focus range and a temperature controlled fan.

 

It seems a more efficient way of converting a wash unit as it draws a lot less power, does not waste as much light and removes the limitation of colour wheels. Disadvantage is 4 channels needed for colour selection instead of 1. Cost is around $120 plus freight. It is cheaper to buy the new led Fresnel instead of modifying older fixtures.

[dmxlights]

Here is the 1 pic that Don was trying to post

 

 

 

 

 

http://i46.photobucket.com/albums/f138/allend12/Mountek150WRGBWFresnel_zps1b384914.jpg

[gyro_gearloose]

That looks like quite a poor optic setup on that Fresnel. I imagine that with the LED that far forwards you'd only get a narrow beam out of it, as the spherical lens on the LED is designed to focus the light into a narrow beam. I doubt the Fresnel lens would do much to widen and soften the beam, not with the LED that far forwards. I could be wrong though, not having seen one of these units in action :)

 

It is cheaper to buy the new led Fresnel instead of modifying older fixtures.

What??? That Fresnel you've posted isn't a moving head, is it? That's what we're talking about here. Not converting conventional Fresnels, PCs, and profiles to LED. Although I do have a pile of old Prelude Fresnels sitting under the bench.... :)

[dashroom66]

Converting tungsten fixtures would be quite interesting, as the filament takes up a larger area than the arc in a MH lamp. This means that it might work with some of the larger 200-400W leds that have a large area.

Got a few quite nice 529w LEDS and some beefy heatsinks lying around, the driver ain't exactly complicated, so I might try making some LED minuette profiles.

 

Imagine 20'000 lumens coming out of one of those little lights.

 

This needs to be done...

 

:-D

[bigclive]

The biggest problem with ultra high power LEDs isn't driving them. It's keeping them cool enough to avoid rapid failure.

 

I'm thinking that it would be quite interesting to see one of the very standard 100W 10x10 LED arrays in a fresnel type fixture where the largish source (about an inch square) wouldn't be as critical as it would in a profile type fixture. The standardisation of LED array formats is quite pleasing.

 

http://www.ledwv.com/en/images/LED%20100W%20LED100W.jpg

[gyro_gearloose]

The biggest problem with ultra high power LEDs isn't driving them. It's keeping them cool enough to avoid rapid failure.

Exactly. That's why I used a large PC CPU cooler on my Miniscans. I don't monitor the temperature of my LED. Instead I rely on the size of the cooler to keep things cool. Seems to be working so far, but I'd feel happier driving a more powerful LED only if I had some means of monitoring its temperature and shutting it down if it overheated. Maybe a tiny Atmel CPU could be used. They can read an analogue signal from a temperature sensor, and use this to control the PWM signal to a computer fan....

 

I'm thinking that it would be quite interesting to see one of the very standard 100W 10x10 LED arrays in a fresnel type fixture where the largish source (about an inch square) wouldn't be as critical as it would in a profile type fixture. The standardisation of LED array formats is quite pleasing.

They seem to be cheap enough, but I'm not sure which supplier to use. I don't suppose it matters for a one-off prototype, but if I needed a regular supply of identical LEDs then it could get a little tricky. I've always used UK or European component suppliers, so I don't know which of the many Chinese suppliers to trust.

[Brian]

I'm out and about at the moment so long posts are a PITA but I just remembered my fresnel shoot-out post I did a few years back...

 

here.

[dashroom66]

On the note of cooling, has anybody seen this method of fanless cooling for computers

 

http://hardware.slashdot.org/story/14/07/30/1710236/quiet-cooling-with-a-copper-foam-heatsink

 

Doubt it would be possible to practically use in lighting, but it's still cool

 

 

George

[James Remo]

Having just bought an optikinetics solar led and being very impressed with the output I'm now looking at my pile of 250's and 100b's and wondering about doing some experimenting with them....

[musht]

On the note of cooling, has anybody seen this method of fanless cooling for computers

 

http://hardware.slashdot.org/story/14/07/30/1710236/quiet-cooling-with-a-copper-foam-heatsink

 

Doubt it would be possible to practically use in lighting, but it's still cool

 

 

George

 

Would be concerned with dust blockage, but cost may be a more immediate barrier...

 

http://www.goodfellow.com/catalogue/GFCat4I.php?ewd_token=AvjSZ7SMqueLGectTAOJboImFpYsJB&n=r7c6MiZHHwv9KL0VOx9ylzeZDUdOgA&ewd_urlNo=GFCat411&Catite=CU003830&CatSearNum=1

 

Think the VLX uses heatpipes to get away from hotspots.

[timsabre]

The vlx heatpipes are to get around the problem of needing the leds at the front of the fixture but the heatsink being on the back. Other manufacturers have the leds at the back with the heatsink and use light pipes to get the light to the front.

[niclights]

The VLX also has the LED's at the back. The heatpipes just provide a way to attach a large cooler to each array in a more suitable location.