Converting discharge lamp to multi-LED for video projector (maybe)

[N_Cook]

Story with pic so far, adding to previous thread text a couple of weeks back
www.diverse.4mg.com/LED_conv1.jpg
7 lenses , each placed over a 12mm brass cup washer, to seat over the
spherical surface of the 5 inch orange ball valve and flat lens fronts
(set in a wheelbarrow innertube of all things)
Tape to quasi-hold in place and glue with superglue spots when posistioned

www.diverse.4mg.com/LED_conv2.jpg
One of the lenses with supplied white holder on left and 2 of my
abandoned attempts to convert the holders to rear mounts for the LEDs,
mm graph squares

www.diverse.4mg.com/LED_PAR_butchery.jpg
Pic of PAR lamp innards , ES base on right with small fuse inside

www.diverse.4mg.com/LED_7cell_1.jpg
Pic at 90mA
KISS, pipe cutter to cut down a large blue barrel pen to intended even
lengths then adjusted by countersink tool to fit over the coni-spherical
rear of the lens , for proper length and provide a temporary swivel for
each , to active focus each before setting as the 7cell. A basic pcb
copper 20mm fuse holder soldered to each LED as a heatsink, then LED set
centrally over the
blue and hotmelted in place with yellow hotmelt. Wire and solder all 7
together.
I did not think about sliding on some sleeving so each link wire
stripped back a bit in the middle for croc clip connection. Set some
perspex correct height over some paper with a square marked on it and
clamped down . Set first LED and mount , held roughly in place by a
light clamp. Then powered up central LED at low current and adjusted
mount for square die image to be central to the paper square, hotmelted
with green hotmelt into place. Powered up next LED only and manipulated
orientation and setting for its image to lie in the square also . Repeat
in turn for all of them .
The luxmeter I have is not to be relied upon as far as calibration.
With some neutral density filters of 4+3+2 stops = 9stops or /2^9
luxmeter read 6400 lux at the centre, this is in non-linear section of
the meter above 3000 but this same meter read 2,100 lux with same 9 stop
filters, horizontal outdoors when insolation from the sun on a cloudless
day was 860W/m2 (includes UV &IR from nearish weather station). Non
neutral density black silicone rubber over the sensor read 140 lux, the
same as with horizontal and the sun (perpendicular to the sun it read
300 lux that day). This was 90mA , in final use with fanned air movement
would by 4 times that, so about 6 times (plus non-linear correction ,
not yet determined) the horizontal lux level of the sun that day. Still
room for 2 rings of 7 each could be added spherically around the 7cell
and chromaticity varied if necessary. I've now got access to the video
projector for a try out , only one opto for "good" signal out, the other
2 must be for lamp trigger , then on receipt of lamp good then lamp full
on control applied.
Assuming simple on/off opto use , have to check not PCM or something awkward
www.diverse.4mg.com/LED_7cell_2.jpg
repeat image with added stopping of camera lens , trying to show that
the 50% brightness contour image lies within the square
www.diverse.4mg.com/LED_7cell_3.jpg
repeat with only 5mA of power to the LEDs difficult to photograph as
available light only. Note the appearance of some graph paper markings
inside the corners of the square

 

[N_Cook]

Bad news for the Jeremiahs and Thomases.
I have a converted HD video projector , as far as I've gone so far.
Hacked the control system so the original lamp supply is disabled and
the control is fooled into believing the missing discharge lamp is lit.
Jury rigged at the moment so the fans not in the right place so running
7 cell at only 75mA. But with a colour bar signal a good image to 2.5
foot diagonal in a darkened room. Good colour gradation across the
luminance scale for red and green , the blue the last 3/10 to black are
black but that was to be expected with 7 yellow-white LEDs with low blue
component and existing adjusted projector settings trying to subdue the
over-prominent blue and lack of red component.
With fans back in position I may try the full x5 350mA rating
I well set the projector back to factory settings via rs232 and power up
again before settling on which flavour of white LEDs to buy this week.
Probably 12 more for the next ring , mid-white LEDs for 19cell and 14
times the power for the normal 5 foot diagonal.
There is room for another ring beyond , 20 ?, but worsening angle to
what TI calls the integrator rod but I call the light tunnel and
significant expense, but even then a lot less that 400+ squid for a
genuine lamp or 300+ squid for grey-market one

 

[N_Cook]

This arrangement has much more latitude as to precise alignment in front
of the light pipe than the parabolic original. I assume because no
direct axial light was going down the pipe with original, instead a
tight focus at the pipe entrance with a high angle of incidence for the
majority of the incident light.
In the process of getting 12 more 4000K, this time, 1.2W LEDs and 8 deg
lenses plus some directional blue if I need them.
Also sorting out a compact power supply, converting a laptop or similar SMPS

 

[N_Cook]

Some pics of the 19cell concentrator compound lens and LEDs
www.diverse.4mg.com/19cell_rear
and very low , few mA, for camera capture
www.diverse.4mg.com/19cell_lit
to the eye the central 7 , 2700K ones seem even more yellow than the 12
x 4000K ones
At 50mA per cell the RGB balance has come up to
R 360 lux, G 450, B 230 through a colour wheel filter
Today I will add a ring of 12x 15 degree high current 5mm blue diodes
around the perimeter to bring up the blue (if necessary) before I mount
in the projector . Then baffle off to divert the fan air over the LED
cluster before trying at 350mA. From further reading of the data sheets,
as fanned, could be pushed to to 900mA with lamp life derating , but I'd
mount a 1N4148 as a temperature sensor in the clust before going there .
Would only be used at the high rating in a couple of summer months , I
hope. Have an adaptable 42V 2 amp compact SMPS that I'll add some
polyswitches in line with the 3 lines of LEDs and droppers . May have to
move to constant current , dependent on monitoring the strings at high
current

 

[[email protected]]

I guess I would have to say that I am at least a little impressed. Really, if you can do this with units that have reliable light engines that'd be great. At tis point I wonder if using LEDs instead of the old lamps, that even tripanel LCDs would last a bunch longer. If you think about it, most of the time the polarizing filters fail early, and certainly UV and heat are doing it. LEDs of any sort should be lower in those undesriable output wavelengths.

Then we got tis ohter issue. Friend of moine was elling me of these asskicking high brightness LED flashlights. They were a coulple hundred bucks ! What are you paying for thesae LEDs ? Just out of curiouslty if for no other reason. If it costs $200 to make the thing work with LEDs, it is obviously worth it if the lifespan and the advantages for the light engine longevity stand true.

If you think the process is commercially viable maybe you should lkook intomarketing kits for high end projos. Like the Barcos I just finally sold, some places charge $500 for the lamps and there are two each. You only burn both of them when you want 3,500 lumens, any other time save them. It's in the menu.

Fact is, with the advent of flat TVs, projos of any kind are of a bit more limited desire commercially. The only thing that'll make this marketable isto put out the retrofits for high end projos that project pictures larger than a commercially viable direct view LCD or plasma.

It is hard to get detailed saervice info on things like Barcos, they want captive service just like anyone else almost. You - or someone - has to figure out how to fake out the ballast. Then in the case of some, you need the remote - to put in a (fake) serial number for the new lamp when the set refuses to run it anymore - to protect you from lamp explosions of course. Less lawsuits and more money makes it every time.

And then of course you are putting your name on the thing. Moreover most ofthese ballasts run directly off the mains, so if that's your power suply then whatever your equivalent of the US' Underwriter's Laboratories is for safety, you will have to comply with all that.

It could just be a one Man operation and you charge what you charge. A thousand, well some would pay it if it lasts. Whatever units you do, then market the kit, for quite a profitable sum of course. As time goes by, others will reverse engineer your product and replace it, but if they do not do the original reverse engineering you did, it might be advantageous for them to just buy the premade kit. You make the thing for the lamp(s), the power sullpy and just make it connect like the original as much as possible and write the instructions.

In the US, if someone really did have a good light engine and the innards are relatively clean, and the ballast is bad, and they like the set....... The same is probably true there. they might just pay if they know all the advantages. I'll write that for 2 %. If you can do it I think I could sell it..

Shit, next year I am going to go around and replace the motor run capacitors on people's air conditioners for money. I will sell them the idea by the fact that it will save them money and we happen to know firsthand. The yearbefore the AC fucked up it was pulling so much power it was ridiculous, and then it quit. With the new cap in there, 90 outside ? Let's set it at 70.HAHAHAHA.

Things like this are GREEN ! The keep shit out of the landfills, keep moneyin the country and I would imagine either save energy or go brighter. One of the other.

Think about it.

 

[N_Cook]

I guess I would have to say that I am at least a little impressed. Really, if you can do this with units that have reliable light engines that'd be great. At tis point I wonder if using LEDs instead of the old lamps, that even tripanel LCDs would last a bunch longer. If you think about it, most of the time the polarizing filters fail early, and certainly UV and heat are doing it. LEDs of any sort should be lower in those undesriable output wavelengths.

snipped for brevity, as I agree with all there.
This 1080 projector was 7 GBP at a radio rally because the replacement
lamp cost 450 GBP, the ink-jet printer business again.
The logical way of making a 19cell cluster is to mould the compound lens
in one. The geometry is not complicated I think, just the moulding
technique and optically clear epoxy . The recessed lens part of each
lens looks like the dome of a clear 5mm LED and that is set in a
non-tapered cylinder that collects the light, perhaps via total internal
reflection off the quasi-spherical rear outer surface.
If good geometry/moulding then perhaps no need to individually focus
each LED.
The main problem is converting the projector, finding the opto-isolators
and defeating their normal control function

 

[N_Cook]

Some pics of the 19cell concentrator compound lens and LEDs
www.diverse.4mg.com/19cell_rear
and very low , few mA, for camera capture
www.diverse.4mg.com/19cell_lit

Erratum, third attempt, bloody thunderbird/e-s/win7

Some pics of the 19cell concentrator compound lens and LEDs
www.diverse.4mg.com/19cell_rear.jpg
and very low , few mA, for camera capture
www.diverse.4mg.com/19cell_lit.jpg

 

[N_Cook]

final run of the Mark1
the original colour bar chart, sort of chrominance on left of white
stripe and luminance to the right
www.diverse.4mg.com/RGB_scale.jpg
RGB approx values
col 1
215,215,255
......
215,255,215
col 6
0,0,255
....
0,255,0
col7
0,0,215
....
0,215,0
col11
0,0,45
...
0,45,0
col12
all 0s
plus odd stripes etc of white

projected
www.diverse.4mg.com/vid_proj_rgb.jpg
slightly skew and not a proper screen , off white cloth as a screen,
with 1 metre ruler slung across but as dark wood not obvious as a ruler
and my camera does not like low light values. The green to blue and red
to blue in the lower right corner is an artefact of the camera unless
brain/eye is wonky too.
Just as well the human eye can accommodate low light values better than
my camera
projecting a white image, the received intensity is the same as a
regular CFL lamp marked 850 lumens at a distance of 5.2 metres.

Not really bright enough so I know what the mark2 version will be.
The problem is getting the light focused into the 4x6.5 mm tiny
aperature of the light tunnel. Those hexagonal lenses are too large, so
by the time you have 19 the off-axis angle to the light tunnel is
getting too great. Will visit a bead shop and try to find mirror glass
beads with about 5mm core that will accept sliced and polished domes off
any old water clear LEDs as lenses set in the mirror tubes and mounted
over the 1.2W LEDs, and individually align to the aperature via an
intitially tiny amount of hotmelt string to the LED rear and then a
matrix of lacing cord and epoxy attached to the 20mm fuse clips added as
heatsink to the rear of each LED. Need as much unobstructed air flow
through those vanes as possible.