Running the XJ-A140 With Missing Diodes

[OzOnE]

I know what you mean about insomnia - not the nicest of things.
I generally sleep OK, but the worst problem for me is headaches and reversed sleeping patterns (I woke up at around 3:30am today).

I bought these PT-54 Phlatlights...

http://uk.mouser.com/ProductDetail/L...%2fwkW3g%3d%3d
http://uk.mouser.com/ProductDetail/L...urnC90Cg%3d%3d

And I think this is a close match to the original Ca$io Red LED...

http://uk.mouser.com/ProductDetail/L...2bJCVfsw%3d%3d

I know they're around £25 each, but they used to cost a great deal more. This is a decent price now tbh - they are of course a perfect match for the Ca$io - even the size and shape of the emitter is important for matching to the shape of the DMD chip / optics / heatsinks. This should ensure efficient collection of the available light.

The way they explain lumens in general can be a bit confusing. Wiki says that "the unit of lumen is color-blind"...

http://webcache.googleusercontent.co...&ct=clnk&gl=uk

Originally I thought this meant that lumens don't take into account our eye's sensitivity to different colours, but this is in fact the opposite - the paragraph above the "color-blind" statement explains that the lumen DOES in fact take this into account...

IIRC, the colour mix for pure white (in LUMENS, not lux) is approx 60% Green, 29% Red, and only 11% Blue...

So, the Green Phlatlight is rated as 1550lm @ 30A - using the percentages above, this gives almost exactly 750lm for Red (which it is actually rated at), and around 284lm for Blue (a bit less than what the Blue Phlatlight is rated at). You can see these are pretty much the perfect amounts for producing a pure white after mixing.

At first, you might think that because the eye is more sensitive to Green that the Green lumens figure would be LESS, but the Green figure is higher for the very reason that the lumen DOES take this into account! The graph on the Wiki makes things a bit clearer.

(ie. The lumen is literally the "perceived" luminosity of a colour - Green "looks" brighter to the eye compared to Red or Blue given the same power).

This took me ages to get my head around when I was researching this stuff a few years back. lol

See what I mean about the voltage drop of each colour LED (on the Mou$er pages)... The Red is only 2.6V, but the Green and Blue are around 4.9V.
This means the poor Ca$io driver will need to dissipate more watts when running either Green or Blue! As an example, given a continuous current of 13.5A...

Red: 2.6V * 13.5A = 35.1 WATTS
Green: 4.9V * 13.5A = 66.15 WATTS !!
Blue: 4.9V * 13.5A = 66.15 WATTS !!

(This is higher than what I recalled in a previous post, but remember that each colour pulsed, so the average current is quite a bit lower).
I don't know how well the Ca$io driver or the peltier cooler will cope with this long-term?? We might need to find a different driver for Green or Blue?

btw, are the Phlatlights actually pulsed closer to 30 Amps in the Ca$io?

My modded PJ is currently running OK with a Green Phlatlight though - the heatsink does get quite hot compared to the Red LED. As long as the driver holds out, we might be OK?

I did attempt to force a Ca$io driver PCB to enable the LED (with just the power cable connected) but I think I damaged the small transistors, or possibly the LED driver chip?

I only have the one spare PCB, so I'll see if I can repair it.
If not, does anyone in the UK or Europe have any spare drivers available? I would also consider grabbing some from the US if the postage is cheap.

OzOnE.

[OzOnE]

Sorry, I forgot the timing info...

I haven't measured the timing accurately yet, but looking at the last o'scope shot...

http://imageshack.us/f/190/redledencidx.png/

...There are two cycles of colour changes per frame (so I guess you could say the Ca$io has a "2X" Colour Wheel). The colour durations are roughly like this (starting with Blue because that's where the colour wheel index pulse starts)...

1.2ms Blue.
4.2ms Red.
2.9ms Green.
1.2ms Blue.
4.2ms Red.
2.9ms Green.

And so on.

(16.67ms total per frame @ 60Hz)

The Blue duration is obviously quite a bit shorter - I guess this is because the laser bank shines directly through the clear segment of the phosphor wheel.
This is easily sufficient for the amount of Blue needed to produce white. The problem is how these proportions will balance out when using only the R/G/B LEDs?

I don't know the timings for the actual DMD bitplanes yet. Not really useful tbh.

OzOnE.

[kobra000]

dang gone man,you have crawled into my brain and catagorically answerd each question and then some...i thank you sir!
yeah i had an idea of trying to keep a color-wheel in time with the DMD color change pulses by ramping its rpm (with the wheel on its own power and controller independent of the projo system) and see if RPMs of the wheel would coincide with the color change..if i could find out if this DMD is used in another projector that uses a regular Mercury vapor lamp..i could probably use its color wheel speed as a reference point and then match it...just as long as the first color segment in the wheel lines up with the first color in the shift (blue to blue,red 2 red,green to green just like you said) but while using a plain white LED ...i dont know...i think its worth a shot...see i dont have the technological experience you guys got with oscopes and logic analyzers (im trying to teach myself with this PDA based Multi-instrament 3.2 program,but..its like reading Greek,while being upside down,on the gravatron,while taking a low altitude chemical flight!)

you said you had tried the high power blue LEDS from ebay...do you mean the planar ones that run at like 30v 1.5 amp 3000-8000 lumens...cause thats what i was going to try (already on the hook for one for 45 bucks).i even found some collimator lenses for planar LEDS on Alibaba.com and some other Chinese wholesaler websights..but not knowing the focal length..no clue which one to buy (their not that expensive though so i was going to run the gamit) but i was interested in your trial and error with these ..what went wrong? too much power? lack of collimation?..they cooked themselves?

so i was correct in assuming that you dont need each color to be equal in lumens. they can be all different levels and still have the same effect..interesting. ..but would it still be beneficial for them all the be equally powered? just perhaps? sounds like if they were all equal..it would eliminate alot of guess work.(and keep the R&D budget down)



its good to see this thread going again..i love it!


Merry Christmas everyone

Keith.

[OzOnE]

I wouldn't worry too much about the colour wheel stuff - if you're talking about simply sequencing the Red / Green / Blue LEDs, it can be done a lot easier than that.

I forgot that you actually only need ONE connection to the driver PCB to be able to sequence the RGB LEDs. You only need the Blue laser diode Enable and the CIDX signal (which can either also be taken from the driver board, or from main board if it's easier).

So, you just do...

Wait for FALLING edge of CIDX - turn off Green LED, turn on Blue LED...
Wait for FALLING edge of LD Enable - turn off Blue LED, turn on Red LED...
Wait for RISING edge of LD Enable - turn off Red LED, turn on Green LED...

And that's it! Can be done with three lines of AVR or PIC code.

(I know I'm repeating a few things here that are in the link below, but oh well)...

I've tried many different types of high-power LED and quickly learnt that you can't just stick a large LED (in surface area) or many smaller LEDs into a DLP and expect decent light output. The problem is down to something called etendue... Basically, unless the light from the LED(s) fits within a very narrow angle, the collection efficiency would be hopeless. It might work fine in an LCD projector, but not with most DLP projectors unfortunately.

In the end, I had to find the brightest possible LEDs, but with the smallest possible emitting area. A couple of years back, the only things publicly available at a fair (ish) cost were the 15W LedEngin emitters.

You then need to find a nice way of combining the RGB colours - you could do it with basic dichroic filters, but I stole a dichroic prism from an old LCD projector instead (see the linked photo near the bottom of the following post)...

http://www.photonlexicon.com/forums/...038#post149038

This made it much easier to just glue the lenses to the prism so that the colours combined into a narrow output angle. IIRC, the lenses were bought from Khatod? The heatsink was a bit trickier though - I don't have any kind of metalworking stuff or drill press, so I hacksawed some CPU heatsinks in half then used thermal adhesive to glue them onto an XB0X 360 GPU heatsink!

It did work, but it's still not usable on a large screen size. It's perfectly watchable in the dark with a 40" wide image, but that's about it. (You need a hideous amount of LED light for a larger image.)

The maximum rated lumen output for the 15W LedEngin diodes are...

Red = 465 lm
Green = 777 lm
Blue = 266 lm

And the max ratings for the PT-54 PhlatLights are...

Red = 1050 lm
Green = 2150 lm
Blue = 480 lm

(although, in the descriptions on Mouser, they are only rated at around two-thirds of that?)

My DIY projector was obviously in no way optimized, so I'm hoping a modded Ca$io will be far more impressive. I doubt it will ever be bright enough for comfortable viewing on my 2-meter wide screen, but it will be fun to try.

I still haven't got around to programming the AVR chip for the RGB sequencer, but I'll try it soon - honest.
The big "brick wall" atm is finding drivers for the Phlatlights. If I can't modify the Ca$io ones, it will be difficult to find (or build) any others.

OzOnE.
EDIT - btw, the lumens proportions of Red / Green / Blue light still need to be correct relative to each other so that pure white is generated after mixing.
ie. you still need to have the 60% Green, 29% Red, 11% Blue lumens relationship.

I know what you mean though - surely this would mean equal power or "lux" for each each colour (as opposed to lumens)?? Very confusing.
I guess in summary, the output "wattage" of each colour would be fairly close?

[kobra000]

what about something like the SMB-160 LED by phlatlight

more data here http://www.china-lamp-suppliers.com/info/show-5912.html

awhile back i talked with some dude at phlatlight and he said the SMB160 and another one (cant remember the number for it) can do 1500 and 3000 lumens of light and can be individually pulsed at different rates or independently addressable
(im still trying to find the emails with the guy it was a few months ago and i get 100's of emails on all kinds of crap a a day ....ADD with ADHD (in stunning 1080i HD!!)..is a real pain in the butt..means i got about a billion hobbies!)
i think i posted some stuff on the SMB-160 and other a few posts back say between numbers 160 and 190..i think im not sure...

OK I found what i had posted ... Running the XJ-A140 With Missing Diodes archive page 8 (cording to google!)

this is a cut and copy of some of my earlier posts ..im sorry for double posting but im reading with one eye here (small bench grinder accident) and dont feel like reading through hundreds of lines of text..so in advance i apologize to those that double posting irks or upsets! and i apologize to everyone for reposting some of my mindless,unscientific and inane rambling....again!.

""""""yeah seems easier to get it to be black and white than anything else.... hey what if we popped a color wheel in front of the light we swapped for the phatlight? all we gotta do is get some form and amount of light to that dlp chip and we got a viable projector...might not as good as it once was but it will be close.or close ish. Im starting to think i might get 3 or 4 lasers and pop back in and see what kind screen i get,just dont know what to do with the remaining laser positions..dupe them with diodes or resistors? or just short crossed the pad. they are roughly priced the same as the leds i have been looking at (LEDs $40-$75 each,lasers $35- $40 each) but im still not wanting to write that monster 100watt 8000 lumen led off
if only 50 percent is getting in and only 20-25% percent of that is making it out the projection lens then that should be about or close to 800-1000 lumens....not bad.but certainly not great. i have found some 300 watt 27,000 lumen....but those things are so hot you can practically weld with them.

I know phatlight has some RGB leds out there...not sure of the currrent or voltage draw...here is a great .pdf file on them..http://www.nefc.com/NEFC_Downloads/N...6.pdf....shows a graphic illustration which states that no color wheel or filtration is needed when sending the light to the DLP (look at page 10)...........(YES LOOK AT PAGE 10! ITS INTERESTING!)

also this pic shows the basic principal i guess its using dichros for color mixing. 28139
and here is a link to the downloads for all color phlatlights http://www.luminus.com/products/color.html
like you said if we can get the phatlight to pulse right...we might have it with a RGB Phlatlight in place of the RED phlatlight..then all we have to do is dupe the ENTIRE flexpcb cable with diodes or resistors or a big resistor to get it to turn on the led.
hmm..that just might be it. and im sure the Phalight is easier to collimate than other Leds.. here is a thread over at LPF showing a dude messing around with a red one(out of the xja projo i assume )
but im more interested in the way he collimated it then anything else. does the red phatlight on the projector have its own collimator lens? cause thats what i constantly see when i see pics of it...this large dome lens in front of it.
http://laserpointerforums.com/f66/ph...tml#post739295
Here is the data sheet on the SMB-160 it says 550 lumens @4amps..but i heard these things can be driven as high as 9 amps.. (wonder if the projector is capable of that..if not i know a external power supply is,just dont know how we would pulse it if it was externally powered)http://www.luminus.com/products/data..._Datasheet.pdf

I see the pros of usin an RGB phatlight like this
its got its own TEC cooler and heat sink
the projector is already expecting a phatlight to be there..so as long as the voltage and current are the same,shouldn't be any different
pulsing a single led might be alot easier than trying to get 2 leds (blue and red)to pulse together,not to mention whatever the phosphor wheel needs.
using the stock red led location,we have more of a straight path to the DLP..a few dichros on the way but that might be a good thing.
dont have to worry about the voltage and whatnot of the Flexcable.cause it wont be used.except for a dummy load to get the projector to turn on.
i really dont think we need the phosphor wheel any more with this setup.

The only con i can see is this
getting as much power out of the RGB phatlight that we can with whatever voltage/current is usually sent to the red led.(if insufficient voltage is encountered,perhaps a transformer/buck boost of some type could be used)
collimation of its light (but i think this is already done with the stock red led lens correct?)
pulsing it at the correct frequency to match the DLP chip.and get a smooth color gamut. do we pulse external or try to let the main board do it?.(i dont think it can be done with the main board.) <----THIS STATEMENT NOW DEFUNCT WITH OZONES WONDERFUL WORK!

i know im going to try the RBG phatlight.wish me luck (YEAH RIGHT! 163 BUCKS FOR THE SMB 160! i get get alot of lasers for that!)


Keith.

ps

just found this....this is neat!http://www.dealextreme.com/p/144w-rg...t-dc-12v-73695
Also..i think something like this would be perfect for dialing in our color intensisty if a RGB led is used...of any type id assume...http://www.dealextreme.com/p/28-key-...dc12-24v-73676 led controller with 28 digit remote control controls dim control and individual LED color control.
something similar..but i belive it allows for faster or slower pulsing..not sure..remote says speed up/speed down. http://www.dealextreme.com/p/144w-rg...p-dc-12v-94206
kobra000
10-22-2011, 15:50
well i contacted the maker of phlatlights and this is what they said in their first email:

"Thank you for contacting Luminus. We do offer 2, standard RGBW products, each having individually addressable LEDs: The SBM-160 is a surface-mount package capable of generating up to 1500 lumens at full power. The CBM-380 is a coreboard package capable of generating up to 3900 lumens at full power.
Product datasheets with much more detailed information for both of these LEDs can be found at our website (www.luminus.com) under the "Products / Color Mix LEDs" section.
Best regards,
Andrei

Second email is as follows.

Hello Keith,
Yes - each LED can be independently controlled. With all 4 LEDs (R, G, B and W) on at full current you will achieve maximum total luminous flux (lumens). Since each LED can be controlled, you can achieve a pure R, G, B image and practically any combination in between by driving each LED differently.
Luminus LEDs can be purchased through our distributors, Avnet (www.avnet.com) and Mouser (www.mouser.com). Please check the distributor websites for stock/availability of these parts (the SBM-160 is quite new; not sure if the distributors have this product yet).
Best regards,

Andrei


Im pretty confident the CBM-380 will do well in place of the Red LED..now we just got figure out how to pulse it in time with the DLP... (think that might be that difficult?)
Now if i can just get over the 163 dollar mouser price tag..thats the price of 4 lasers....yeah the idea of putting the lasers back in is starting to look real good.

Keith.




Ok that was the archive from page 8...(weird how everything you ever say out there ..still...is...out....there!) again sorry for double posting....im headed to bed!)

merry x mas yall.


Keith

[kobra000]

http://www.youtube.com/watch?v=OqNIWRzDUeE

check out this video of the CBM-380 in action....now im just spitballing here but i think this LED shows real promise..and cant we just use the orginal red LED phatlights collimator lens?.....now just gotta address each LED in time with the DLP..(with this new data that Ozone has graciously provided us..i dont see this as being an impossibility!)..

has anyone ever heard of an Arduino?..if anyone has could you give me the low down on it?

[kobra000]

Ozone


so the whole problem with using the High power led thing was lack of collimation? ie not getting enough light in a small enough angle to pass through the dlp?

[OzOnE]

Hey, I've apparently got Asperger's too. Welcome to the club! That's probably why I take on too many projects as well. I've had to train myself to switch off sometimes and limit the number of projects to something manageable.

so the whole problem with using the High power led thing was lack of collimation? ie not getting enough light in a small enough angle to pass through the dlp?

Basically, yes. Collimation won't help if the surface area of the LED's emitter is too large though...

If I'm understanding it correctly, the "etendue" value defines maximum angle or "cone" of light that can fit through any optical system. This is especially important in a projector...

If you imagine drawing a straight line from one edge of the LED emitter, through the lenses, onto the edge of the DMD chip, then out through the main lens, there will be a maximum width of the LED emitter before the straight line won't fit all the way through to the main lens any more (even if you change the optics).

I think you can use a slightly larger emitter if the DMD chip is bigger, but the emitter / light source still needs to be fairly small. This is why the metal halide lamps in DLP projectors generally have a very small arc width compared to a typical LCD projector lamp.

An RGB LED like the CBM-380 would probably work OK as it's small, but you could end up with slight non-uniformity of the colours in the projected image because the different colour emitters are slightly offset.

I did try a cheap "Prolight" RGB LED a few years back, then a Lamina Ceramics Atlas RGB - they worked fine as a proof on concept, and showed that my FPGA sequencer was working, but the image was VERY dim (as you would expect).

It depends what you really want out of the project - if you want a usable projector (probably with a small screen size though) then I would still recommend going with all Phlatlights. If you'd just like to see the projector working with an RGB LED at first, then it's quite easy to do.

To be honest, the SBM-160 is a bit pricey for the amount of light you'll get. Unfortunately it will be near impossible to make use of the White LED because the Ca$io only uses RGB anyway.

For the same price, you could get the Green and Blue PT-54's instead - these will be matched to the original Red, and will hopefully work well with the rest of the optical path. The original Ca$io heatsink and lens is also a perfect fit for a Blue or Green PT-54.

Remember, you'll only get a fraction of that light out of the projector's lens!

I can build you a board if you like which will sequence the LED colours for you. It would contain the AVR chip for the sequencing, plus the drivers for a small RGB LED on the same board.

The drivers would be based on the simple LM317 type regulator (same type I used for the LedEngin LEDs)...

http://www.instructables.com/id/Circ...h-Power-LED-s/

The board could connect to the Ca$io PSU (should have enough spare juice for this).

One issue is that I don't know how to bypass the Red Phlatlight yet (so the Ca$io driver board still outputs the timing signals we need). You might need to leave the Red Phlatlight connected for the time being?

Another issue is the dichroic filters in the Ca$io both let Blue light pass and the first filter would reflect the Green light in the wrong direction. If you kept the Red Phlatlight in place, then only drive the Blue and Green sections of the RGB LED, you would need to rotate the first dichroic filter by 90 degrees to get the Green light to reflect towards the second dichroic.

Also, the Blue light would need to pass all the way through the colour wheel, off the mirrors, then through the second dichroic (like the original lasers do). I don't know how well the Blue light would match with the Red and Green light if this happens?

There are very few configurations which would work for this LED experiment because the Ca$io essentially has the wrong "type" of dichroic filters.

btw, I use Arduino boards all the time now. I have one sitting here as my 3D glasses controller. I don't like the Arduino's own software, so just use AVR Studio instead. (ie. the chip is generic, it doesn't care which AVR software you choose). To me, the Arduino is just a cheap and simple way to buy an AVR chip on a nice PCB with a clock crystal, voltage reg, power socket, pin headers and integrated USB-to-serial chip. It's all programmable in C and program the actual chip via USB with simple software.

Let me know if I can build you an RGB LED sequencer and I'll see what I can put together. I can't promise timescales though.

OzOnE.

[kobra000]

Grrrrrr....we have the LED's we have the DMD....heck im thinking about gutting it and recasing it so i can have more room to move and install and whatnot...I picked up a back of lenses like 60 or 70 of them ball park (some were stuck together which i thought were just large lenses so in short i have quite a few on hand.)..from a army surplus store...dude said most of them were from rifle scopes,microscopes,telescopes,binocs,range finders big lenses,small lenses,circle lenses,oval lenses,flats,fresnals (i heard these work good for collimators!)...and the like....i figure.there has got to be something in there that might help me either one or a combo of some,im already on the hook for the Blue led i got from ebay (4000 lumens 470nm blue)....so i gotta work it.who knows perhaps ill stumble across it..im thinking of taking all the optics..the dichors everything out..taking the dmd and mainboard and and all the critical components and recasing it..then i i need/want/or have to i can hook right up to the dmd or as close as possible.
yeah i see what your saying with the collimation/entendue...ive been talking to a projector expert and what he said is basically what you said...but i think the problem with this projector is all the bouncing it does,with the dichors and this and that...i wanna start trimming the fat of this piggie...

surface area is too large....hmmmm..what if we took a condenser or smalle projector lens...like say from a cell phone projector,or toy projector or pico projector...and used it in reverse...what im thinking is this we virtually "shrink" the emitter size down to a useable size..and use the lens to project/throw the light here and there...i dont know..rough idea im having trouble explaining. with all the lens combos and devices out there ...there has gotta be a solution..and hopefully a cheap one.

as for the sequencer...right now i dont have the money for it..(i have a fiber optically thin budget man its pathetic!) so something like that ill have to put on the back burner for now...(what would the ballpark price be?)...
hmmm the only exp i have with C programming is the "learn C++ in 21 days" book i use as a paperweight (bought it in like 99 or 2000)....and i have never once even looked up a Arduino video....(but im very curious about them,i figure every other project on the net calls for one,why not this project ya know?)
i did find this wierd thing on one of my all night internet surfing jags...its a variable lens Iris...kinda like the eyes of the Robot Johnny 5 in Short Circuit...they were rather expensive 78 bucks a pop..but the interesting part is..they had a very short focal length (less than 25mm or 1 inch) but also when you rotated or "focused" them via a movable wring,which closed or opened the iris,and another would move a collection of lens back and forth..very minute adjustments!!! they said it would change the angle of collimation..even stated that it could take it to precisely 1.5mm..i dont know crap about collimation or stuff like that,but thats gotta be in the ballpark.ill have to hunt through my bookmarks to find it (i have literally thousands!).i dont know what they were used for..lasers calibration,some other type of projection,Video camera equipment...but they were only 1.5" in dia and like 2" long....i thought it would be perfect for something like this...(a variable collimation lens?? awesome!)...when i find the link i will post them. might be worth the 78 bucks.

the data and mods you have provided have given us a huge huge step in the right direction...now if we can just nail this collimation or entendue problem,we would have this thing in the bag man!..
Do you still have the Hi power LED's you bought?

thanks

Keith.

[ekeup]

Question/recommendation...

Wouldn't it be easiest to change/modify the phosphorescent light wheel with a RGB wheel and just use a white light source via the blue light path?
Using the existing wheel would allow the projector to keep it in sync, right?