I have a crazy idea that well... it's not really crazy, it's just insanely expensive!

Anyways, roughly how many lumens do we need hitting the LCD panel?

I just need a rought answer... a guestimate from the experienced builders here. Roughly how many lumens are hitting the LCD in a standard 400W setup?

use the inverse square law which states source intensity divided by 4*pi* radius of object from point source = intensity or I = S/(4*pi*r^2)

I'm too tired to plug anything in.. going to bed

take how many lumens hit the screen, divide by 0.75 to account for triplet, divide by 0.07 to account for lcd (neglecting fresnels)

eg 200 lumens out = 3800 at LCD

now tell us your idea

I'm not sure I want to be laughed at just yet..

Basically I'm thinking of a diffused light engine with either 3528 Lumens or 10584 Lumens, at a cost of $825CDN or $3665CDN respectively.

LOL - OK, now that I typed that out I feel foolish for even researching the idea, but still...


4000 Lumens is approximately what we get at the LCD panel? hmm..

well, figure a 5% lcd panel, 5000 lumens at the lcd(assuming unsplit) will give you 250 lumens on the other side and after triplet will give you a shade lexx than 200 lumens on screen..... so I'cd go for the 10,000 myself....more like 400 lumens onscreen then which makes it very watchable in ambient light....

Well, maybe reality is setting it.. I would give it a shot but I think the RGB balanced would be @$#%ed. So I'd waste $4000... It's so tempting though! Maybe I'll take some pics later today.

One question, when in a comercial projectors says that have 1200 lumens (for example), what refers that lumens? screen, LCD or lamp?

Thank you.

Generally it's called ANSI lumens on commercial units. It refers the the brightness at the screen, using an average of 9 different points. (at the centers of the boxes of a tic-tac-toe board)

cheers,
gs

I think ANSI lumens is screen based, but based on simple colours,, like cartoons is the brightest version,,, however coming back to movie replay and the millions of colour combos,, most PJ's are reduced to a lower level,, much lower,,,

as to ANSI,,, google it,,, wikipedia has reams on the subject,,, as per above,, simple of my statement is...

Not sure exactly what you are wanting to achieve, but lux on the far side of the lcd should be atleast 7,000 on a 17" 10,000 on a 15" and about 20,000 on a 8" lcd. That again is LUX on the opposite side. I feel it is too hard to get the lumen rating on a close side of the bulb. It isn't gonna be completely accurate I think, but just my opinion,

Keep in mind that if you've got a diffuse light source, the panel will eat all of the off-axis light, so even less than 5% of that 3528 or 10584 lumens will make it through.

Unless, of course, you've got an efficient reflective light recycler too.

Hope this helps,

Well... I am asking for it, but here it goes, a LED Light Engine.


#1675 - 10"x17" Perfboard


#1677 - Mylar to be placed between the LEDs and the Perfboard


#1678 - Superflux LED's, 3 or 9 Lumens each


#1680 - High Efficiency Difuser



I would need 1176 LED's for this setup.

1176 x 3 Lumens = 3528 Lumens

1176 x 9 Lumens = 10584 Lumes

hmmm thats illuminating,, heheheheh pun intended

hey you stole my veroboard idea,,,, geez I have one of those breadboard units gathereing dust,, have not used it in 12 years or so,,

except she has power rails in built twas used to build proto model s up with

P.S.

I am taking DONATIONS at PayPal address: dberladyn@hotmail.com.

All I can say, is I wish you wouldn't waste your money like that, but good luck.

Me too!

I think I could get myself to try the $800 version, but that's kind of rediculous too eh? I just love the idea, it's so clean and simple (and SMALL to boot!).

Anybody else have any comments on this idea? Good, bad, ugly?

If you don't use a rear fresnel to collimate the light then it must be done elsewhere or use a light source
providing same. Light passing through the LCD should be perpendicular to it, +/- something small, like 5
degrees. Because you have fairly large diameter LEDs with distinct spacing between them I think you'll end
up with hotspots patterned as the LED mountings. I also believe a diffuser will help, but not enough.

I'm all for a LED setup, but using many, many, small diameter LEDs with small output cone angles mounted in
the tightest pattern possible.

hey dont knock,,, imagine the view when i went 150 watt instead of 500 watt,,, heheheheh,, nothing is impossible,, just improbable till proven otherwise,,, or your money runs out

Yes, I think you might be right... but, I think the same principle would apply as did in the parabolic setup. If there was a little bit of distance between the leds and the difuser, the field fresnel would probably only see one large illuminating surface.

Now, how bright that surface would appear after the LCD panel is the question. At least to me.

The other problem is RGB colour balance. I purchased several types LEDS, the 9 Lumen Superflux appeared to be the "whitest" light. However, a person might be better off using RGB LEDs and controlling each colour independantly.


Like this LED.

It contains a ground wire, and a lead for each colour: Red, Green and Blue. I don't know what the Lumen output is, but each colour is 20mA, so I assume that by combining the three it should be near the 9 Lumen output of the "white" Superflux.

I have several 5mm "white" LEDS here, but the colour is blueish, I fear the projector wouldn't display reds correctly like Tiam's projector. Look carefully at his pics.

Anyways, I don't know if I want to do this or not - I'm just spending money researching the idea. With some encouragement, I might pursue it.

Lucky

Thank you for your reply.

It is very interesting this thread.

Thank you.

Well, it depends on what your goal is. Mine is atleast 300 lumens 100 lux on the wall or it isn't satisfactory. But again, it depends on what you goal is. I know your projector seems to be very clear, so giving up brightness for clearness is another option. To each their own, but I have proven that a LED light source will not give you anywhere over I think 50 lumens, no matter what at this point in time. Because again, it isn't about the culmination of all the leds. it is about the other side of the lcd, and what lux you get. You are still only going to be able to achieve about 5 to 10 lux, maybe alittle more on the other side of the lcd, which I get 20,000. But I am not going to go into again. Good luck, and keep us informed. I was just merely hoping to stop someone from throwing money out the window.

I haven't seen your "proof", but I did see some heavily opinionated posts in Tiams thread. Anyways, currently I am blinded from looking at LED's. I just picked up a protype board to do some testing. Maybe a LED light engine is not yet financially feasible, but I think it's close to becoming a reality.


#1687 - LEDs mounted on a prototype board.


#1691 - The very RIGHT LED is a 9 Lumen LED, the rest are 3 Lumen.

#1688 - 5mm 50,000mcd LEDs


#1689 - Absolutely blinding!

The reason I prefered the 4pin Superflux LEDs was it's brightness and colour. But another major reason was the ease of mounting. Anyways, maybe Paladin is right, maybe the spacing is too much. The 5mm LEDs can be placed closer to each other, but they need to be filed on all 4 sides to fit properly on the PCB.

Yes, they need to be filed to fit, but you can pack 'em very tight. I did this for a small ~250 LED array to drive a 4" LCD, but that project has been stalled for a while now. It simply wasn't bright enough, but it also wasn't drawing enough current, so I need to fiddle with resistors.

Read all about it in my PLOG, including photos of the ground-down LEDs

I worry that the way I've wired these LEDs means that I'll never get sufficient brightness and reliability. The strings of parallel LEDs are just so long. We'll see.

Hi Lucky,
I do not think that it is a creazy idea. I you look at the large LED TV panels in Vegas they will blind you even during daylight. I modified some panels for YVR and even though I used only 5,000 mcdl LED-s I had to use a brightness control for night operation.
The problems I see are the following: The white LED-s have UV chips, use phosphor and they will burn almost as fast as the plasma displays. Also, the CRI is poor with the white LED-s.
The better way is to go with the 3-chip RGB designs but presently they are expensive.
The third problem is that you will have to provide agressive cooling to the leads to prevent premature aging of the chips.
What I am planning is the following: I will make my PJ modular so I will be able to attach any light engine at ease. I will start with my 575W set-up and until the bulb works I stay with it. I would like to see some more smaller ceramic lights on the market as well as more RGB LED-s.

I'll read your PLOG but you are NOT suppose to wire them in parallel unless you have a resistor for EACH LED. 250 LEDs wasn't very bright, but how big where you trying to make the projection? For my panel I would have anywhere from 1500 - 3000 LEDs, but... I think my biggest concern is having too much Blue light and not enough Red or Green.

EDIT: Ok, you knew about the resistors and parallel etc. Anyways, where is your pics?!?!? I'd like to see.

EDIT2: How did you grind your LEDs, on a grinder?

OK, good luck.

YVR? Do you live in Vancouver?

The one thing I seem to be excellent at is dreaming up ideas and wasting money. This morning I ordered 10 RGB LEDs to experiment with. I never thought of the CRI... I was only thinking of the Colour Temperature. Good point.

Just do me one favor before you spend too much more money. Just do what I have said before. Buy enough LEDS to make a 1" square of them, Then put your lcd up against it, and then read the lux on the other side. My final word. I won't bother you any more on this.

Look, your probably right about me wasting money on this, but you have to admit it is interesting. Is it not?

Go ahead and bother me, your just being that pesky voice of REASON. Anyways, I have only been buying LEDs in quantities of 10, so... I've already wasted more than I should have, but I haven't been excessive (yet). That's a great point, put the leds up to the LCD panel, but I refuse to strip my WUXGA for the time being (it's still possibly for sale).

If anyone is local and wants to try a test array on their stripped panel, feel free to speak up.

Actually, the reason I post about this, is because i have tried about everything, LEDs, Fluorecent, 1000w, 875w, halogen, 400w, and many other things. I am seriously only posting, because it is a waste of money, and wish now, I would have spent all this money on doing it right the first time. I would have been watching what I have now, instead of not being able to watch comfortably all this time. I now know what I wish I knew back then.

OK, this is the kick in the ass I need to make me pick up on further experimentation. I'll mess around with it a bit next week and get some pics up.


I have a belt sander that can be clamped to the bench. Just take the LED in hand (by the leads) with the belt sander running at a slow speed and run the edge against the belt as you turn the LED. Keep your fingers back from the belt! I only lost 1-2 LEDS out of the ~250 I modified like this.

I have a hard time letting go of the LED dream as well.
I have spent many hours studying this idea. The more I read the less I wanted to spend money on the experiment.

Miedosoracing has already suggested what I was going to say. Build a small prototype and place it next to a LCD. We would all love to see that experiment.

The final conclusion I came up with is that if all the light from the LEDs could be concentrated and sent back to the LCD it might work. But now you are back to using a rear fresnel, so whats the point in LEDs. My goal was to remove the need for a rear fresnel. In fact that is still my goal, just not with LEDs.

I would think LEDs are definitely the future of projectors though.
If you try it you can call yourself a pioneer!

I know the feeling, I am the guy that did this:
Parabolic Reflector/Projector TESTS

I would think fluorecents would be the future for this reason. If you take those thin tubes, and put them side by side, all the way down the lcd, you would have more lux to them then led's. I just don't know how to get all the light to the lens in a small dot. Thus about equal watts, more lux and maybe alittle less life, but still good life.

Wow, this thread is moving in real time. Every time I refresh there are more posts.



Fluorescents, Good idea. Had not thought of that. Should have, there use them for laptop LCDs. Although I would think that if the manufactures had a chance to remove glass tubes from the design they would go for it.
Do they make fluorescents with unbreakable plastics?

My last lcd I tore down had two Fl-'s on the ends that would be perfect. I think 12" long, and only 4-5mm in width. Then there was a clear plastic window that grabbed the light from those, and lit up the lcd. But if all the lcd was covered by those Fl's, it would be very bright in my opinion. Just getting down to the point is where you have the problem. Also, I also figured out something else out while retrofiting a commercial projector. The lights, we keep on looking at the arc length. They have a pinpoint arc, because, no matter what light they use, they direct it length of arc forward and back, and the ends point towards the actual lcd. Thus, the arc point is a small point, the corners all get the same amount of light from each side.

The great thing about flourescents is that they run cool... I wonder... lol

I live in Nanaimo but consult at YVR at every second week for 3 days.
If you use the RGB units you will have to use 3 adjustable regulated power supplies and will be able to do the color balancing by varying the supply voltage of the red, blue and green chips.
If you want I can put a small supply together for you with 3 regulators and 3 voltage controls. All I need is the voltage and currant rating of the LED-s. I can set up a supply for you for the 10 lamps.
Let me know if you are intersted, it will be my contribution to your project.

It's entirely up to you, I will PM you my address.

The specs are as follows:

Case: Superflux waterclear
Viewing Angle: 100°
Power: 2,1V red, 3,3V green, 3,4V blue
Wave Lenght: 640nm red, 515nm green, 465nm blue
Current: 20mA per color

Perhaps, I should have just ordered regular 5mm LEDs instead of the SuperFlux. Anyways, if you make this I will post pics and perhaps even a "mini-movie" of the effects on changing the colour balance. Whether, I or anyone actually takes up this project of 1500-3000 LEDs with 4 independant leads... is well, that's a tremendous amount of soldering. 6000-12,000 solder joints, just for the LEDs.

But anyways, I am interested to see. 3 voltage controls would really make things simple.

I am going back to Nanaimo tomorrow night, need a couple of days to do other things and than I will make a the supply for you. I will contact you early next week for your address.

I guess i'll throw this information out, for whatever it's worth...

I have a WUXGA panel, a/g stripped, using 330/650 fresnels with a LL Pro Triplet (500mmFL). For the light engine, i'm currently running an Osram HQI-TS lamp on an overdriven S51 ballast (~550W), a 4.5x6.5 condenser from Surplus Shed, and a LL pro reflector (touching the lamp housing). My screen is 112"x63" and I have an average of 39.2 lux ===> 39.2 lux x 4.55m^2 = 179 ANSI Lumens.

1) Measured lux at surface of exit side of LCD (triplet side) = ~5500 lux
2) Estimated lux at surface of input side of LCD (lamp side), assuming 5% LCD transmittance = 5500/.05 = 110,000 lux
3) Measured lux before collimating frensel (~40mm before LCD) = 123,000 lux (this makes enough sense due to frensel loss, and being ~40mm before the LCD)

This would mean that you would need about 115,000 - 120,000 (or so) lux before a 5% LCD to get a similar lux (lumen) reading on a screen of this size.

The only LED light source I have is a Petzl Tikka Headlamp (about 5-6 years old, running off 3 AAA batteries). The headlamp does have a type of reflector that the LED's kind of "stick through". It contains 3 LED's which are about 4mm in diameter. They are aligned in a row, with an end to end measurement of 17mm. (basically the 3 LED's would fit in a rectangle of dimensions 4 x 17mm) Measured Lux @ 1" distance ~ 10,000 lux.

Now... for some assumptions...

The probe on my lux meter is about 23mm in diameter. I believe that I could easily place 9 of these LED's in an array to increase the total coverage of the lux meter reading head, giving me 3x the lux value. If so, i'd be getting 30,000 lux for the 9 LED's... furthermore, these LED's are based on 5-6 year old technology... if the technology has devised a way for increasing lumen effieciency 3 to 4x, then we are talking 90,000 - 120,000 lux... this would put me near or at the lux (lumen) readings i'm experiencing now.

BTW, move the lux meter to 1/2" from the LED's and the readings jump to ~12,000 lux. (36,000 / 108,000-144,000 for subsequent calculations)

Now... I'm not saying that LED's are going to be cheap... because they are not. They are currently very expensive (at least for the type of array sizes needed for large panels), but I do believe the technology is there, i'm just not sure it's good enough to be feasible. Also... and this is a big 'also'... my 'blind' assumption that LED technology has come so far as from when my little headlamp was manufactured... Someday my assumptions will be realized, but until then, I have no idea what I have, nor if my assumptions make sense. I agree with medios that the best (most prudent) action is to build a small array, and measure the lux output at 1/2", 1", 1.5", etc... to see what kind of lux numbers are being generated... this may give you an idea of what type of output you might expect from your LED array.

Another thing to consider is that LED's may be more effiencent in panel transmittance as they may be more closely matched to the peaks in the color filters of the LCD, and overall may lead to better color and contrast. Although there is little information for birghtness (lux) measurements, there are at least 3-4 others that have agreed that the new 150W CDM lamps are producing "better" results (color and contrast)than 400W setups... if LED's are producing similar results, then having roughly 1/2 the lux (lumen) values stated above, may give similar visual results that are being seen with using the short arc CDM lamps.

Just my thoughts and humble opinions.

cheers,
gs

I did a quick "proof of concept" experiment before I set about wiring all those LEDs in my array - I took a 3-LED headlight (powered by 3 AA batteries) that I use on my bicycle and tried projecting an image using the CRT lens and LCD. Obviously it projected only a small subset of the image, but there's no question that you could see stuff projected. For a small screen in a darkened room where color correctness is not critical, I do believe that a well put-together LED-driven PJ would make for a neat toy.

Whether you can actually end up with a high-quality unit filling up a 6' or 8' wide screen or not is another question altogether.

1) Well, then the DIY projectors here, has 200-300 lumens maximum (in screen), and have more quality than comercial projectors, how???? Because de "normal" projectors has minimum of 1000 lumens...

(and I saw more quality in your projectors, than the medium/high quality comercial projectors with few lumens... why?).

2) And other thing to say is, what distance they use to calculate the ANSI lumens?, because if the distance is big, the lumens are so much less.

Thank you.

When you say commercial units have a minimum of 1000 lumens, are you stating their rated values ? I would say yes to that question. This is because manufacturers "tweek" the PJ's to their limits to get as much brightness to the screen, then claim 'x' lumens... Once this same PJ is installed in an actual location and adjusted properly for watching, the "real" lumens on the screen are around 500 or so... this is well documented. Even so, 500 lumens is still quite a bit better (brightness wise) than our DIY units. I would say quality comes more from the fact we are using 15" 1024x768 (or 17" 1280x1024) panels, while their LCD's are in the sub 1" range.... what this means is that a 15" LCD (or 17") will have much less screen door effect... yes, commercial units use optics to mask the screendoor, which in turn reduces the picture quality.

It sounds like you're trying to compare lumens to quality... lumens (brightness) is only one aspect of the picture...


This is the nice thing about lumens... they (more or less) stay constant as lumens is lux / screen area. As you make the screen larger, the lux go down, but the area goes up, keeping lumens more or less constant. What do manufacturers use for measuring lumens ? ... not quite sure, but I'll bet you anything, it's whatever settings the projector allows that gives them the highest value !

Not sure if i've answered your questions/concerns or not ?

cheers,
gs

Thank you very much for your good explanation.
Then the failed subject of the DIY projectors are the brightness, maybe one day we'll have DIY projectors with 1000 lumens. The post of "Ceramic Lamps 150 Watt Type" of elken2004, show us could be the future, teach us, the performance of the 150W new ceramic lamps, almost working like 400W LL kit. I heard that the 400W version of this lamps coming the next year... would be the future with amazing brightness and performance?

Greetings.

I recieved my shipment of RGB LEDs. These are very cool, from the three different colours, Red, Green and Blue I can make any colour including White. I think the best way to do this was through modulation which I don't have the means, but using resistors I can vary the voltage and create different colours. I think with the right electronics to control a super array of leds, a LED light engine is definitely possible (although very expensive).


#1701 - Green


#1701 - Blue


#1701 - Red

PS. This is the same LED in all of the pics.