896 leds,20ma,3.3v,12000-15000mcd,32*28 leds,with 12-15v dc power source,a cpu fans.
lcd:8.4 inches,toshiba,800*600 tft.
cost list(chinese RMB,1us$=8.1 RMB)
lcd with control panel:780
1000 pcs leds:350
PCB board:20
12-15v DC power:100
lense:320mm,dia.:67mm 160
fenier? lense(I can't say it in English):330mm,60

I am a Chinese,I hate me,when a learn English,I uasully sleep.I don't think someday I must say it.I am very sorry for my poor English.

no more words I can say,see pictures please.

the led board

led board in a wood box

led board in a wood box

it works

60 inches blue result picture
you can see more pictures at: http://diy2005.aa.topzj.com/viewthread.php?tid=44145&fpage=1

60 inches result picture
you can see all 42 pictures at : http://diy2005.aa.topzj.com/viewthread.php?tid=53643&fpage=1

Is the wave effect in this shot (Post #6) from the led's or from the camera used?

Joe

Looks like there's 2 issues there. A bit of unevenness in lighting from the LEDs and also some from what appears to be the screen.

Without seeing the screen in "normal" lighting, it's hard to tell exactly how much of what is from what.

the bad pictuers is my first led board with 192leds,the better one is 896leds board i use now.

Scrolled through most of the pics he has posted in the link he gave and the wave effect definitely doesn't appear to be a factor. Maybe it is just noticeable when the screen is a solid color? Maybe it is only being picked up by the camera? It actually looks pretty good and at just 60" it is reasonably viewable under ambient light (at least "bar room" quality anyway). There does seem to be a color issue, in the reds in particular. Blues look great. Blacks are decent. Green is good. Yellow may be a bit muted, but reds are definitely lacking- they're almost brown really. Maybe this can be adjusted/compensated with the lcd controls?

Anyway, nice job Tian. Definitely shows a good result with led lighting. I might even try it myself if I could find 900 leds for a reasonable price (say, $100)...not likely in the states...I've seen 6x6 arrays (36 total leds) selling for about $20. That would make your array cost $500 here...ouch...

this picture is under 25w old lights and 2*7w new lights(I can't say them in English)

Yeah, in that picture you can see some wrinkles in the screen, which can be seen in the blue projection.

The LEDs seem to produce slightly uneven light. Still looks better than I thought that it would.

hi everyone,it is 0:51 at china,sorry I goto bed, sleep now.
I will do my best to answer your question tomorrow.

Can anyone source some LEDs for a project like this? The cheapest I can find them is about a buck a piece or 100/$40 on eBay.

Building the PCB to hold the LEDs would be a simple matter with our CNC machines.

放映机看起来好从图片, 但您会采取有些, 或许一些从DVD 吗?

He posted this link before. 1,000 for 60 dollars

http://www.besthongkong.com/product_info.p...5c6ed6ad0c706d6

Joe

Those LEDs are 10,000k color. Most would consider that very blue.

we found a factory to make the prebuild led boards,we sale it in china now,and we want to sale this led board in ebay for everyone in the earth,after we test it more than one month,at about january 2006.
the price is 139us$,include:
1.prebuild led board with 896leds
2.12-16v,5A DC power.
3.other 50 same leds for free
4.other 25resisters for free
5.a small slience fans for this led boards

the leds we use is:3.3v,20ma,25-30 degree,12000-15000mcd,7000K,5mm.

139us$ not include mail cost,you pay it by youself.
if we sale it in ebay,we will give the link.
thks,everyone.

Can we see some pictures using your LED projector to show how crisp the projected image is?

yah...can we see some results..

this result picture is 100 inches(16:9),use this led board at 55watts,by my companion "carpow".
he and i make this 896leds board togather,we know each other in internet,his city and my city more than 1000 miles distance. ,now I don't know carpow is a male or a female,we contact all in internet.

zedrik,are you from hongkong?if you can read chinese,please visit:
http://www.jd-bbs.com/viewthread.php?tid=8...page%3D1&page=1
you can see how i and carpow make this 896leds board from a 192leds board step by step.
by the help of more than 100 persons from whole china in internet,we just can did it successful.

That is pretty amazing

Is the 896LED board being used with the same 8.4" LCD ? If no, how large is the LED array ?

Any plans to make one for a 15" or 17" LCD ?

Is there a possbility of sourcing LED's in the 6000-6500°K range ?

From the photo you posted, it certainly looks like corner brightness is very much improved. Nice work... keep it up.


Thanks for the info,
gs

did you measured the lumens output? do you plan on doing it?

nice work

gadgetsmith,the led board is large a little than 8.4 inches,maybe 8.5 inches,wo haven't check it.
leds can be 1000-10000K,we choice 7000k.
we have no plans to make for more than 8.4',because the numbers of leds maybe up to 2700 if we do 15',it ts too large to us now.when we use this 8.4' led board more than 2 or 3 months,maybe we will plan to do a 15' led board.
thank you,gs.
tian

we have no machine to check the lumens output,but we check it with a PJ has 150W metal halide lamp.I think,our 50w led board=100w metal halide lamp(output lumens)

no.1 picture is made by our DIY PJ with our 50w led board.
no.2 picture is made by a "made in china" PJ with 150w metal hilide lamp

these result picture size are both 100 inches(16:9),and photo by same camera with same mode.

we guess our 50w led board has about 100 lumens output.

I want a 17" version with 6000k

I dont see why u cant do a 17" for more cash... i mean its like me buying three 8" boards and then cutting them all to fit and doing some splicing... i can imagine the temperture in the box is no problem anymore and there should be the ability to make the boxes shorter when using an array like this... also with led lifetimes ranging from 5000hr to 10000hr i can imagine an array like this lasting a very long time.

I questioned the brightness and then realised this is a 8" screen doing 100" so i'd say thats pushing the size limitations of that screen for clerity... a 17" would be astounding. Or atleast thats what i think.

""""""""""""""""""""""""""""""""""""""""
I'm not a member yet so not allowed to post in the Gallery. But Tian's thread was elsewhere earlier and I did post the lumens output of his board there as well as a number of other posts too. Anyway...his board produces 12,096 Lumens calculated as:

Output per LED = 12000-15000 mcd (or 13.5 cd on avg)
Total Number of LEDs = 896

Total Lumens output = 13.5 x 896 = 12,096 Lumens

""""""""""""""""""""""""

this was a private mesage I received after my question about the lumens output. Just thought abou posting here because of the big mistake on the information here;

the each indibidual led has an stated output of 13,5 candelas on average. This is correct, but this does not mean there are 13,5 lumens. There is a maximun brightness at 13,5 candelas on the stated light output aperture of the each led. You must multiply this "average" candelas balue with the stereoradian aperture. This is how light output should be calculated. Work out the solid angle first.

I bet there are not 2000 lumens at all in all the 890 leds. And I would say also there are not 80 lumens on the screen as well. Just my opinion. I am not triyngo to break your efforts, good work anyway .

Yes Rox is right about this. But this project is tremendous fun, and we are glad to see it finally done. There have been so many questions about LEDs since I started this site.

Here is a LED candelas to lumen conversion but you need to know the beam angle LED Candelas to Lumens
Here is the chart from that site to make things easier

beam angle cd/lm
__________________
5º 167.22
10º 41.82
15º 18.60
20º 10.48
25º 6.71
30º 4.67
35º 3.44
40º 2.64
45º 2.09

Tian said that he was using 30º LEDs so the with 12,000-15000mcd ratings taking an average of 13.5cd per LED there should be about 13.5/4.67 = 2.89 lumens
Using 896 LEDs * 2.89 lumens/LED = 2590 Lumens

This compares with a typical LL setup with 33,000 lumens at bulb
13% angular efficency = 4290 lumens
+25% for reflector = 5360 lumens
-10% for fresnel transmitance = 4826 lumens
As the MH arc isn't a point source I'm not sure how much of this light makes it thru the LCD.

The LEDS are not as bright as the MH solution but not as far off as many have guessed. The 30º LED beam will also mean that some of the light won't get through the field fresnel so there are additional losses there.

LEDs don't require rear fresnel, ballast or distance between light and LCD and 55W is much easier to cool than 400W. This is all with 896LEDs moving upto 1500 or so makes them close to equal solutions. There have been quite a few people with 250W MH solutions that seem to be happy so this may not be far off.

does 30 degrees mean that the light is in a 30 degree cone as it leaves the LCD? Irongecko has calulated that light entering the field fresnel needs to be no more than 7 degrees off normal incidence to make it thru the field fresnel in useable form... (from the diffusive light engine thread)

It means from the center of the LED it is 30° cone outward. seems like a 10° to 15° would be better but also depends on how far back from LCD it is.

If Tian only sells 896 led boards, then 4 boards should fit a 17 inch lcd almost perfectly.

yeah I thought of that, if it was right on the lcd it might be ok?

If you look at the web site it looks like there are no white LEDs at lower than 30 degrees. This may be why he stated 30 degrees as the best angle. A lower angle should be better if you can find one.

My only other concern would be that the light is not collimated. At higher resolutions the image may be noticably fuzzier than with a point light source. Only one way to find out is to try it and see.

Edit: I was wrong about the availability of smaller angles. It looks like there are 20 degree leds also.

thats where the 7 degrees comes in, and yeah, I think not only does it get fuzzier, but we collimated light to within 35 degrees usiing a non-fresnel method and discovered that only 1/24th of the light got thru the field fresnel in a useable form(able to be projected) because of the bad angle (not collimated enough) the light was coming in at

Tian,

Where did you find the 7,000K color temperature LEDs. The ones on the web site you showed us were all 10,000K Color Temperatures.

more mistakes here as well;

the 13% light effectivity, this is 2D (4290 lumens) not correct either. Should use stereoradians as well. Lokk for solid angles on google.

The leds arrays, lets say will output 3000 lumens in the best of the cases. But this is like difusive light, very hard to control so it enters the projection lens... I think there are somewhere 100 lumens on the projection. Anyway, it looks like the light efficiency has been improved if we compare projected lumes/lamp wattage;

It coud be great if you could measure the real light output with a luxmeter.

A lenticular array could help in this regard.

Nothing is perfect in any of the cases, we don't have an accurate model of what the frenels can really capture any way as well as non point sourec efficencies. I have suggested before that light from arcs longer than 20mm isn't going to get into a standard triplet and the greatest brightness is at the electrode tips. Just trying to ballpark lumens to lumens. many have said that LED can never work, its been proven by Tian that this can be made to "work" but probbaly still well behind the standard MH solution. I was just tring to point out that while the LED's aren't as bright they are radiating in a dircetion rather than in directions not usable by our optics (fresnels) and that the brightness while less is approaching the MH levels. Tian has shown that this can be made to work.

Pagercam, your calculations may be correct or not I'm not sure but it's a bit unfair to compare Tian's PJ with a typical LL setup because he only used a 8.4" screen. You don't need a 400 W lamp for such a small screen in a LL setup. So if at all, we should be comparing Tian's output with a 250W Metal Halide lamp.

For a 15" screen, typical of a LL setup, Tian could use TWO of those boards. In which case, if your calculations are correct, we have 5180 lumens in Tian's PJ vs. 4826 lumens in a typical LL setup. And this is using less than one fifth of power (36*2=72 watt) of a 400 W Metal Halide lamp.

Another advantage with LEDs is that they typically have a CRI around 75-76 compared with 65-70 of Metal Halide lamps.

Looks positive, I noticed that the LED's are leaded type means they are soldered to PCB through holes. Since LEDs are sensitive to soldering temperature (because the inside joints are via bonding wires to dies) that individual LED may not be same bright like others.
If this project become more mature it is preferred to locate SMD LEDs, the cost should be down if right source can be found, and same to assembly cost.
Uniformity of LED brightness can be easily acheived if from good source and during binning process (to seperate LEDs by their brightness, color represented by wavelength and bias current) they are from same bin or neighboured bins.
I am not an LED manufacturer but I am running an electronic assembly house in China near ShenZhen, and I have visited a lot LED house so knew a little about it.

Thumb up to Tian, I think he did this DIY project quite well and should enjoy fruitful result soon.

LED efficiency is improving by leaps and bounds practically every few months. It may not be long that we see LEDs take over practically every kind of lighting application from projectors to laptop displays to automotive headlights to home lighting.

Check out this thread from /. from earlier this year when news of development of a LED projector came out. A few days ago Toshiba launched what might be the first commercial LED projector. It isn't anywhere near anything like other projectors in light output but its a sign of the direction in which things are moving.



And it runs on a battery.

I think you missed my point. The point was that Tian has while maybe not equaled a full LL projector he has made it work. I've been read these forums for more than a year and everytime someone has suggested LEDs they have been shot down. Tian has proven that he can be done. LEDs have some limitations and MH has others my calculations were simple to gain a basic idea of relative lumens and there are significant benifits as listed to LEDs over MH (less heat smaller size etc...) More LEDs with a smaller beam angle could surpass the MH solution before too long someone needs to put the effort in to make it work. The high performance LEDs anrn't cheap but should last longer than MH and won't have the metal halides problems of getting dimmer over time. People often list the life at 20,000 hrs but that is the 50% brightness time so if you want to keep 90% brightness you really only get 3-4,000 hrs.

I believe that this is a DLP projector and DLPs are something like 60% reflective while LCDs are 5-8% transmissive so they can use a significantly weaker source. If this is anything like the small Samsung projector, it may only be able to project up to 60" images at 200 lumens not bad but no where near the traditional comerical projectors.

I got that perfectly. But I was saying he might have actually surpassed LL setup in terms of the light source intensity. That too at less than one fifth of power.

yes, I agree with the things are being said here.

But I would say as others, that the 30 dergees aperture on the leds output is too much. There is light that is not drected to the triplet because of this.

Anyway, I still would like to know if it is posible for you to measure the real light output on the projection, further than any asumtion we could make, the real measuremnt is a very important information in my opinion. The Real measuremnts for a lumenlab are in the range of 150-200 lumens typically.

anyway, do not missinterpret my words. I recon the leds are improving each time I close my eyes, but I would say the DAY has not come yet . Good work, as I have said more than once .

new points on the calculation again;

the 13,5 candelas MAX average value is only valid for the peak brightness at 0 degrees. It is not constant at full output angle (30 deg), the angle is stated so the edge (15 deg I guess) is half brightness (somewhere 7 candelas), So taking this information in consideration, we can conclude somewhere 10 candelas as AVERAGE in all the led's output (as the best case). This average value considering leds solid angle (30 flat cone degrees = 0.21 stereoradians), average candelas*stereoradians=total lumens...... 10*0.21=2.1 total lumens on each indibidual led.

2.1*896 leds=less than 2000 total lumens.

Now how much would be the effective lumens there directed to the triplet throw the lcd?, this is hard to work out. That's why i was asking for a posibility meassuring the real projected lumens. Sincerelly I don´t think there are more than 100 lumens in any case. Will have an eye here if a measuremnt is planed .