[DIY] DIY/modified DLP projector for page-flipping

Tutorials on how to create your own rigs, pics, movies, and everything that has to do with S-3D at home!
Post Reply
User avatar
Petrus
Two Eyed Hopeful
Posts: 95
Joined: Thu Mar 04, 2010 12:50 pm
Location: France

[DIY] DIY/modified DLP projector for page-flipping

Post by Petrus »

When my 85Hz projector's colorwheel explode, I searched for a replacement projector and I don't found any 120Hz DLP projectors in the market that follow my needs of throw ratio, optical offset and brightness. Then I thought about hacking a not-120Hz projector before considering that it will be way too complex for me.

But there is a growing interest here about modifying DLP projectors with custom hardware and LED lighting so I made this topic to talk about it.

For inspiration, you can see this High-end LED+DLP projector dissection from a German website : http://www.cine4home.de/tests/projektor ... o_Test.htm" onclick="window.open(this.href);return false;
OzOnE2k10
Sharp Eyed Eagle!
Posts: 436
Joined: Wed Jan 13, 2010 7:12 am
Location: UK

Re: [DIY] DIY/modified DLP projector for page-flipping

Post by OzOnE2k10 »

Hi,

Now THIS is a great area of interest for me. ;)

@Petrus: I know I sent most of this info via PM, but I'll try to include some of the more practical stuff here...

The following is NOT a complete project guide, but it gives a general outline of what I used to build my LED projector...

(Please let me know if I'm bombarding people with too much typing.)


Modifying a DLP projector with high-power LEDs was the project which got me started with FPGAs about 6 years ago. I needed a way of sequencing the RGB LEDs so that the colour switching would match the original (removed) colour wheel...

I bought an old Proxima projector at first, then wrote my first Verilog code to interpret the serial data being sent to the projector's colour wheel motor drive chip. Many DLP projectors use the Allegro A8902 (or similar) motor chip...

http://www.datasheetarchive.com/pdf-dat ... 60607.html" onclick="window.open(this.href);return false;

I did try to monitor the serial data using a PIC chip initially, but I think the serial speed was too fast. The FPGA worked perfectly after some minor bugfixes.

Of course, many projectors have an opto sensor to sense when the colour wheel rotates, so that signal is generated too.
Some projectors (like the Proxima) also require a "tachometer" signal, which is just a square wave which toggles each time the wheel rotates. (I had to cut a small track on the PCB, then input the tacho signal from the FPGA directly to the CPU on the projector).

More modern projectors use a different method to detect the wheel rotation - there is often a photodiode with a yellow filter on it - this acts as a colour sensor to tell the projector when the blue segment of the colour wheel has just past (and Red / Green is about to start). This signal was also simple to generate.

If the LEDs are bright enough, you wouldn't need to connect this signal because the PJ would sense the colour changes directly. I did find it to be a huge help when aligning the lenses though (the image wouldn't freeze or go blank etc.)

I looked far and wide for the brightest LEDs around at the time (many years ago). I soon found out about the "Archilles heel" of optical design - Etendue...

http://en.wikipedia.org/wiki/Etendue" onclick="window.open(this.href);return false;

Basically, the light source(s) need to have a very narrow angle of emission in DLP projectors. The point from where the light is emitted also needs to be kept very small. If you don't do this, any light outside of this "cone" will be wasted.

In other words, the LEDs you use need to have very high brightness output from a very small surface area. (You can't just bunch many LEDs next to each other without loosing massive amounts of light efficiency.)

With LCD projectors, the Etendue problem generally isn't as bad because the LCD panels are often bigger than DMD chips. The DMD requires a narrow-beam light source too (because of the way the light is reflected off the mirrors).

Then, for full-colour you of course need to combine your RGB LEDs using a dichroic prism block or filters (also in a small angle). The article which Petrus posted is an excellent example of how an LED projector "should" be built (great Web site).

So, I had to find some suitable lenses with a narrow output angle, then build an RGB combiner using the dichroic prism from an old LCD projector. It looks like this (FPGA board removed for another project)...

http://img31.imageshack.us/img31/9523/file0085.jpg" onclick="window.open(this.href);return false;

I used the brightest LEDs that I could find for the money (a few years ago) which had a small emitter surface. I won't mention the name, you can see them in the photo. :D They are rated at 15 Watts per LED!

The image brightness with this setup was pretty impressive. I wasn't expecting spectacular results because I'm not an optical engineer and I just used whichever lenses gave the best results. The image was perfectly watchable in a dark room, especially if you keep the image size around 4 to 5 feet wide.

I also had to build my own high-current LED drivers (top-left in photo). They basically use KA350 adjustable regulators (higher power versions of the LM317) and some low-ohm resistors for current feedback. My drivers are only ON / OFF switchable, but you can actually adjust the relative brightness of each LED colour using PWM anyway.

The power supply for the LED drivers can be a simple laptop PSU provided it has sufficient current output.

Another point is that you'll need to bypass the PJs original lamp ballast signals. This is usually fairly straightforward, but some projectors need specific timings to work properly.

So, here are my ancient project files for Altera Quartus. Remember, this is the first Verilog code I ever wrote and much of the design is done as a block diagram. The code would be FAR neater and more compact if I was to tidy it up now... :oops:

http://www.mediafire.com/?8eda3p7mx1m7p1z" onclick="window.open(this.href);return false;

The code needs to be modified to match the specific timings of YOUR original colour wheel. I tried to keep the angles in Degrees, so you'll need to measure the angles of your colour wheel segments as accurately as you can. The code is quite well commented, so you'll see how basic it is.

You'll need to see how your donor projector detects the colour wheel rotation too. Many PJs still use an opto sensor next to the wheel itself.

I can't guarantee that this code will work with your specific projector - there are probably many different clock speeds used for the motor chip (which would affect the speed of the wheel rotation). Modern PJs can also have some rather complex colour wheel sequences and many use a dynamic Iris etc. Lots of issues may arise depending on the projector you use.

I added some PWM code blocks to the design so that the LED colours can be "tweaked". I found that this almost always made the image worse - you can only ever reduce the overall brightness when tweaking the colour too.

A few months ago, I "stole" the FPGA board from the LED projector for another project - I couldn't afford to buy another FPGA at the time, but I'm tempted to put it back together now. ;)

btw, the PJ in the photo is an IBM iL2215 (re-badged InFocus LP335). It was one of the more light-efficient PJs I could find (ANSI Lumens vs lamp wattage).

Important note!...
Unfortunately it is very difficult to modify the original firmware of a commercial projector to speed up the colour switching (unless you have the original source code :( ). So, even though the LEDs can be switched MUCH faster than your average colour wheel can spin, you will still have a similar amount of rainbow effect etc.

Also, I'm not sure how much brightness a DIY LED DLP can achieve, but please don't expect miracles. With the newer high-power LEDs it might just be possible to get very impressive brightness - you bank account will start to suffer though.

Speeding up the actual frame rate that a projector can display is almost impossible. The only method I could think of which doesn't involve trying to modify a projector is ColorFlipping (still work-in-progress after a six-month break!)...

http://www.mtbs3d.com/phpBB/viewtopic.p ... P&start=60" onclick="window.open(this.href);return false;

OK, I think that's enough info for now; My typing fingers are aching. :ugeek:

OzOnE.
User avatar
tritosine5G
Terrif-eying the Ladies!
Posts: 894
Joined: Wed Mar 17, 2010 9:35 am
Location: As far from Hold Display guys as possible!!! ^2

Re: [DIY] DIY/modified DLP projector for page-flipping

Post by tritosine5G »

I think even 150 lumen with a 8-10gain screen would be da awesomesauce ( hardware brightness slider FTW ) .
This is my favorite topic when it comes to curved screens:
http://forum.dvdtalk.com/dvd-home-theat ... light.html" onclick="window.open(this.href);return false;

Or you could paint something rollable with aluminium if you have space constraints standing against curved screen, for around 3.0 gain. I have 3.0 recipe , only I can't buy those "Behr" paints here, its basically white basecoat, metal flake reflective coat, then white topcoat in a given amount. Vacuum tensioning is beyond my tought process. Maybe I can get screens measured by local profesional reviewer ( he liked 120hz dlp for gaming too, afterall, lol).
So I would want to help you guys with screen too. With LEDs , its sortof necessary too.
BTW, I know a courier so I could send projection screens over to France and UK easily and very cheap (thats how I obtained half ton vintage JBL, and Quad electrostatic LOL ) ;) We need a recipe yet...
Plus enter surround :
http://www.widescreengamingforum.com/fo ... 65&t=18394" onclick="window.open(this.href);return false;

ok thats just plain crazy.
He sells these for 1800 bucks. :P
Fpga knowledge would be def. good to do something like this. ...and all that EDID stuff too nowdays, what a mess... :P
-this overpriced surround, silverscreen business is begging to be disrupted imho.
-Biased for 0 Gen HMD's to hell and back must be one hundred percent hell bent bias!
User avatar
tritosine5G
Terrif-eying the Ladies!
Posts: 894
Joined: Wed Mar 17, 2010 9:35 am
Location: As far from Hold Display guys as possible!!! ^2

Re: [DIY] DIY/modified DLP projector for page-flipping

Post by tritosine5G »

Bump. Any comments about PhlatLight sets? I have an idea of professional paint meanwhile ;)

My other favorite silver screen article: http://lennylipton.wordpress.com/2009/1 ... en-part-1/ (plenty to read on that site!!!)
-Biased for 0 Gen HMD's to hell and back must be one hundred percent hell bent bias!
OzOnE2k10
Sharp Eyed Eagle!
Posts: 436
Joined: Wed Jan 13, 2010 7:12 am
Location: UK

Re: [DIY] DIY/modified DLP projector for page-flipping

Post by OzOnE2k10 »

Hi,

I've probably repeated myself quite a bit here, but it IS 3:00am. :) ...

This silver screen stuff is a bit beyond me atm. I'll have to do some more research on it. I read the article you posted and it was quite interesting.

The Oculus3D system appears to use a fairly simple prism / lens assembly - I wonder why they don't use something similar on digital projectors? Surely the resolution of the digital projectors is getting so high now that it shouldn't be too bad to split the frame in two? That would probably increase the brightness compared to many other methods and bypass the need for shutter glasses and the higher refresh rates?

I'm not sure where to go from here regarding the DMD board? - Should it be just a DMD + Reset chip on a board, or should it include the full FPGA + DDR SDRAM??

If the board can be kept simple (DMD + Reset), then you can use whichever FPGA and memory config you want. This allows for ease of FPGA / input upgrades and for changing to a better DMD etc.

I've been looking at the basics of how Geohot's code works and how to send data to the DMD. I think it will be fairly trivial to control, and the code will be far simpler without the USB stuff. I've also been messing with changing the FPGA to an Altera chip. This is just a quick test, but it's a bit of a nightmare to try and route it under Eagle...

http://yfrog.com/7fdmdboardtestozonej" onclick="window.open(this.href);return false;

http://yfrog.com/bddmdboardschematictestozj" onclick="window.open(this.href);return false;

If your PCB friends can help with the actual routing of the board, then that would help immensely. The above pic also demonstrates that maybe it really is a good idea to keep the FPGA separate? :lol:

An FPGA development board can be bought cheaply nowadays, and even a basic board should be fast enough to transfer some video to the DMD from an HDMI chip. (I'm trying to think of the quickest and most realistic way of getting something up-and-running).

So, I might as well just add some connectors to Geohot's board to directly control the DMD / Reset chip. I don't know which connectors are best or how to calculate the impedances, so can your friends also help with that too?

We'd then have to calculate the bandwidth needed to send the bitplanes to the DMD to work out how fast the RAM needs to be. This is quite a complex process because it depends on DMD resolution / frame rate / colour changes per frame / bitplane splitting / minimum time for mirror reset etc. I'll have to work on this one, but I'm not entirely sure what I'm doing tbh. :?

OK, about the Phlatlights... A few months ago I wrote down the prices of some Phlatlights vs the approx "White" brightness when the RGB is combined.

Unfortunately, it doesn't look like they allow the PT "Projection" type LED's to be sold on public sites, so I looked at at the single-chip "Illumination" LEDs instead. I'm now sure how much light would be lost when using the Illumination types for DLP, or whether it would even be worth trying them at all??

The prices are probably out of date now but it gives a rough idea of price vs lumens. These were the highest brightness "bin" of each colour I could find at the time...

Single, rectangular chip. Prices in UK Pounds. I think the prices included VAT but I can't remember?...

RED CBT-120-R-C11-HM100 1400lm £62.36.
GREEN CBT-120-G-C11-JM200 3100lm £65.08.
BLUE CBT-120-B-C11-KK300 600lm £54.76.

Total: 5100lm "White" @ £182.20. (28lm per £.)


Or, for the next brightness down...
RED CBT-90-R-C11-HK100 1150lm £46.05.
GREEN CBT-90-G-C11-JK200 2350lm £43.94.
BLUE CBT-90-B-C11-KK300 400lm £45.02.

Total: 3900lm "White" @ £135.01. ( 28.9lm per £.)


My current LegEngin 15W emitters are rated at 560, 460, and 780 lumens for Red, Green, Blue respectively. This gives around 1800lm total...

I don't think many of these lumens find their way to the lens, but it gives a fairly bright image in a darkened room. The best thing of all is getting rid of the mechanical colour wheel (and the noise it makes), but my optics are FAR from optimised of course. :oops:

If you can get hold of the PT-120 Phlatlights, then great. But - it will be so expensive in the end, is it really worth it when commercial 3D projectors are starting to fall in price already?

I still want to try and get a DMD board made either way, but you can see how pricey LEDs can get when you want a triple-figure lumen output.

The silver screen idea might be a great solution for LED modding (if the screen can be made simple enough to construct). If you know where I can get some sample sheets, I can try it out on my LED modded PJ. ;)

OzOnE.
User avatar
tritosine5G
Terrif-eying the Ladies!
Posts: 894
Joined: Wed Mar 17, 2010 9:35 am
Location: As far from Hold Display guys as possible!!! ^2

Re: [DIY] DIY/modified DLP projector for page-flipping

Post by tritosine5G »

Again, with 10 gain screen , 150 lm would be more than enough. For a 3-4 gain screen you 'd want 300+ :)
Matte white is bad anyway ;)
Check out these wonderscreens (Look for VUTEC):
http://www.avsforum.com/avs-vb/showthread.php?t=262466" onclick="window.open(this.href);return false;

Then again, very small LED pj's do 300+. Also I want ultimate blacks, so the light routing should be no sub optimal. So LED luma is no problem ;) we want no ultimate brightness here imho.
I saw those 3LED 's on a single chip. How about those. Im goin to talk to my italian friend and ask him about them.

Thats why I tought about using something simpler than those watercooled sets. And that oculus thing, can't we use wobulation for cinemascope 21:9 somehow ? LOL.
( You know , in TV's they use 960*1080 resolution DMD's, and its doubled. Dunno if it works with normal DMD's? )

+ I still don't like the non depolarising screen idea, if we won't have paint, we have to use that one with depolarizing diffuser, then again maybe its also better than paint.

And I have a 1500k xilinx fpga kit with USB you guys can have if you want!
-Biased for 0 Gen HMD's to hell and back must be one hundred percent hell bent bias!
OzOnE2k10
Sharp Eyed Eagle!
Posts: 436
Joined: Wed Jan 13, 2010 7:12 am
Location: UK

Re: [DIY] DIY/modified DLP projector for page-flipping

Post by OzOnE2k10 »

I don't think you'll ever get decent brightness unless you use three separate LEDs (R/G/B). You also don't need to water-cool the Phlatlights initially, it's only those fancy Sim2 projectors which seem to use that atm.

I did try a cheap 20W RGB LED (combined) when I was first testing the LED projector - it wasn't as dark as you might think, but I suspect it will start to cause some serious light uniformity issues with higher power LEDs.

I've given much thought to wobulation type ideas in the past, but what they actually do is to show half the image's pixels first, then physically move the entire DMD chip by a tiny amount (half-a-pixel width, usually diagonally), then display the other half of the pixels. This can be achieved because DMD chips are so fast. It proves once again that DMD have always been fast, it's only things like mechanical colour wheels and white light sources which are the limiting factors....

http://scr3.golem.de/screenshots/0709/T ... dlp3d3.png" onclick="window.open(this.href);return false;

I'm not sure that you could do 21:9 because the chip only moves by an incredibly small amount. (I know you weren't entirely serious, but it's something to think about). 8-)

If you could rapidly move the entire image output of a DMD to a separate light path / angle or lens, maybe some interesting things could be done? Would need to be VERY fast though - you could use a wobbling or rotating mirror (like in a laser printer) to shift the image, but I doubt you could ever stop the inertia long enough to produce a flicker-free image.

The future of projection looks like it will use extremely fast but small MEMS chips to scan RGB lasers. TI has already started to produce basic MEMS chips and I think they will improve rapidly to the point where they become standard. Now that high-power Blue laser diodes have dropped in price HUGELY in the last few months (my "other" hobby :) ), it looks like laser projectors will be the way forward. (Excellent colour purity, and no focusing required!!).

I've been staring at Geohot's code a bit more - it looks far simpler than I imagined to control a DMD. I just wasn't sure at first what different parts of the code did. It can be simplified by a large amount.

You just send 64 bits of parallel data to the DMD at a time, so it takes 16 clock cycles to fill a whole 1024-pixel line for an XGA chip. Once a whole "block" is filled with data, you can then run the DAD1000 through a small reset sequence to actually move the mirrors to display the new data.

You can also choose which one of 16 "blocks" to send data too or reset. A "block" is a horizontal strip of the DMD, so there are 48 lines per block on an XGA DMD (49,152 pixels per block!). The tricky thing after that is storing real video frames to RAM, then timing the whole bitplane / load / reset / colour change process for the DMD. The RAM accesses will obviously need to be fast enough to support the video frame rate we want.

So, I'm just going to start with getting a PCB made for an XGA DMD (based on Geohot's design). I need to get hold of some DAD1000 chips though. Also, if you can easily find some better DMD chips (720p, DarkChip etc.), then we can think about trying to control those too at some point.

I think we need to start with XGA chips first though. They are FAR more common in faulty projectors than 720p chips, and have a high enough resolution not to completely suck (like SVGA - eewww).

I can't see the data rate being very high - I know there are many bitplanes which need to be loaded during a video frame period, but 1024x768 is only 96KBytes of data per bitplane...

1024*768 = 786,432 pixels or "bits".
786,432 bits / 8 = 96,054 Bytes.
96,054 Bytes / 1024 = 96 KBytes per bitplane.

The average reset time for many DMDs appears to be around 2us, but we don't need to load 96KB in 2us. This is probably way too fast to even see the image, so most projectors hold the LSB bitplane (Least Significant Bit) for a much longer time. This gives us time to load the 96KB of data before the mirrors are reset. Does that make sense? I'm not even sure myself any more. :lol:

@tritosine - do you think your friends can help with the actual routing of the PCB design? I can probably get a board routed, but the tracks will be quite randomly placed - luckily I don't think the data rate to the DMD is amazingly fast, but it would be a great help if they are experienced with impedance control?

Thanks,
OzOnE.
User avatar
tritosine5G
Terrif-eying the Ladies!
Posts: 894
Joined: Wed Mar 17, 2010 9:35 am
Location: As far from Hold Display guys as possible!!! ^2

Re: [DIY] DIY/modified DLP projector for page-flipping

Post by tritosine5G »

PhlatLight: how about these: http://s.taobao.com/search?q=Phlatlight ... rFineness=" onclick="window.open(this.href);return false;
DAD1000: im goin to ask here http://www.utsource.net" onclick="window.open(this.href);return false;
yeah I can try to get a look on your PCB. Add me to msn pls if you want ( brighter1@vipmail.hu - not an email , amsn is a good alternativ client).
I want to talk about this, in combination with silverscreen:
For projectors the indicated Contrast Ratio can generally be further increased by 100 percent or more by constricting the optical path with an iris, which reduces brightness and the overall optical efficiency (lumens per watt). LEDs can also behave like a super-fast dynamic iris, so they can dramatically improve Black Levels and Contrast Ratios (but with constraints that generally introduce some gray-scale artifacts, Part B).

What you can try right away is the decor paint, the standard metal flake spray stuff thats cheap to buy, must be some 5 gbp. Or you can ask for screen samples.

For example I met with a chinese who claim to have nanotube rear projection screen, that has 3 gain, wide viewing angle, and appears dark grey. This screen would worth a LED pj ;) , & that would be some 380 usd for 70", fer sure I'd sample first. Screens like VUTEC are hopelessly expensive though ( and depolarising!!).

http://www.picoprojector-info.com/inter ... er-and-cto" onclick="window.open(this.href);return false;
Don’t get me wrong, I think showing videos and cartoons is fine as a part of any pico projector demo and we show this type of content as well. But I think that pico projectors are going to be showing all kinds of content including videos, web browsing, text documents and presentations. I found the demo content being show for laser beam steering to be more than a bit contrived to avoid showing image quality issues with LBS.

There are a lot of other serious issues with LBS including light output limits due to eye safety, image uniformity, distortion, and alignment of the laser beams that will take more time to explain.
http://www.picoprojector-info.com/short ... diants-cto" onclick="window.open(this.href);return false;
At its core, the LaserVue TVs are DLP RPTVs. The lasers are just a fancy (and undeniably cool) light source. So the contrast ratio is going to be limited to the native contrast of the DMD. In theory the lasers can be turned up and down to increase or decrease the brightness (even turning them off to create an absolute black), but this is functionally the same as the auto-iris in other RPTVs. So the full-on/full-off contrast ratio will most likely be impressive. But the ANSI contrast (a checkerboard pattern of 8 white and 8 black boxes), which is a more accurate representation of what's on the screen, will no doubt be poor. This is because the laser is only able to ramp up and down the brightness of the whole image, not parts of the image (like say a local dimming LED LCD or what plasma and CRTs do inherently). In addition, all RPTVs have internal reflections that further lower the ANSI rating. Mitsubishi RPTVs typically measure very well compared to other companies as far as ANSI contrast goes, but we're still talking numbers in the hundreds, not the thousands of say a Pioneer KURO.
http://www.hemagazine.com/node/Mitsubis ... e_Laser_TV" onclick="window.open(this.href);return false;
Lol.

hmm . Didn't know about phlatlight is unobtanium as well.
-Biased for 0 Gen HMD's to hell and back must be one hundred percent hell bent bias!
User avatar
tritosine5G
Terrif-eying the Ladies!
Posts: 894
Joined: Wed Mar 17, 2010 9:35 am
Location: As far from Hold Display guys as possible!!! ^2

Re: [DIY] DIY/modified DLP projector for page-flipping

Post by tritosine5G »

Some useful links I found meanwhile:

http://cds.linear.com/docs/Reference%20 ... c1470A.pdf

I love Linear tech, they have true 18bit r2r dac in plastic package, quite a feat, Texas Instruments has no such thing afaik.

http://diy-community.de/showthread.php? ... 611faa203c" onclick="window.open(this.href);return false;
3x PT181 (4:3) , "paid samples", I can arrange this without VAT & customs imho.
Thank you I have paid for the LED's $ 200 plus shipping about 180 € including VAT on imports. Das ist der Sample-Preis. This is the sample price. Ich weis nicht, was die normal kosten, da man diese bis jetzt blos von Luminus direkt beziehen kann. I do not know what the normal cost because they can be up to now can only directly from Luminus.
-hmm!!!!! Hopefully we're not too late to the party :lol: , translation:
http://translate.google.com/translate?j ... 611faa203c" onclick="window.open(this.href);return false;

DLP_link glasses starting to get really cheap:
http://www.vidimensio.eu/catalog/images ... umfang.jpg" onclick="window.open(this.href);return false; <- this is said to be only 40Eur.
-Biased for 0 Gen HMD's to hell and back must be one hundred percent hell bent bias!
OzOnE2k10
Sharp Eyed Eagle!
Posts: 436
Joined: Wed Jan 13, 2010 7:12 am
Location: UK

Re: [DIY] DIY/modified DLP projector for page-flipping

Post by OzOnE2k10 »

Hmm, interesting. Did you manage to find the photos of "mmkeule's" projector build? They don't seem to appear for me, only the photos of the LEDs in the first post?

I wonder what type of optics he is using too? With the Phlatlights, I think the optics are actually a bit simpler than with other types of LED (the optics in the cine4home article looked relatively simple).

As it happens, I actually had some proper PCBs made for some LED drivers based on the LT3478. I did start off using the LT3475 (dual driver), but realized I might as well use single drivers because you'd almost always use only three LEDs (R/G/B).

The drivers did "work", but the problem was that the switching frequency ran at a different speed to the DMD update rate - this caused massive amounts of flickering in the projected image...

There is a "Sync" input on the LT3478 but I wasn't sure of what sort of clock to drive it with... Ideally, it would need to be timed to when the DMD bitplanes are updated, but again, without hacking the projector's firmware it would be fairly impossible to find a proper signal to drive the Sync pin with. If you just used the Vsync signal to drive the Sync pin, it would be way too slow to operate the driver chip properly.

Maybe it could be made to work with a specific clock frequency (and larger capacitor values)? In the end, I abandoned the PWM drivers and built my own analog versions using KA350's (3 Amp versions of the LM317). I only needed them to switch full ON / full OFF, and I can still drive them using PWM from the FPGA for controlling the brightness of each LED anyway (high clock speed, so no flicker at all).

The analog drivers are very simple to build, but the other issue with Phlatlights is that they operate at a low voltage and require massive amounts of current. The use of Phlatlights is unfortunately going to add complexity to the type of high-current PSU needed etc. (I only require a simple laptop PSU to run my current build.)

I'm very surprised that "mmkeule" got hold of an Osram Rapcur driver; it is very rare to see them for sale anywhere!

@Tritosine, you must be good with finding stuff on Google. :D I thought I was a determined searcher, but that German thread is an interesting find.

On a related note, I'm currently finishing off my laser projector build. It uses some new Flexmod P3 drivers which could also be used for many types of LEDs. They are well built and cheap linear / analog drivers with bias and gain - up to 4 Amps per driver, pretty good, but not enough if you're using Phlatlights :(.

It would be nice to use laser diodes for DLP, but the number of diodes needed is extreme...

http://www.cine4home.com/tests/3-tests/ ... f-uhp.html" onclick="window.open(this.href);return false;

You can see that the Casio uses 24 one-Watt blue diodes! (this is where my diode came from!). It uses a phosphor wheel for Green, and a Red Phlatlight too.

OzOnE.
P.S. Attached is a screenshot of the LT3478 driver layout (for what it's worth). I would have posted a photo of an actual driver, but I can't find them atm.
You do not have the required permissions to view the files attached to this post.
User avatar
tritosine5G
Terrif-eying the Ladies!
Posts: 894
Joined: Wed Mar 17, 2010 9:35 am
Location: As far from Hold Display guys as possible!!! ^2

Re: [DIY] DIY/modified DLP projector for page-flipping

Post by tritosine5G »

Hey, I can get SMPS psu in any shape & size. My friend makes them and I bet he's eager to help ( skilled in both smps & class-d , then there's Pafi on diyaudio forums , the other switching amp maniac, he's a friend). So I vote for PT120 / 121, bcuz:
On the 4th page of the german forum they talk about 1000 lumen with PT120. Whew. I calculated something like:

4-500 is needed for shutterglasses 3d with 10.0 gain screen, but that has to be curved. If 700-100 lumens can be reached, 4.0 gain flatscreen paint is shiny too. 8-)

In fact my 120hz DLP does 800 lumen right now with its bulb. With matte screen its dark, but playable well. On/off contrast sucks tho, and I dislike mattewhite because its all about reflections in the room, sorta ambient lightning machine.

EDIT: Breaking news on the screen front. Someone told me an unbelievable idea , this will be a mega-suprise . It's flat, it's 3+ gain, its retroreflective, plus dirt cheap. I'll tell more once I tried it. Maybe you can bug me on MSN :mrgreen: :mrgreen: :mrgreen: :mrgreen:
-Biased for 0 Gen HMD's to hell and back must be one hundred percent hell bent bias!
User avatar
tritosine5G
Terrif-eying the Ladies!
Posts: 894
Joined: Wed Mar 17, 2010 9:35 am
Location: As far from Hold Display guys as possible!!! ^2

Re: [DIY] DIY/modified DLP projector for page-flipping

Post by tritosine5G »

Look at that: 1000 lumen with PT120 (playable with matte white screen very well but not very bright)
http://www.entexpinc.com/truvue/pdf/TruVueTechSheet.pdf" onclick="window.open(this.href);return false; . They are simulating the white segment I guess?
So far we can achieve:

1. huge contrast advantage
2. "zero bulb" advantage
3. brightness advantage up to 2-30% compared to bulb 120hz, and 4-50% compared to "aging bulb" (!!!)
4. my latest theory: motion resolution advantage

-my comments in bold:
According to that test material, the 240Hz Sony XBR7 (4ms hold time method,with black frame insertion, bad for flicker sensitive ppl) did in fact reduce motion blur significantly compared with 120Hz displays , so I'm willing to believe claims that 240Hz is less-blurry than 120Hz. In case you're wondering, the XBR7 delivered between 900 and 1,000 lines of motion resolution, which matches the result of a typical plasma. ( damn sure plasma have 4 ms hold time,regardless source material, "governed by pwm" , http://www.avsforum.com/avs-vb/showthre ... st10641564)

That compares with 500-600 lines for a standard 120Hz model like the Sony KDL-52XBR6 and 300-400 lines for a 60Hz LCD ( FullHD huh? bahahaha). But those results were with test patterns. The real question is whether you actually tell the difference in everyday viewing? For most viewers, the answer is "no."(so what)
-according to this, right now, with my 120hz dlp and ~8 ms hold time, I have 5-600 :/
But I want my full 720 lines. ;) Its clear what we have to do. (120 +hz )
______________________________________________________
ok im done with my part of researching, next up: devil's nest, dive into TI 's patents.
-Biased for 0 Gen HMD's to hell and back must be one hundred percent hell bent bias!
User avatar
tritosine5G
Terrif-eying the Ladies!
Posts: 894
Joined: Wed Mar 17, 2010 9:35 am
Location: As far from Hold Display guys as possible!!! ^2

Re: [DIY] DIY/modified DLP projector for page-flipping

Post by tritosine5G »

Found optics stuff:
http://glighting.en.alibaba.com/product ... ptics.html" onclick="window.open(this.href);return false;

http://www.crystechcoating.com/optics/& ... ee-sp.html" onclick="window.open(this.href);return false;

I want to go for those pt120's, and ask linear tech for the driver PCB gerber's. Also would talk to chinese. Ozone what do You say.
-Biased for 0 Gen HMD's to hell and back must be one hundred percent hell bent bias!
User avatar
tritosine5G
Terrif-eying the Ladies!
Posts: 894
Joined: Wed Mar 17, 2010 9:35 am
Location: As far from Hold Display guys as possible!!! ^2

Re: [DIY] DIY/modified DLP projector for page-flipping

Post by tritosine5G »

sorta Offtopic, https://www.ideals.illinois.edu/bitstre ... sequence=2" onclick="window.open(this.href);return false;

or not ? :woot
-Biased for 0 Gen HMD's to hell and back must be one hundred percent hell bent bias!
cyberheater
One Eyed Hopeful
Posts: 7
Joined: Thu Jan 21, 2010 10:28 am

Re: [DIY] DIY/modified DLP projector for page-flipping

Post by cyberheater »

tritosine wrote:EDIT: Breaking news on the screen front. Someone told me an unbelievable idea , this will be a mega-suprise . It's flat, it's 3+ gain, its retroreflective, plus dirt cheap. I'll tell more once I tried it. Maybe you can bug me on MSN :mrgreen: :mrgreen: :mrgreen: :mrgreen:
Tell me more...
User avatar
tritosine5G
Terrif-eying the Ladies!
Posts: 894
Joined: Wed Mar 17, 2010 9:35 am
Location: As far from Hold Display guys as possible!!! ^2

Re: [DIY] DIY/modified DLP projector for page-flipping

Post by tritosine5G »

cyberheater wrote:
tritosine wrote:EDIT: Breaking news on the screen front. Someone told me an unbelievable idea , this will be a mega-suprise . It's flat, it's 3+ gain, its retroreflective, plus dirt cheap. I'll tell more once I tried it. Maybe you can bug me on MSN :mrgreen: :mrgreen: :mrgreen: :mrgreen:
Tell me more...
Where are you at?
Thats the "autobahn" stuff ;) . Found cheap silver screen since, too. Said to be ~1.8 gain, looks like PVC you can tension, with silver on one side. Asked if they have it on stock then I buy it as fast as I can. This is much easier for me than glass beaded , so Im going for silver first, also I can just tension it over my current frame, huge plus, angular reflective, also a huge plus right now. I don know much yet , but I will tell you guys , I promise.

Autobahn screen info , thanks to Petrus for finding these french websites, he was ahead of me with all this actually, I simply got this idea from a repairman.
Hi, thought you might be interested about my cheapass glass beaded screen research (seems much better than diy silver so far), some private msg-s of mine:


Hi, french ppl did it, they say grain is not visible beyond 1 meter, and its a fine 2d screen too . Grain is because of uneven distribution. Literature says da lite hi power is good because of 9 micrometer beads, and commercial not good because of 60 uM. I think you can obtain an inbetween value and sort those somehow. French ppl also used a light grey paint behind beads to gain amb. light rejection.

Here are their pictures:

http://img154.imageshack.us/img154/5103/1001465mw9.jpg" onclick="window.open(this.href);return false;
http://img92.imageshack.us/img92/4481/1001456px9.jpg" onclick="window.open(this.href);return false;

Im goin to obtain some 10lb 's of bog standard autobahn stuff next week and I'll try.
Re: DIY Glass Beaded Screen
http://www.allinbox.com/technique/neops ... psycho.htm" onclick="window.open(this.href);return false;

thats quite good result for first screen of his. Lets not talk about my curved silverscreen attempt, hahah.
more stuff: http://www.diyaudio.com/forums/diy-proj ... use-3.html" onclick="window.open(this.href);return false;
http://en.wikipedia.org/wiki/Tulle_netting" onclick="window.open(this.href);return false;

this looks cool to sort the beads , no ?
-Biased for 0 Gen HMD's to hell and back must be one hundred percent hell bent bias!
User avatar
tritosine5G
Terrif-eying the Ladies!
Posts: 894
Joined: Wed Mar 17, 2010 9:35 am
Location: As far from Hold Display guys as possible!!! ^2

Re: [DIY] DIY/modified DLP projector for page-flipping

Post by tritosine5G »

Wobulation type DMD ( 960*1080) frequently breaks down , due inadequate cooling, seaside weather, many spareparts float .

http://www.audiovideo-parts.com/scripts ... 0Favorites" onclick="window.open(this.href);return false;

http://losangeles.craigslist.org/lgb/el ... 17149.html" onclick="window.open(this.href);return false;

I know fullHD sounds overkill, but, you could use some anamorhpic lens arrangement for 2.35 : 1 ;)

Ozone , all ok ?
-Biased for 0 Gen HMD's to hell and back must be one hundred percent hell bent bias!
User avatar
tritosine5G
Terrif-eying the Ladies!
Posts: 894
Joined: Wed Mar 17, 2010 9:35 am
Location: As far from Hold Display guys as possible!!! ^2

Re: [DIY] DIY/modified DLP projector for page-flipping

Post by tritosine5G »

Fixed myself 2 sets of PT120 so far, someone buys them for me in US in exchange for something. I won't loose money anyway now. I might be fixing a few sets moar. For now we'll manufacture those linear tech switching power supply pcb's, and I'll use LED's as lightning . ;)

Dudes, we need to reverse engineer that xHd5 chipset ...

Im reading about diy anamorphic lens, seems perfectly fine choice for xHD5.
-Biased for 0 Gen HMD's to hell and back must be one hundred percent hell bent bias!
Post Reply

Return to “Do it Yourself!”