Electronics DIY LVDS controller - all soldered, no video?

i know this has been posted about 53256254 times before but i haven't seen anyone ask my specific question.

of course, i have here an lcd from a laptop. compaq armada 7400, to be (somewhat) precice. it's a 13.3 inch hitachi tx34d65vc1caa and the only webpage i found about it other than someone trying to sell one ($300US ?!?) the only post i found about it was on a dutch modding site. the person was pointing to hitachi's pdfs but saying that his lcd wasn't listed. about 3 posts down someone posted info on a samsung lcd. well it turns out that one of the sites trying to sell the hitachi lcd that i have says the samsung listed in the dutch thread was compatable with the one i have.

so i have a rough pinout of the lcd. it uses lvds. big suprise there, so the method discussed here: http://forums.bit-tech.net/showthread.php?t=51818
won't work. (incidentally that dutch site i found googling is the same one linked in there)

i didn't want to be defeated. i'm not gonna shell out $100+ for a controller card. so i started plugging in random chip numbers from the motherboard on the laptop and it turns out that one of them scores a near direct hit on the TI website. the hit i got was a newer generation chip than the one i have, but further searching within the TI website gave me this http://focus.ti.com/lit/ds/symlink/sn75lvds84.pdf

so now i have a pinout for the controller chip as well as for my screen.

i'm sure it's probably not worth it in the long run to try to take the (ohmygoshtiny) smt 45 some odd pin chip off the motherboard and wire it myself, but assuming i was able to and got it hooked up to the lcd correctly, and had the video inputs ready to accept signal, where would i go from there?

page 11 of the above pdf shows a rough schematic of how to hook up the controller chip, but it has 3 or 4 inputs for each color. IIRC vga out only has one output for each color, so i know i'm not going to go off the back of the video card directly. but would i be able to go off the vesa header on my voodoo3 as in the above thread here on bit-tech?

ultimately i'd rather not spend any money on this, other than maybe a pcb etching kit and a finer tip for my soldering iron. but with the information i've found i'm starting to think it might be possible to make my own controller with a normal video card and have myself a nice (even if only 16 color ) little display for what will eventually be my media pc (mythtv emerging now in gentoo on it).

anyone care to point me in the general direction i need to go?

 

Well I've had a look at the PDF and to me it seems like the chip takes an 18bit digital signal in. This could be tricky to find but I expect that somewhere on your graphics card there will be points where you can get the raw signal. This will likely be out of the GPU before it enters a chip to convert it to a signal fit for a monitor.

This really won't be easy but should be very interesting indeed if you get something working.

 

hrmmm.... well i have an AGP ATI rageIIc 4mb that i'd be perfectly willing to sacrifice, and should i get something working on that i may see if it'll port over to the voodoo3 3500TV that will be the basis of the 'media' part of this computer... off to find ati schematics.

it'd be so much nicer if i could just cut out the controller, video chip, and power stuff on the laptop and use all that, but i'm sure that wouldn't work.

thanks for taking a look at the pdf, and for the encouragement

 

Ive also had a quick look at the pdf. Im no expert on this but it looks like the VESA method may work. The controller inputs include the H&V syncs, clock and enable mentioned in the VESA thread. You would however have somne extra data inputs, as you'd only use 4, but you could try tying them low. May be worth a try if you have the time/motivation I know im going to have a go over the summer when ive got some free time

 

i just might try that, too, but as said above, i think it's expecting several raw digital inputs.

on another note, i did the google random chip numbers trick with my voodoo3 and found (among other things, like the video decoder and audio decode for the video inputs) on the back of the card an xilinx square IC and the pdf for it here: http://www.engin.brown.edu/courses/En163/xc9572XL.pdf it says nothing about video input/output but it looks like it's a generic encoder/decoder circuit. would this be the chip used to translate the digital video from the voodoo3 chip to vga? it's yet another ohmygoshtiny SMT chip and again, who knows if i can come close to soldering that small, but if it looks like a good starting point it's something, right?

i haven't looked at the ATI card cause it's still in the computer which is currently compiling qt. so it should be done sometime later this week .

thanks again, i'm hoping i can really pull this hack off, and if not that my info might help someone else get a laptop lcd working from a normal video card.

 

just re-read this reply, what do you mean by "tying them low" ?

 

tying them low = just connecting them to ground

 

hrmm, well on the spec sheet it says if you're not using all 4 of each R/G/B connector (where it shows the difference between 12 bit and 18 bit connections) it just shows that the extra inputs should be left not connected. i'd think that tying them to ground might mess something else up in the chip.

but i'll try it both ways...

 

Yes indeed it does I only had a quick peek at it! Unused inputs tend to attract noise but I think thats more with CMOS chips which this doesnt seem to be. Like you say cant hurt to try it both ways (im sure theres a joke in there somewhere )

 

remeber to tie all pins with cmos.

(10k should do).

Looks as thought you might be able to use the VESA method because if i remeber rightly thats 12bit, also your controller chip is 5v tolerant (even thou it likes 3.3v supply), so bonza, might be worth sticking a 1k resistor in line with the outputs of the gfx card even so, just to be safe.

 

do you know the actual high/low output voltages on the VESA connector?

egads i feel like a n00b all over again. "tie all pins with cmos" ?!?

 

I would imagine the VESA bus uses 0 and 5v outputs, as theanimus says the chip will take it.

Basically make sure that every input pin on a CMOS chip is connected to either the high or low rail (ie 0 or 5v) If not they act as 'aerials' picking up all sorts of background noise, floating between high and low and generally causing problems TTL chips float high internally so you can get away with it, but i was always taught it was good practise to make sure an input pin is connected to something, no matter what the chip is

Be very interested to see how you get on with this, Ive been itching to try this ever since i saw the VESA thread, as soon as the exams are over ill be scouring ebay

 

ok, cool, thanks that makes sense. (one of my more recent electronics experiments was a small amplifier kit that failed to amplify microphone input but picked up local AM stations very faintly)

as for VESA outputs, i'll have to sit down and start probing with my multimeter to see what kinda outputs it's really putting out and possibly add resistors from there.

thanks for all the help guys!

 

awesome, it looks like it's 3.3v signaling on the VESA outputs.

 

well i have the preliminary schematic and circuit diagram for the controller
[edit] { images deleted, newer ones below } [/edit]
(made in EAGLE by the way, i started with the windows version then copied what i had into the linux version, both work quite well once you get past its quirks)

the IC is the controller chip, the pads are for the wires from the video card (nice how a ribbon cable motherboard header for a printer fits over the VESA pins, only had to brake one of them off, which is either ground or not used depending on the pinout you believe)
and the 50 pin connector is for the LCD, though in real life it's much smaller and SMT so i'll have to find an smt one in a library (anyone wanna make one for me? i haven't gotten that far yet).

I've been poking around on the connector to the lcd and i know what goes between the chip and the lcd, and i know which pins are ground, now it's just a matter of finding which pins are power then i should be all set

the traces on the board are ugly because i got bored of waiting for the autorouter to try to figure out how to route the circuit on one layer so i did it myself. tomorrow i'm gonna try figuring out where power goes into the lcd, i'm pretty sure it's all 3.3v and i know where the IC gets it's 3.3v from so i should be set there.

who knows when i'll be able to get a pcb kit and actually start making this, fortunately it's small so i'll be able to make multiples out of one kit

 

ok, did some more probing today, and looked at a few more things on the motherboard, and desoldered the chip to see the traces under it, etc.

I also built my own 50 pin smt cable header because the library didn't have any, and that changed the board layout quite a bit.

the board at what will be realsize, about 2 inches by an inch and 3/4, i.e really frickin small. ignore what look like touching traces, the real big printout doesn't have them. (click for a slightly bigger monochrome version, still has a few traces that look like they're touching but i'll be printing at higher resolution)



the big blob on the top left is ground, thick trace on the bottom left is power for the ccfl (still haven't figured what voltage), heavy trace on the right is power for the lcd, above that is a big pad for power for the chip.

i managed to get the whole thing layed out with only one jumper wire needing to be put in place, towards the right of the chip (stuff that has to be grounded, i'll just solder a wire over the middle power trace)

here's the schematic (again, click for fullsize version)


i get payed wednesday so hopefully i'll be able to get a pcb kit then. at the size this is and the size of the sheets in the kits at radio shack, i shuould be able to make several if i mess up the first one.

 

this now points to a (huge physical size, 40k disk size) 1400dpi image of what my circuit board will look like, which prints flawlessly on my epson c82 at 1440x1440dpi. i've had to tweak a few things in the gimp to get the pinouts for the LCD header right, and it's now accurate


now for the questions... on the motherboard between the chip lvds outputs and the header for the lcd itself there are 5 ohm resistors on each of the 8 signal traces. i know 5 ohms isn't much in general day to day stuff, but is it going to make a huge difference when dealing with something like LVDS?

second question, i know the chip takes 3.3v power from it's spec sheet. based on that information, i was able to figure out that the lcd power is also 3.3v (0 resistance between pins on the lcd for power and pins on the chip for power) -- but i can't for the life of me figure out the voltage for the ccfl. from what I can tell, one pin is definitely ground, and one is a thick trace off to what looks like a mosfet (small tubular thing with a center metal ring tied to ground, parallel to the one that the 3.3v for the lcd/chip trace back to) so i know what those are. on the pcb for the ccfl inverter, the pins corespond to what look like power and ground.

the 2 other pins are odd though... one has about 2k ohms of resistance between it and ground on the motherboard, and on the inverter board looks to connect to a transistor that has the main power running through it, so i'm thinking contrast control there? the other one is just as if not odder than that, i can't tell where it connects on the inverter board but on the motherboard it appears to have a 104k ohm resistor between it and ground. i have this resistor placed in the schematic and a space for it on the board, though i'm not sure what the point is.


anyway, can anyone shed some light on these questions? thanks in advance

 

coughcoughbumpcpugh

anyone have a guess on wether i need the 5ohm resistors and/or what voltage the ccfl might be? I'll probably be buying the etching kit tomorrow after i get payed, and while i'll have plenty of real estate to work with (enough copper clad board to make at least 8 of my board up there) i don't want to have to solder/desolder a bunch of stuff if the first try doesn't work.

acrimonius? zap? you guys are the resident electroncs geniuses

edited to add that the 5 ohom resistors between lvds output and input are not listed on any of the schematics or datasheets from TI. they may have been something added by compaq to cope with something else, but i have no idea.

 

bah, i'll just make the circuit and futz around with it, one of my friends suggested that the ccfl inverter might be running at battery voltage, so i'll have to check what that is.

i'll be buying the pcb kit today and probably printing off some copies of the circuit on transparency sheet i happen to have, I just have to remember to bring the transparency in to work tomorrow.

drat, i forgot i don't have an iron. eh, maybe that'd be a good thing to buy anyway.

anywho, wish me luck, i'm sure i'm going to need it.

 

doh, won't be getting payed today, the paychecks got stuck in another state for some reason. drat.

guess it's tomrrow then