Screen Tablet malarky => Build Logs => Topic started by: DaBotz on June 22, 2017, 02:56:08 PM

Title: Unneeded-IQ
Post by: DaBotz on June 22, 2017, 02:56:08 PM
A little time ago, I ceded to a temptation - an Intuos 2 12x12 USB, for 1euro (+10 of shipping fees).

The guy selling it wasn't sure it worked, it lacked a button (not really... just that odd cover that is in place of a Reset button used in some iteration of the serial connection tablets) and had no pen, but the light went up when attached to the usb, and the computer "saw" it.

Being educated in the ways of Bongofishism, I gambled on it having nothing more than marks and scruffs.

For a while, it has sat around in my drawing place, waiting for the future.

Then, a month or so, I was struck by yet another attack of Cintiquism, and went on ebay for a cheap 15" with external power brick - possibly with DVI connection.

The few I saw asked too much money, for my taste (this would be my sixth build, after all), but I finally saw an IBM 9493-ag1 for 25 Euros, shipping included.


I checked its service manual, found on-line ( ( ) , and saw that it had an internal (hidden below a cover) DVI input socket, that went to the outside through a DVI to D-Sub (VGA) cable.

It also had replaceable CCFL lamps (which were likely in need to be replaced, as this is a business monitor and must have logged in 8 hours a day since the day it was bought).

Alas, I didn't peruse the manual long enough, or I would have seen that the LCD is a LG LM151X2-F2MN,  attached to the monitor module through a Hirose DF9-41p connector, and a whole twelve cm or so  (say, 5" in imperial units) of cable.

I so hoped it would be a 30 lines LVDS on a 1mm pitch FFC (like in the three LG monitors that I have opened).   


That is going to bite my ass...

Also, it is a 6 bit LCD (whoops - this wasn't going to be a "main battle Cintiq", anyway, just another "fast selection" screen, if it is not too messy, jitterwise - also, I hope that IBM did a good job with dithering, in the monitor firmware).

It arrived in a couple of days, and I immediately realized that the CCFL were - if not almost dead - very well ripe.

It took a minute or so to reach full brightness, on switch-on, and it wasn't much bright, anyway

I decided to invest some more 9 $ to buy one of those "Upgrade to LED" kits, before opening up the screen.

I am sorry that I did not take pictures of the disassembly of the monitor (extremely well put together... it has more screws and RF cage than those present in all the monitors I dismounted to date), but it is actually detailed in the service manual (so, it would be a a bit redundant).

When I reached finally inside the monitor, I saw the damn "Hirose on a short leash" connection, then saw that the back metal plate of the LCD was simply attached to a white plastic frame, and could be removed without opening the LCD panel (Good).

The CCFL are in pretty sturdy assemblies with a steel beam-channel containing them, and slide in-out of the panel pretty easily - so I could replace them with the LED bars, in a matter of an hour or so, ONCE I understood which cables are what.

In fact, instead of the 4 cables I expected, the inverter assembly - a Samsung LG1501 - was connected to the monitor's module with a ten pin connector, and ten individual lines (note: even if on my monitor, a permanent marker line marked the monitor module side, it doesn't matter if one's inverts the head or tail of the ten-lines, as they are flipped in the travel from one side to another)

Initially, I was perplexed, however a couple of continuity tests at monitor unplugged from the power grid and some final corroboration with it switched on and off, gave this pretty simple pin-out

     +5V       What the hell are these for?

     0V - GND
     0V - GND
     +4.9 V (on) / 0V (off )  -  enable , or enable+d¡m? I failed to see any difference in the value, though, when I dimmed down the brightness

     -0.1  V (?????)

     0V - GND
     0V - GND

Notable miss: which is the damn dim line? No idea.

I removed the LCD backplate; probably, sanding/filing down the plastic sprockets that hold it in place could lower its height 1-1.5 mm.

However, I saw something that just a couple of years ago would have been a snafu, for me - it has two PCBs, one folded back for the column driver, a more components heavy one for the row driver, connected by a 20 lines, 0.5mm pitch flexible PCB.


I bought and extender, from an Italian seller, and some 15cm (note: it is a bit tight; it would be better some ten cm, or so, more) of the correspondent FFC.

Once I got the extender and the FFC, I folded out the main PCB, and extended the lines of one CCFL, to see if this mod works.

It does.



(Note that, at full brightness, Photoshop retouch wasn't needed to see around the screen)

Yesterday I received the LED kit, and I proceeded to replace the CCFL.

(Retouched with Photoshop; even this way, the "aura" of the screen, with the LED, is well visible.)

It works as expected, at least twice as bright as before, but at the moment I have no dimming whatsoever.

Someone has any idea / suggestion?  

Title: Re: Unneeded-IQ
Post by: Ertew on June 22, 2017, 06:30:00 PM
DVI input:
I checked schematic and cannot found digital input. Looks like main electronic board have only VGA input. I'm not sure whet to say about DVI->VGA converter but I'm sure that it doesn't exist on schematic.

Inverter voltages:
Most CCFL drivers works from 3.3V to 12V. If higher supply voltage is essential, controller may be powered from "enable" line or from main input through resistor and zener (simple voltage stabilizer). First solution is common in laptop designs. In Your case I'm sure that 24V goes only to CCFL transformer and power transistors. 5V goes to driver chip.

Inverter pinout:
look at page 40 (of 41). Inverter need single 5V line. Second one are PWM  8)

Title: Re: Unneeded-IQ
Post by: DaBotz on June 22, 2017, 11:50:44 PM
Thanks, Ertew... bummer for the analog-only input.  Why on Earth did they decide to have a DVI connector there, instead of a classic D-Sub, escapes me.

I will check page 40, thanks.

... oh, it i the line with the blue cable (the only one different, all the others are grey)

I feel more than a bit dumb.

Being able to dim the brightness will be very nice, as at full power it is almost too bright and heats up quite a bit.

Title: Re: Unneeded-IQ
Post by: DaBotz on June 23, 2017, 07:17:27 PM
Checked the line...

Output is

 0.1 V   with brightness at 100%

 5 V with brighness at 0%

I fear that it is not a PWM , but a direct analog output, as the LED board doesn't seem to respond much to the brightness control.

(but it heats up less the LED, so, maybe it's just that my etes do not see the difference)

If it is an analog, I suppose I could wip up an analog-PWM converter with a capacitor, a resistor, a diode and the usual double op-ampli chip... I should really need some kind of signal analyzer.   

Title: Re: Unneeded-IQ
Post by: Ertew on June 23, 2017, 09:30:44 PM
I looked at monitor documentation once again. Schematics are for main PCB. This board have analog RGB input only.
There are also second PCB called power board and contain input connectors. If DVI->VGA converter exist, there must be chip for that job. If You cannot found converter chip (min. 20 pins) on power board, this monitor have only DVI-A input.

I'm 100% sure that dimming are realized by PWM output. No one use DAC for analog output. There are only two options: PWM or analog via PWM + resistor + capacitor.
Look at PDF, page 38, top left corner. There are a lot of signals. Somewhere in the middle You can found DIM-ADJ (7) signal. It's connected to two resistors and capacitor.
IMHO You should use DIY analog->PWM converter. Yes, You can get PWM signal by removing capacitor or by connecting directly to chip (left side od R61 ? ). But there are second problem, PWM frequency. LED driver have limited frequency capacity (often above 100hz, rarely above 1khz). On the other hand monitor chip can easily produce PWM signal with range 4khz - 45khz, depend on selected options.

Title: Re: Unneeded-IQ
Post by: DaBotz on June 24, 2017, 09:46:32 AM
Thanks again, Ertew. I see no converter on the power board, though I still have to un-mount it so it could well be on the other side(for the moment, I have no access to my usual working table, so I will have to shelf this project for a little while)...

More than a bit of a Bummer, for me.


Yes, I already supposed that I had to build some Analog to PWM converter.

I found a schema, on-line, that should do the trick and, with a small adaptation, being able to provide a tunable frequency


(just using a variable resistor for R4 should allow a certain degree of tune, but I'll look it better when I have a bit of time to install a SPICE and play... if I remember my electrics well, that RC time constant is way too low, off by a couple of magnitudes).

Title: Re: Unneeded-IQ
Post by: DaBotz on June 27, 2017, 01:04:07 PM
In the end, I think that I will replace that capacitor with one of 100nF and the resistance with one 2.5 fixed and a variable (an ungrounded potenziometer) of 0-10k one.

it should give a PWM with a frequency tunable between some 200 hz and a low khz...


I changed the .asc extension to .txt ( it is a text file, after all) - I used LTSpice, from Linear Techcnologies... hence, the choice of the OP-Ampli (I am going to use an LM358 or something the like, in the end).

Title: Re: Unneeded-IQ
Post by: Ertew on June 27, 2017, 07:06:59 PM
Fingers crossed :)

btw, LM358 have very poor performance. Oscillator with this chip will always run slower than with modern op-amp. You may need to tweak values before get it working.

Title: Re: Unneeded-IQ
Post by: DaBotz on June 27, 2017, 08:47:38 PM
Thanks for the tit-bit about the LM358, Ertew (though, I thought that a 1Khz max output frequency was pretty basic...).

I can't find that at the local shop, nor the AD8359 that I used with Spice... I'll have to decide for something else anyway, it seems.

On a side note, I have bought a couple of DF9-41-V(the plugs to the DF9-41-P sockets)  that will arrive in a week or so... when they'll arrive, I'll have to try to whip up a longer connection cable for the lcd.

that is going to be frustrating. or fun.

Or frustrating.

I admit that, as it is a bit on the supernumerary side of things, I am taking a lot more risks than usual, on this build.

Title: Re: Unneeded-IQ
Post by: Ertew on June 28, 2017, 07:29:13 PM
Oops. I forgot about frequency. I build similar oscillator and notice significant frequency error at 10khz. At 100khz circuit are unstable and way off from simulation. For 1kHz You may expect less than 10% error at frequency. Sorry for scare You.

More important thing: LM358 aren't rail-to-rail amplifier. I expect non-linear distortions near minimum and maximum settings - may be noticeable but not critical.

Title: Re: Unneeded-IQ
Post by: DaBotz on June 29, 2017, 06:30:13 PM
My idea was having a PWM frequency tunable between some 200 and 1000-1200 hz... pretty basic, as far as electronics go.

Fast enough to avoid seeing the light flicker (would it be best some exact multiple of the refresh rate? no idea), slow enough that the on/off times of the LED can be ignored hen setting brightness at minimum , and with a chance of tuning it out of resonances with the tablet, should these manifest themselves.

From what you told me before, the LED kit probably can't handle things going much faster than that, anyway... and my soldering ability is not stellar, so slower is going to be better.

Title: Re: Unneeded-IQ
Post by: DaBotz on July 07, 2017, 10:14:52 PM
Waiting for the OPAmps (and my fortitude to re-group), I tried to solder an extension for the Hirose DF9 etc... No way in hell I can manage 82 solders on pins with 1 mm pitch.

I am stuck with the original, ridiculously short cable.

As a result, the relative position of LCD vs The board is all but fixed... two cm to wiggle around, or so. Also, the connector is ridiculously prone to unlatch on his own (also, the cable is as rigid as I would like my nether parts to be, which definitely does not help).

Never ever again will I consider using a panel with it...

I have decided to place the screen "vertical", to be a the middle of the finished build, which has led me to some unfortunate discoveries...

This allows me to place the monitor main board, the power and video inputs board  and the Intuos secondary board in the same underside recess, and keep the front relatively
thin (relatively is the main word, here... if I guess correctly, and the XD-0912 is similar to the XD-0608, a build with one of them and a modern LAptop LCD can be as thichk as 1.5 cm; this one at minimum is going to be 32 mm.)

My 12"x18" build is 61x43 cm (it could be trimmed town to 59.5x40.5, simply avoiding the second order of side walls that I added to have the width for the side screw threads.

If I decided to go on, this 15" (12"x9") was going to be 43 cm tall, for... 48 cm wide. Much bigger than I expected or desired. 

Also, having it open and all, I was finally able to do some tests and, in the end...

Both the column rows driver boards interfere with the tablet, and even with these out of the way, the build closes on par with the old "UBiq" center area, at which point... I may be better served building a smaller case for that, wide as the screen only... it would have a similar final footprint, a bigger drawable Area (about 1100x1050 px, vs 1024x768, or 16" vs 15").

A video of the test (

A bummer, but I learnt how to replace CCFL with LEDs, at least. 


Title: Re: Unneeded-IQ
Post by: Ertew on July 08, 2017, 07:09:56 PM
If I see correctly, DVI connector miss many pins. This must be DVI-A input plus some funny pins for PnP, monitor ID and hot-plug detection. Definitely no digital input.

Nice work with backlight. If I see correctly, this screen are designed to easily swap backlight elements.

P.S. linked video doesn't work.

Title: Re: Unneeded-IQ
Post by: XDjackieXD on July 09, 2017, 12:49:17 AM
If I see correctly, DVI connector miss many pins. This must be DVI-A input plus some funny pins for PnP, monitor ID and hot-plug detection. Definitely no digital input.

Nice work with backlight. If I see correctly, this screen are designed to easily swap backlight elements.

P.S. linked video doesn't work.

one half of the pins is only needed for dual link dvi so it is possible that these pins are enough for single link dvi (but I haven't looked up the pinout so it could very well be that the connected pins are only for EDID)

Title: Re: Unneeded-IQ
Post by: DaBotz on July 11, 2017, 01:09:54 AM
If I see correctly, DVI connector miss many pins. This must be DVI-A input plus some funny pins for PnP, monitor ID and hot-plug detection. Definitely no digital input.

Nice work with backlight. If I see correctly, this screen are designed to easily swap backlight elements.

P.S. linked video doesn't work.

I tried the link, and it works for me.

The back-light replacement is very easy, in this LCD - one unscrews the CCFL assembly, and they require a very modicum force to slid out of the Panel- no need to open all up, and expose the various layers inside the lightbox or the LCD.

The back metal plate, too, was attached to the internal plastic frame and could be removed without opening the panel.

Also, the assemblies holding the CCFL are pretty sturdy  - to take out the lamps and glue the led strips in their place ( with a non-conductive, transparent "mass" glue resembling a sticky silicone that I use often as "DIY" rubber) was maybe half an hour of work.

At full power, the LED driver had an inductor that heated up quite a bit - I glued a 4x4 cm piece of aluminium (a radiating surface from a scrapped PSU) on top of it, and it cools it down nicely.

The main issue with this screen was the Hirose connector: it is beyond my ability to jury-rig a longer extension for that, and the cable inside the monitor is just too short to place the main board in any sensible position.

Combined with the fact that both the comlumns driver and the rows driver boards interfere with the Intuos, requiring two RF cages, it made the project simply too much hassle, for little gain.

However, I mounted the monitor back together (if I can, I always try to keep things reversible till the final build), and I  am happy to say that, for the moment, it works flawlessly - the LEDs are a bit too spaced, and it shows on blacks in the cm or so above the LED strips, but I think it is acceptable.