Screen Tablet malarky => Build Logs => Topic started by: Forestfrog on February 25, 2017, 01:48:02 PM

Title: Ribbitiq - trying not to break everything.
Post by: Forestfrog on February 25, 2017, 01:48:02 PM
Hey there,

getting in line of the "giving making a DIY-Cintiq a try"-ers.

A little background: Actually I got a drawing monitor some time ago, the UGee 2150. Maybe someone is shaking their head, going "why would that idiot want a DIY-Cintiq then??" Well, it is a mixed bag.

The screen/hardware itself is very good, as far as I can judge, screen is very crisp, great colors, little heat, pen pressure works ok, little lag, but  - and here I am looking at the other half - the driver is quite special.
Not many options - that I could live with. Calibration is...tricky. I can't get it right, for example I started to fudge on the height after a lot of tries, clicking the pen about five mm below the calibration marks, so the cursor would line up a bit better with the pen tip, as usually it would hover 5mm above it after a finished calibration.
Then, sometimes there is a loss of function - I have to open the driver panel by using the mouse, do a pressure test there, and then the pen would work again for some time. And some other things that were just a bit off, so I found myself more fiddling with tuning than actually drawing, which started to irritate me. Enough that I considered other options, and having used a Graphire and Intuos 2 A5 earlier I did really like the Wacom driver setup.

Now I found an Intuos4 used on ebay for a steal, only had to buy a new pen (got the classic, not the grip pen) and that made me think of trying to convert it into a monitor tablet, so I pounced.

Got a used screen for another 50€, disassembled it, and here I am.

The hard stuff:

Tablet: Intuos4 XL (PTK-1240) - that thing is huge!
Monitor/Panel: LG IPS 22MP55HQ, with Panel LM215WF3 (SL)(K1). 21,5", IPS, LED-Backlight, external power source - took some time to find that combination without shelling out too much money. Now I hope it isn't one of the hopeless cases that will always play jitterbells.

Additional stuff: Got a roll of copper tape and alumium tape each, will try both for shielding. Also several sheets of plastic foil for insulation.

I did the usual first, place the stripped screen onto the tablet enclosure to check if the pen signal connects - sadly, not at all.
So I removed the top cover of the intuos, put a sheet of plastic foil onto the digitizer and the screen on top of it - better, pen is recognized, but reading height is very low, I'd say only about 2-3mm above the screen. So possibilities of putting a protective sheet on it are "slim" .. haha. Maybe that will improve with shielding.
So, all in all right now: The good - it works, mostly.

Jitter: There is some. Very very little in the middle of the screen and towards the top, quite some more towards the bottom (where the long panel-circuirty is) and to the right, where the LED-strip is placed. Which brings me to:

Heat: The LED strip is getting very hot, to a degree where I only left it on for a minute or so, as I started to smell something, likely something being gassed or melting from a plastic by the heat. I had glued the strip(using the remaining glue on its backside) onto the thin black plastic bezel(which I use to keep the panels layers together) that came with the monitor. Originally it was attached to the side lip of the metal backside, but as that had to go of course, now there is no means of getting rid of the heat. I guess I will have to find another way to help with heat dissipation, maybe glueing it to an aluminum sheet angleded outwards?
A minor issue: The LED-strip is not powered by a flat cable, but by some bundled-up wires, which stick out from unter screen (and thereby lifting it up a bit at that corner) due to the angle of the connector. As I can't drill a hole into the digitizer to make room for them, I guess I will try to put the strip at the other side of the screen, so the connector points upwards, there should be more space.

I will take some more pictures and add the next steps planned.

****** Later this day:

I re-positioned the led-strip to let the wires go up instead of down. But this does not solve the problem as expected - the top  two layers of the screen are a bit wider than the lower ones, which I failed to notice before. To place the LED-strip directly against the diffuser-layer, I will either have to
a) displace the upper/lower layers against each other so there is no "step", diverging from how they were inside the original screen
b) exchange the wires against a flat cable - I found where to order them, but unsure what kind of connector I need to look for.

Also I gave shielding a first try: A piece of copper tape (insulated of course) between the panels circuits and the digitizer. Then I noticed (which may have something to do with it or just me not paying proper attention before): The jitter happens(now?) only to the sides of the screen, but there in almost equal strength. So: Middle of the screen (complete top to bottom) is almost jitter-free(I guess 1-2 pixels at most), at the sides (1-2cm) there is some jitter. Placing another strip of copper tape next to one side gave no change.
Random clicks also only appear in those small side-zones occasionally.
Right now, the tip of the pen doesn't seem to click - the cursor is moving, but no input if I touch the surface... but maybe I only changed something in the driver settings - will try again tomorrow.

Title: Re: Ribbitiq - trying not to break everything.
Post by: DaBotz on February 27, 2017, 02:56:26 PM
Calibration is often the weakest point of the "would-be Cintiqs" - none seems to work all that well.

Note that with a DIY, it is going to be a lot less easier - it is not the driver that does it for you but, as it is NOT it doing the work (in some half-botched way), you can finesse it over time.

The panel looks a lot like the one in the LG IPS224v that I used to overhaul my second build, but I am perplexed from the lack of reading height.

None of my builds (the four of them that I have at home...  I know, that's called mania; when I see a tablet and a screen, I feel this urge to rape the second on top of the first) has less than three mm of reading height above its glass cover (mostly, 3 mm thick), and they comprehend an unopened Intuos 4, with a laptop panel and a 2mm glass.

Check well the geometry of the assembly - with a build that big, it is very easy to "lose" mm of height, more so in this preliminary stage, that can be recovered with a careful final construction.

Title: Re: Ribbitiq - trying not to break everything.
Post by: Forestfrog on March 01, 2017, 08:44:12 AM
Thank you for your reply, DaBotz  :)

I would decribe the driver/calibration issue with the cintiq-alternatives like this - with the DIY-cintiq there are already some regions where it is perfectly good, and others where it is off, but there are driver functions that can be adjusted, hopefully correcting this. With the cintiq-alternatives, it is always and everywhere slightly off, and no way to correct it. (And of course parallax is already factored into it - it did not affect using the tablet for me).

Also I just checked, your monitor and the one I used were added to the prad-database with about 2 years difference, I don't know how long a certain panel model usually is used, but indeed they look quite similar.
I guess I will do a height check with the stack of paper on the digitizier and no panel in between, just to check if there is something off.

And coming from a totally different direction: I don't know if grounding might be and issue - the wall outlet the setup is connected to has little/bad grounding (old house). I will connect it to the one new outlet in the room and see if this does make any difference.

If it would not be too much of a hassle - could you drop some more keywords regarding what you call geometry? I am always thankful of getting a hint about stuff I may have been totally oblivious of before.

And finally, something completely different: I have an account issue, sent messages to Bernard, Aerendraca and Drewid already, but no reply since days - do you know how to reach someone who can access the "higher functions" of the forum?

Title: Re: Ribbitiq - trying not to break everything.
Post by: DaBotz on March 04, 2017, 02:49:30 PM
By geometry, I simply men the way you lay the stuff in the build, or details in the frame of the LCD.

Check that the Wacom board and its back metal pate  are plane and at the right distance one from each other (without the back metal plate, the reading height goes down significantly, and I have read somewhere here that some managed to augment it in their bit, by carefully augmenting the distance between the plate and the antennas board).
I realize that it is hardly going to be your case, as you were testing it "bare", but I have taken this habit of not moving the LCD, unless I have reassembled it enough that it is safe to handle without an excessive care.

In the case of my last build, I carved out the centre of the LCD backplate and remounted the remaining frame. It made my stripped LD almost as solid as it originally was, but it also meant that I conserved the half mm ridge that it has all around.

Then I connected it with the TCon board, so that those precious FFC didn't have to fear any mechanical stress. using two stainless steel flanges - and I added a couple of mm there.

Then lose another couple in the way the Wacom sat inside the case.

In the end, the first test run was terrible, because it looked like the new screen (which is 3 mm thinner than the old one, the glorious HP LE2201) gave me a worse reading height and more false clicks.

Then, I realized - and fixed - those small mistakes, and the end result is nice enough.

As far as how come nobody outside Wacom seems to be able to do a decent self-calibrating driver, it would just be really nice if they added a manual calibration interface for the ones like us.

Our panels may not be the same, but are still more similar that the ones in the IPS225v (which use a flat ffc instead of single cable, to power the led).

Also, being engineers a shy race averse to changing things just because, I would expect them to share as much characteristics as possible (for example, the LVDS timing characteristics).

On the other hands, my builds use Intuos 2 which, I suspect, where designed with more tolerance for EMF interference, and for going behind much thicker screens.

I never had to contact the admin or mods in this forum, so I have no idea...

Title: Re: Ribbitiq - trying not to break everything.
Post by: Forestfrog on March 05, 2017, 09:40:10 PM
Thank you again for your reply, DaBotz. I am learning things  :D

No time for any progress this weekend, but I did a bit of testing right now, with results I cannot really classify.

As the reading height was the thing that irked me the most right now, I did the paperstack-test.

Paper directly on digitizer, nothing in-between: 19,5-20mm reading height
Panel layers(turned off) on digitizer, paper on panel layers: 19,00mm reading height
Panel layers(turned on) on digitizer, paper on panel layers: 17,00mm (!) - which is acutally pretty good, compared to what I got before, when barely lifting the pen above the ~8-9mm of panel height(with bezel) gave me dropouts.

Also with panel turned on, little jitter almost everywhere (did not test consistently, as I only have A4 paper and did not want to move it all over the panel surface.)

There is another issue with the pen: It seems as if the tablet does take some time to recongize it, or if this has to be triggered by opening the driver window. Before doing this, then pen is dead for about a minute after I have the windows desktop. I still have to test if it is one or the other of the above or both. This also happened when working with bare digitizer, so it seems to be no interference issue. Once "found"(and within reading height) the pen/cursor glides as smooth as expected of a wacom product)

(I had to buy the classic pen separate from the Intuos4, as it came without any pen. Opted for classic pen instead of grip pen as I prefer a slim stylus)

Title: Re: Ribbitiq - trying not to break everything.
Post by: DaBotz on March 09, 2017, 09:47:28 PM
Let me know something about the Classic Pen .

I have a Intuos 4 Grip pen  - my first build, that I still use at night, was a "TabletMod L" derivative, witn an Intuos 4  large and a 15.4" AUO panel - and I hate it.

Not only the form of it, but also the fact that I feel the tip of it fitting a bit more loosely than in the Intuos 2 Pens (which already are not a Rotring RapidoGraph), and its pressure sensor not only is not very linear, it tends to "weaken up" after half an hour or so.

That, and the prices, are the factors that decided me to stick with the Intuos 2 (that I bought just to make a try for something with better colours, to use at night - it's... complicated) till I can.

I may opt for an Intuos 3 (the inch of screen or so that the Intuos 2 12x18 does not cover are a bit of a bother)... some day.

I've seen that newer Wacoms have 8092 pressure levels.

I hope that they got them more linear, because if they are like my I4 Grip Pen, I see no real improvement in more precision (which is not the same as accuracy, by the way - see the 5080 lpi - 0.005 mm precision - vs the 0.5 mm accuracy of the tablets beyond the Intuos 2... by the way, for a DiY-tiq spatial accuracy is as important as the precision, or more).  

How does feel, the I4 classic pen?

Title: Re: Ribbitiq - trying not to break everything.
Post by: Forestfrog on March 12, 2017, 02:09:39 PM
Regarding the classic pen - I like it. It is like a slim ballpoint pen, very leightweight, comfortable for my hand. Also, in what little testing I did, the pressure curve was alright. With the Ugee tablet I had to install lazy nezumi to adjust the pressure curve in PS (CS2, so afaik no suitable in-program-tool) so I did not have to press down so hard.
Somewhere I do have an Intuos2 A5, if I can find it (and install it side by side with the Intuos4) I can do a little more detailed comparison between both pens.

Attached a crappy picture of all three pens, top to bottom: Intuos 2 pen, Intuos4 classic pen, Ugee pen.

As for progress - I made a temporary heatsink for the LED-strip, just some bent aluminum plate. It works well as it is getting quite hot, so removing heat from the LEDs, but the thermal pad I used to stick both together wasn't sticky enough, so they came apart after a few minutes.

Still, during this time I was able to play with the monitor frequency a bit, moving it above 60Hz reduced jitter notably - now there is only a light wiggling at the left and right side of the screen.
The sweet spot starts at 62Hz, up to 64Hz I see no change and I didn't try higher frequencies(also no smaller increments, as the driver only allows full steps.) 59Hz made it worse, so I didn't go any further in that direction.

Also, I removed the copper tape covering the screen-circuitry at some point to check how much shielding this might play into it - absolutely no change, so jitter and shielding(that part) seem to have no connection right now.

There still is that issue that I have to open the driver window by mouse-click to get the pen recognized, before that the pen is being ignored.

Edit: The thing that is bugging me the most right now is the placement of the LED strip. At the moment I only manage a "visible, but quite dark" setup, especially at the right side (LED is on the left) of the screen it is very dark. I guess that has to do with finding the best height so the LEDs shine directly into the diffuser layer.
Does anyone know, if "closing" the other three sides (with black tape etc) so no light can escape would improve things?

Title: Re: Ribbitiq - trying not to break everything.
Post by: Ertew on March 12, 2017, 05:30:34 PM
Edit: The thing that is bugging me the most right now is the placement of the LED strip. At the moment I only manage a "visible, but quite dark" setup, especially at the right side (LED is on the left) of the screen it is very dark. I guess that has to do with finding the best height so the LEDs shine directly into the diffuser layer.
Does anyone know, if "closing" the other three sides (with black tape etc) so no light can escape would improve things?

If I understand You right, You have a problem with backlight diffuser. LEDs are on left and left side are far more brighter than the right side.
Yes, there are may be problem with alignment - LEDs emit light above diffuser, maybe directly at one of focusing layers. Proper diffuser should reflect small amount of light (from high light beam) at left side, more (from moderate light beam) at the middle and all remaining light (smallest light beam) at the right side.

The escaping light shuldn't be a problem. IMHO light leakage at remaining 3 sides will be less than 10%. Reflecting them back shouldn't improve the overall effect.
If light leakage disturb You (by illuminating buttons or desk), black tape are OK for catch the light. In Your case You may check something reflective to reflect light back into panel - silver tape or at least white tape.

Title: Re: Ribbitiq - trying not to break everything.
Post by: Forestfrog on March 13, 2017, 09:10:11 AM
Hello Ertew,

thanks for your answer - just a short reply from me before heading off to work:

I found part of the problem (or rather noticed what was right before my eyes....) the diffuser does have an orientation. Which in this case means, that at the small sides of the "glass"-block that is the diffuser, only one is open(right side), the other three are covered by thin strips of a white material.
Now as I rotated the LED-strip to get the wires coming out on the upper side, I also put it on the other side(left) of the panel to have it shining towards the diffuser again. But as I did not rotate the diffuser as well, it was always shining through one of those thin white covers, which likely swallowed a lot of light. Yay perceptive me  :P

So next step will be trying to rotate the stack below the LCD-layer by 180% to check if this solves the problem without creating any other. (I would like to try to switch the wires powering the LEDs for a flat cable, but I can't find a matching connector. Starting with not knowing how it is properly called so I can search for it.)

The correct alignment of LED strip and diffuser is still another point, of course.

Title: Re: Ribbitiq - trying not to break everything.
Post by: Ertew on March 13, 2017, 09:52:36 PM
Well. Diffuser have direction because LEDs are only at one side. If You 'fix' the problem by removing silver tape and glue it on the other side, the diffuser will still spread more light on right side and less on left.

Yes, when You move LED strip, You need to rotate diffuser layer.
What about upper layers? IMO You need to rotate only diffuser section which may contain only 'glass' or 'glass' and single sheet above. You can easy test all layers. If layer have different structure/density at different regions it's part of diffuser. If layer have the same structure at any point, it may be lens or something similar. You not need to rotate that.
And please pay attention to layers order and top side. If You mess with that, You will never found the right order.

BTW, why You need to rotate backlight and all optical layers? IMO You can rotate LCD glass without moving frame and backlight, then rotate whole module.

Title: Re: Ribbitiq - trying not to break everything.
Post by: Forestfrog on March 19, 2017, 01:33:56 PM
In the end I did almost what you said - removed the LCD glass, rotated all other layers together (to avoid dust getting in or mixing up any layers and to try and change as little as possible) and then put the LCD glass on top of them again. Great improvement in lighting - I have to do some further testing, but right now I'd say, the screen is workable this way (only regarding distribution of light).

Also I made a new heatsink for the LEDs, this time a U-shaped bar of aluminum cut to size and a different pad to stick both together. This works fine as a temporary solution, so I do not have to hurry that much when testing for fear of anything getting too hot. The final solution will be a massive bar of aluminum to get as much heat away from the LEDs as possible to not shorten their lifespan from overheating.

Also I was pleasantly surprised to find how little jitter I have right now, and with no shielding at all. I hope I am not too optimistic when I think I will try to improve this a bit more with shielding and grounding (another thing I will have to read up on here in the forum again, so I don't connect any wrong stuff) - and after that there is only left to build an enclosure (which I know will still take a lot of time, but mechanics I can handle, it is electronics that I know almost nothing of)

Pics will follow.

Title: Re: Ribbitiq - trying not to break everything.
Post by: Forestfrog on March 19, 2017, 05:36:53 PM
So, pictures:

The LED-strip attached to the aluminum-bar:

And from the top:

A little test in PS:

And a closer look:

The small artifacts along the lines are due to me not having the screen and tablet properly aligned I guess, the angle likely is slightly off - something to pay very close attention to when building the enclosure.

Title: Re: Ribbitiq - trying not to break everything.
Post by: Forestfrog on May 01, 2017, 08:27:47 PM
Hi everyone,

long time no update annnnd - likely not one for a good chunk of time to come.

In the meantime two things happened: I got my hands on a 21UX (Intuos3 base) only to find out that it wasn't the revelation I expected it to be. The less than stellar screen, the drawing surface and the light jitter ot the top/bottom of the screen all came together to make it "ok" at best for me. Plus the hefty weight, the special power-plug... it is sitting on a shelf now until I have decided if to sell it or keep it for testing/comparison.

The other thing - I gave my Ugee tablet another go and noticed, that it is almost spot-on without any calibration if I go with windows-driver only and install no AMD GPU driver. Keeping it like that causes other issues, but all in all, with the far better screen and closing in on a working solution, I decided to put the Ribbitiq on a very low backburner. I might fiddle with it now and then again, but as I mostly want to draw and may have a solution at hand - it isn't as urgent as before.

I will still check now and then if there are questions or anything, but updates may take a long time.


Title: Re: Ribbitiq - trying not to break everything.
Post by: DaBotz on May 07, 2017, 10:06:01 AM
The important part is being able to draw... if fiddling with the Ribbtiq and then comparing it, the 21UX and the UGee got you to find a way to use the latter to better results, I think that it is just right