Bongofish
November 19, 2017, 01:14:24 PM *
Welcome, Guest. Please login or register.

Login with username, password and session length
News: You can attach photos hosted by the forum rather than using an external image hosting site, this means they will stay forever and not disappear after a year or two.
 
   Home   Help Search Login Register  
Pages: [1] 2 3
  Print  
Author Topic: Gigantiq - a tale of desmeasured greed.  (Read 12254 times)
DaBotz
Sr. Member
****
Posts: 227



View Profile WWW
« on: March 20, 2014, 06:18:31 PM »

In the past, I have mounted a variant of the TableMod with a wood case that I called CabinetiQ.
It went too well, so I decided to go for a Bigger tablet this time.

I bought an Intuos2 A3 (yessss - my preciousssss), this beast:


* DSC00963.jpg (76.31 KB. 1024x768 - viewed 351 times.)


This time, I was very undecided about following Don Shole's specs for his TabletMod XL;

The Intuos2 A3 is more a 16:10 than a 16:9, so the two AOC monitors he suggests falls or too wide ( the e2243f) or a bit too small (the e2043f). Also, in many places I read that each tablet/monitor  combination is slightly different...

I finally entered in Ebay looking for monitors  22" in 16:10, with LED back-lighting.

And here is where I made my first mistake.  I bought an HP LE2201w that was marked as LED back-lighting by the vendor.

I checked some online reviews, where they spoke well of its colours (and one of the reviewers also defined it a LED Screen).

Less than a week and it was mine.

I checked that it worked, and it works well, with quite brilliant colours.

It is not an IPS, but I didn't expect one in a 16:10 format ( it's kind of old, I gather;  though I still prefer it to the frigging 16:9). Still, the colours are one magnitude better than the ones in the CabinetiQ.

I went on to open it up and see what I had in front....   first surprise: two CCFL cables.


* DSC00955.jpg (123.1 KB. 1024x768 - viewed 307 times.)

So, LED my ass. LE was for Lenovo, evidently. Damn.

There is no separated inverter, it is integrated with the power source, but the driver board is separated...


* DSC00958.jpg (101.85 KB. 1024x768 - viewed 348 times.)

this is not so much a cause of horror as what I realized soon after.

The LCD panel is a Samsung LTM220MT05


* DSC00960.jpg (83.38 KB. 1024x768 - viewed 286 times.)


, whose most worrying characteristic is its... DEPTH!

It's 13mm tall (17 with the case covering the circuit board) , and I think it cannot really be slimmed down below 12mm with even the most crafty stripping on my part.

Adding even a lean acrylic glass means It'll end in the 14-15 mm range... this, hypothesizing that I could get rid of all the metal without affecting any functionality (why one of the cables of the CCFL is bigger than the other? probably because the metallic case involving the panel acts as a common ground; I had seen the same with the AUO of the Cabinetic; I had need to ground a metal strip to the panel's circuit board, in order for the back lamp to work properly).

So, now, I do not really know what to do... mount the monitor back together, sell it and look for an AOC e2043f?

Try to dismount the Wacom and see if it can read reasonably the pen position at that height?

But I doubt the sensors table is more than a 4 mm below the surface, and the specs for Intuos2 and 3 states a maximum pen height of 6 mm... even stripping, I would be some 2-3 mm short before placing a glass?

Grrr - I feel stupid...

Anybody that has done a project on an XD knows how much reading height is there, above the circuit board?

[Update from after : with a 3mm glass on top. the pen position is still recognized to an height of some 5 mm... a 13 mm thick panel proved plenty usable]


« Last Edit: November 20, 2015, 11:31:15 AM by DaBotz » Logged

The most incredible artist of... Barbanza?
Aerendraca
Administrator
Hero Member
*****
Posts: 1068


View Profile
« Reply #1 on: March 20, 2014, 08:09:29 PM »

Doh, I've made this mistake before too.

I have an Intuos 2 A4 12x12 and I've measured a workable reading height of 14mm above the Wacom plastic case. Whilst this is workable it doesn't leave much room to hover above the surface.
Logged
DaBotz
Sr. Member
****
Posts: 227



View Profile WWW
« Reply #2 on: April 06, 2014, 04:42:01 PM »

Thanks, Aerendraca.

Update: I've spent some time pondering upon what to do, re-read your post on the 12x12 intuos 2, Aerendraca (which makes clear that, not much may be, but there cold be enough height above the tablet), and finally I decided to strip the LCD.

I did it as carefully as I could, and the result is pretty neat.


* DSC01002.JPG (149.35 KB. 1024x768 - viewed 372 times.)


... I can still piece the monitor back together and re-sell it, or use it (it's a pretty good monitor) as a "colours mirror" for the Cabinetiq (My philips 226v is a 16:9, which make is it a bit problematic on that respect. Also, I can't seem to find a good calibration for its colours, no matter what; The HP looks a little bit better) .

I carved out most of the backplate of the panel, and kept a vestigial metal frame all around it.

It was inevitable, as the top and the bottom also acts as case for the CCFL, and so couldn't really be eliminated.  Also, it would have been too frail, otherwise.

It is possible that it makes me lose some pixels on top and bottom of the screen. That said, I usually have a mouse at hand for dealing with the main screen anyway...

Width-wise, the screen is also a couple of cm bigger than the active area, (unfortunately, I failed to locate any 21" 16:10). I'll just have to be creative with the "mapping".

As far as I can tell, the remaining frame does not interfere with the pen signal, and I have some mm of "air" above the screen, yet a 1mm acrylic would be better (unfortunately, I think the thinnest one I can find here is 2mm) .

However, I do experience plenty of  jitter and false clicks (well, plenty - it's comparable to the CabinetiQ's in some of its worst days, before beefing up the common mode rejecters solved the issue), which is quite a put off.

I read some quotes of the fact that linking together the panel board and the Wacom base plate can alleviate this symptoms.

I am restive about stitching together directly two things that are not designed to be joined...

I expect that some "trick" is necessary, to avoid polarisation current issues.  Something like a good old capacitor to be placed between the two, as "big" as possible; I'll rummage a bit more, till I find the relevant thread.

Also, I hope the same common mode rejecters that I used with the CabinetiQ could help
(by the way, is it possible that the jitter is also caused by ghosting currents passing through the capacitor  "created" by the tft matrix and the wacom antennas?  the sensibility to the presence of current mode filters on both lines  - at least on the C...Q - would suggest it , in my opinion).

I've also ordered some CCFL extenders from NYJTouch. On the C.Q I just soldered a couple of 220V cables, after having intertwined the strands (soldering did work to carry analogic satellite signals, when I installed my satellite dish, so I did not see how could it go wrong for something so low a frequency)    

They are going to be here in ten days or so.

Anyway, looking through the site, I also found this post - http://forum.bongofish.co.uk/index.php?topic=2302.0 - where it is indicated that the LG IPS224V could be a pretty good match for the Intuos2.

Damn, had I read it last month (it's a f***ing 16:9, though).

On the LG site, I see it has been superseded by the "6" series (LG IPS226v). It's a bit too steep for my budget, but I'll keep it in mind.

In the meanwhile, I'll go on with the "case" works and see if the jiiter is still as bad, with no CCFL involved (as far as I can tell, the ccfl are not a notable issue) and with a grounded sheld in front of the power board
(If I have to switch to a NIJtouch controller, I can as well sell the thing and go for the LG anyway. Moneywise, it would be sensible enough).

As it is going to be "thicker" than the C...Q, I went for a thicker (10mm) and stiffier plywood base, in which I drilled a hole the size of Wacom's secondary board.

I'm going to cover the hole with a metal shield that I will connect to "earth" (the power board for the panel is going to be right below it, unless the CCFL extenders prove very but very effective, in which case I could move it around a bit).


* DSC01004.JPG (161.91 KB. 1024x768 - viewed 325 times.)
 

* DSC01006.JPG (104.67 KB. 1024x768 - viewed 324 times.)


For now, let's keep my fingers' crossed.



« Last Edit: April 12, 2014, 10:50:48 AM by DaBotz » Logged

The most incredible artist of... Barbanza?
DaBotz
Sr. Member
****
Posts: 227



View Profile WWW
« Reply #3 on: April 12, 2014, 12:45:32 PM »

Update:

While I wait for the CCFL cable exenders, I built a sling to support the panel circuit board. I fixed on top of it the original board cover, and I hope that the resulting Faraday cage will shield some of the noise.


* [1024px]-DSC01025.JPG (84.19 KB. 1024x768 - viewed 355 times.)


I acquired a burnt PC power unit and salvaged the ferrite (and soft-iron) nuclei of two of tis transformers, that I plan to use them as common mode chokes on the Wacom USB cable and the VGA cable.

I'll look if I can find another ferrite choke for the fcc that goes from the controller to the LCD panel, and for the CCFL cables (I shall investigate more before trying this last trick, though - while the noise on the CabinetiQ was very much a product of common mode currents seeping in the USB and power cables, I have no idea if the same can be said for this new one.)

That said, even if the panel did prove unsuitable, and I'll have to switch to the aforementioned LG monitor, I still have to add hotkeys and wheels to the thing to get it usable for me.

So, I spent a couple of hours "modding" three mouses (in the end, I'll use three wheels: zoom, brush size and LazyNezumi mode).

By the way, as time goes by, I finally picked a "clean" way to mod the mouses.

It all start buying a small mouse (mini mouse) - in my case, from NGS.


* [1024px]-DSC01023.JPG (133.79 KB. 1024x768 - viewed 309 times.)
     

* [1024px]-DSC01024.JPG (151.23 KB. 1024x768 - viewed 311 times.)


These  small mouses use a rotary encoder - instead of the usual set of IR diodes that you can find in normal sized mouses - to obviate the lack of space due to the smaller wheel.

The encoder is less durable but a lot more handy than the diodes  

Unfortunately, these encoders are originally with a horizontal axle (obviously), while I really need it to be vertical.

The first time I modded one, I made an unholy mess, with some 80 cm of cables connecting a pcb with the encoder and the three buttons to the original circuit.

Of course, these picked up EM "noise" wonderfully and fed it into the USB system (which is, I gather, not very RF-noise proof).

The second time, it was slightly better.

Third is a charm, as they say.

I decided that less is - in this case - decidedly more.

After having de-soldered the encoder, I bent its three input 90 degrees forward (really, I did it as I was de-soldering them: it's the cleanest way I have to do the job, warm up the thin and bent the damned things out - the pin holes are open on one side ).

Then I de-soldered also the left and right click buttons (one of them, I needed to de-solder anyway, to have a platform onto which to anchor the turned-vertical-encoder)

I soldered the "structural" pin of the encoder to a pretty small piece of PCB, that I glued at 90ยบ from the mouse board (well, actually, it is glued to a small cubic piece of wood that is then glued to the PCB - I'm a wood guy and I didn't have any Lego around; if someone know how to bend PCBs, let me know) so that the encoder axle was vertical (not exactly, damn me). Once the glue was dried and the whole shtick was solid, I re-soldered the encoder's signal pins and verified that it works.

I also placed the removed buttons on two auxiliary PCBs connected with some cm of cables to the mouse board, and verified that they, also, work as they should.

Which is not so obvious: on this mouse, both buttons share a common line, and is the value of a resistance in series with the switch that let know the mouse chip which button is pushed.

As a result of this design (which, among other things, grants that the left click "overrule" the right one), a flimsy solder may transform the left click in a right one.

As knobs, once again I will go with polypropylene office chairs' wheels.

I can't find my preferred - 40mm x 20kg -  , so I have  to use the 50mm (30 kg) I had lying around  (I have bought some to replace the ones under my chair that I accidentally destroyed a while ago - I always buy a spare sample).

Now, there is a moderately simple and very cheap trick to connect these to the encoder...

The hole in their back is slightly less than the girth of a classic aerial coaxial cable, so you can take one, stick it into the hole, cut it after 1-1.5 cm while leaving the copper core intact, and extract the copper core from the cable.

Gently nudge a little more the de-cored cable into the hole (with a hammer, but gently), and then you can carve away the external shielding, leaving a short, reasonably rigid and smooth nylon pivot, pretty rigidly connected to the wheel.

In the place where originally was the copper core (a bit too thin and smooth), I place (very gently, hammering it down, but gently) a tooth-picker that then I slim down a tiny bit hexagonally.

The tooth-picker is strong more than enough to carry the movement to the encoder but its wood is also tender enough that the plastic of the encoder can shape it when posed into the pivot siege, and I count on the nylon to handle side forces (though, the wheels of the Cabinetiq are running on tooth-pickers alone since I modded the first wheel), as I can place it into inside a small hollow screw (used to lodge 6mm screws - threaded for metal and easy removal - in wood).


* [1024px]-DSC01022.JPG (111.68 KB. 1024x768 - viewed 322 times.)



* [1024px]-DSC01016.JPG (126.9 KB. 1024x768 - viewed 363 times.)



* [1024px]-DSC01018.JPG (136.28 KB. 1024x768 - viewed 314 times.)


I have an idea of how placing them together, but I feel it is better that I see if I have to move the power source away, before start cutting plywood.

FOr the moment, the disposition could be something like this



* [1024px]-DSC01015.JPG (117.43 KB. 1024x768 - viewed 318 times.)


But, as I said, refinement may still come            
  



* [1024px]-DSC01010.JPG (135.73 KB, 1024x768 - viewed 302 times.)

* [1024px]-DSC01011.JPG (98.85 KB, 1024x768 - viewed 284 times.)
« Last Edit: April 12, 2014, 01:09:44 PM by DaBotz » Logged

The most incredible artist of... Barbanza?
DaBotz
Sr. Member
****
Posts: 227



View Profile WWW
« Reply #4 on: April 26, 2014, 02:46:43 PM »

Fast update: received the CCFL cable extenders from NyJtouch.

Tried a fast run, with power souce aside the screen, tablet upside down, "armor" around the LCD board and two ridiculously oversized common mode chokes on the usb cable.

Jitter is down to 3-4 pixels and no apparent "false clicks". I think it can be done.

I need to think well at the auxliary inputs (three wheels, their buttons,  a touch-pad and a compact keyboard), now.
Logged

The most incredible artist of... Barbanza?
DaBotz
Sr. Member
****
Posts: 227



View Profile WWW
« Reply #5 on: May 01, 2014, 01:27:24 AM »

Update: I have build it.


* [1024px]-DSC01066.JPG (190.13 KB. 1024x768 - viewed 463 times.)
 

* [1024px]-DSC01068.JPG (211.28 KB. 1024x768 - viewed 384 times.)


And it works relatively fine.

As I was saying in my last update, jitter with the power source out of the way seemed reasonable.

I have decided, finally, to build the table alone, and leave "stubs" to attach auxiliary inputs after.  This choice, in turn, is due to my decision of using this with Linux - Ubuntu, to be more precise.

Now, in Linux HidMacros does not work, so I'll have to whip some kind of "overriding driver" for the aux keyboard and the wheels.

From the documentation, it could be relatively forward to modify the reference usbmouse.c and usbkeybd.c "base" drivers, but I rather not tie all to such an hazardous step.

Right now I think I should rather draw for a while and see where to put things, before having another hardware spurt.

And, anyway, the proiority seems to be for "enahncing" the stand...

I'd rather not go for an "Ergotron", but I need to have a stand able to present the screen at about half a meter from its base....

Logged

The most incredible artist of... Barbanza?
DaBotz
Sr. Member
****
Posts: 227



View Profile WWW
« Reply #6 on: May 01, 2014, 02:40:26 AM »

I should add that I decided to call this UbiQ , now.

"GigantiQ" was a good - slightly self-derogatory - name when I thought it would not work, but now that it does, it is too much bombastic.

Being wary of jitter, I decided to place the power source "aside" the tablet/lcd panel (instead of below) ) and to leave the main connectors on the same side (at the "top").

* [1024px]-DSC01054.JPG (239.8 KB. 1024x768 - viewed 338 times.)


This has resulted in a big case (65x62 cm) , because the power source was an horrible 15x18 cm.

Note: one of the many differences between this and the "CabinetiQ" is that UbiQ has no "privileged direction", so "top" is just for the moment.


As I hate the idea of having to manhandle directly the Intuos usb cable, I placed all of it "inside" the shell, with a good length around some ferrite nuclei I scavenged from a burnt down PC power supply, and left just an head to connect with.

... To allow for expansion, I added a powered usb  hub ( to which the Intuos2 is connected), which expose three usb slots on the UbiQ left side (right below the screen buttons).


* [1024px]-DSC01078.JPG (148.21 KB. 1024x768 - viewed 294 times.)


The aforementioned usb "head" doesn't come from the Intuos, it is the hub root.

When I will have ready the hotkeys "console" (or whatever) all that will be needed is to plug it in.

In the meanwhile, a numeric pad will do (as it has always done with the Huion 610, to date) as hotkeys hardware.

The Linux driver has almost no GUI interface, but it is very configurable through a command line tool called xsetwacom.

I'd rather not touch the true configuration files of the Wacom, as making a mistake there breaks the X11 server and  I know for experience that it may be quite bothersome to "rollback" from that state - in my system, it combines with the CIFS initialization that fails at boot because it starts before the network service, and it all makes things a bit less straightforward.

Among the many "valuable pluses" of the linux driver, it allows to set an "active area" bigger than the "physical" active area of the tablet.

This means that it has been possible to calibrate the UbiQ - that has a screen wider than the physical area - without touching the projection on screen, just resizing its logical area.

Which, in turn, makes easy to handle multi-monitors, as it just need to use a mapToOutput command to link the tablet to its screen.

The driver also allows to set the number of samples over which the pen positions (and pressure) are averaged ( a sliding door average).

Selecting a value of 5 samples, jitter goes down very nicely to around 2-3 px, which I can work with, without making the pen too sluggish, so I proclaim myself content of the result.

Finally, but not lastly, the driver allows to set a pressure curve that is a 4-points bezier, so I have compensated for the "horrible" off-the-shelf pressure curve of the Intuos2.

Anyway, as an example, here is the current configuration for the UbiQ,  

* configure-ubiq.sh.txt (2.07 KB - downloaded 138 times.)



For the moment I am pretty happy with this build, though I'll have to create a  new stand for it...




* configure-ubiq.sh.txt (2.07 KB - downloaded 138 times.)
« Last Edit: May 05, 2014, 09:23:40 PM by DaBotz » Logged

The most incredible artist of... Barbanza?
Aerendraca
Administrator
Hero Member
*****
Posts: 1068


View Profile
« Reply #7 on: May 01, 2014, 04:02:21 AM »

Amazing stuff, an excellent job well done! Now you get to have a play! I'm a bit jealous.
Logged
DaBotz
Sr. Member
****
Posts: 227



View Profile WWW
« Reply #8 on: May 06, 2014, 01:08:28 AM »

Thanks, Aerendraca. I am playng with it... more fiddling than drawing, though.

I am getting "around" the "issues" of the UBiQ.

I replaced the stand that I used initially, the "Moloch" from the CabinetiQ, with a pivoting wooden arm attached to the library in my sitting room.

It is ugly as hell, but it works (and I do not know if I have the "tecnology" to d a better work; yes, an iron tube could be smaller, but I do not care that much for soldering steel beyond not-stressed works like the shield I placed around the power unit

I then proceeded to modify some of the pieces in the old stand, and use it as a height-regulable support.

Even uglier.

I have managed to place together a python script that allows to change the Wacom driver smoothness from inside GIMP (and, by the way, plug ins in GIMP can be associated to key short-cuts) even if, maybe, the best solution would be to use a bash script and an alias for it, so that the fas setting can be invoked system-wide.

Right now, the biggest "problem" is that I see no fast way to create wheels in Ubuntu.

I could create an overriding driver for the class of usb mouse that I use for building the wheels, but I can't seem to be able to convince Linux to give me all the headers that I need (or something).

That said, should I count it as a successful build?  It does work fine...


* 00_ubiq_plugin (copia).py.txt (1.01 KB - downloaded 127 times.)

* 00_ubiq_plugin (copia).py.txt (1.01 KB - downloaded 127 times.)
Logged

The most incredible artist of... Barbanza?
Aerendraca
Administrator
Hero Member
*****
Posts: 1068


View Profile
« Reply #9 on: May 06, 2014, 07:52:06 AM »

I agree that this is a successful build and have used my new privileges to move the thread. Once again great job!
Logged
DaBotz
Sr. Member
****
Posts: 227



View Profile WWW
« Reply #10 on: May 14, 2014, 07:30:21 PM »

Fast update: I have built the keyboard/wheels control, here it is:


* 00-[1024px]-DSC01184.JPG (199.28 KB. 1024x768 - viewed 476 times.)


Getting the wheels to work in Ubuntu has proven relatively simple ("simple" being an euphemism, of course).

I achieved it using the xinput set-button-map command, and a small utility called imwheel (which allows to replace wheels events with keystroke macros, depending also on the active window).

I remapped the "usual" buttons of my mouses to button IDs beyond the reach of "normal mouses", and used xbindkeys to bind them to a script that raises/lowers the Wacom driver smoothing factor, and to  the brush hardness (in all, the mouses do what I need, wthout interfering with the works of the normal pointers).

For the moment, I'm still fighting with the keyboard, as my "main" keyboard, a wireless Logitech K400, does not accept to keep its own layout and follows whichever was that of the last keyboard used.

My idea was to set up a bogus keyboard layout made of "out of boundaries" unicode characters, and to use Autokey to bind macros to the keys, but this would end in the K400 emulating the same unusable layout the first key I press on the command keyboard, so I would have to have a third keyboard (even just a numeric pad) to reset thek400 to a usable layout...

A hassle, in a word.


Here is, also, the last version of the UBiQ configurator and of other scripts I use, to whomever it may concern.

And a bogus logo, of course.


* .configure-ubiq.png (7.44 KB. 100x100 - viewed 955 times.)


P.s. - Yes, Yes, I realize now what that border around the keyboard and the wheels looks like... I suppose I'll change it, the next time I have to modify any thing.

* .configure-ubiq.sh.txt (0.03 KB - downloaded 121 times.)
* .imwheelrc.txt (8.28 KB - downloaded 132 times.)
* .xbindkeysrc.txt (2.4 KB - downloaded 135 times.)
* ubiq-rawsample-down.txt (0.4 KB - downloaded 126 times.)
* configure-ubiq.txt (3.64 KB - downloaded 157 times.)
« Last Edit: May 14, 2014, 11:05:33 PM by DaBotz » Logged

The most incredible artist of... Barbanza?
Aerendraca
Administrator
Hero Member
*****
Posts: 1068


View Profile
« Reply #11 on: May 15, 2014, 07:42:47 AM »

 Cheesy I totally hadn't noticed about the border until you pointed it out!

It looks pretty well made, how is it in terms of position? Is it natural to use?
Logged
DaBotz
Sr. Member
****
Posts: 227



View Profile WWW
« Reply #12 on: May 15, 2014, 06:13:02 PM »

Position-wise, I still have to figure where to place the screen.
 
For the moment, as I use it on my sofa, the command board is in a pretty good position, i.e. where I naturally left the hand before building it.

If I had to use it in the same place where I have the Cabinetiq, I'd have to raise it some ten cm and place it parallel to the main screen.

It would take a couple of minutes... under the left border of the Screen I have left 4 threaded support for 6mm screws, so I can wiggle it around quite a bit.

At the moment, the only thing I miss is some labels to put on the command keyboard, as I do not remember what key is what.

I'm still bending my mind around the k400 issue, though...

That buggers me, because it look some kind of not-complyingness on the Logitech part (the fact that it does not notify itself as a keyboard and a pointer is "odd").
Logged

The most incredible artist of... Barbanza?
DaBotz
Sr. Member
****
Posts: 227



View Profile WWW
« Reply #13 on: May 21, 2014, 08:49:33 PM »

Here, a video with the beast in action:

http://youtu.be/vTE7vYDXT80

.
Logged

The most incredible artist of... Barbanza?
Aerendraca
Administrator
Hero Member
*****
Posts: 1068


View Profile
« Reply #14 on: May 22, 2014, 08:04:31 AM »

Great video DaBotz, it's nice to see it fully in action. Looks great and seems like it works very well. Very nice.
Logged
Pages: [1] 2 3
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.21 | SMF © 2015, Simple Machines Valid XHTML 1.0! Valid CSS!