test1
March 25, 2017, 08:40:23 AM *
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 ... 15
  Print  
Author Topic: Serenitiq - Intuos2 12x12Oversize & erm... Screen TBA  (Read 39814 times)
Aerendraca
Administrator
Hero Member
*****
Posts: 1057


View Profile
« on: September 22, 2012, 07:51:00 PM »

Note to new readers: There is a lot of useful information in this thread which some of you might find interesting, however since it is a long thread I feel it is my duty to provide a pre-warning that this project did not yield a finished product. Please don't let this put you off browsing through it though since there are some inspired moments which may well spark an idea in you own DIY Cintiqs. Stay tuned for the Serenitiq2.

Also, I have sifted through the information provided within and provided corrections to statements/calculations where necessary. I will continue to do so until this thread is dead. (10/05/2013)





So I've been in the shadows of this forum for some time soaking up all the great information that many of you have provided, and for that I thank you all. Whilst in the shadows my build had already begun so I've already made some progress which - as the title suggests - is based on the intuos2 12x12.

Here's a little about what's been happening so far and maybe a little about how I see this project going.

First off the insides of my gear so far sprawled rather neatly across my table:


Keen eyed builders might spot that the wacom is upside-down, this is deliberate since after some consideration I've decided that it would be more useful to have the quick key buttons at the bottom rather than the top. Eventually I plan to relocate the power LED to the adjacent end, but for now it can stay put.

The screen is shown in it natural orientation ontop of the tablet board and has been stripped of all but the reflective sheets and plastic of the back light;  The plastic bezel was just too tall - some 15mm - total height minus all the other stuff is 9mm. Originally I had concerns about the thickness of the panel but after testing the working height of the tablet it seems I'm good up to about 14mm allowing for some distance to move the pen around.

Testing the screen on the tablet revealed my worst fear - The Jitters!! And they were bad, oh so bad. So, I got hold of a copy of Powerstrip and began tweaking the horizontal frequencies to see if they could be soothed. After a couple of hours of fumbling around with the software, accidentally resetting instead of accepting changes, and the occasional crash of the panel I finally found a range of frequencies that tamed the jittery beast quite a bit (although not completely). At this stage I'm not too concerned though, there was no shielding and I have plans!

The forum has taught me that most of the interference comes from the power supply and ccfls of the back light, and my power supply is a switch-mode power supply as I'm sure most are. Unfortunately this type of power supply produces radio frequency interference (RFI) typically in the range 100-150Khz which seems to be perfect at disrupting the Wacom circuitry. As well as this, the ccfls also emit a certain amount of RFI, Fab!

So, my plan?

Well, since the power supply includes the inverter onboard, there's not much I can do with that, so in the metaphorical bin it goes. Don't forget the inverter was also onboard, bye bye ccfls! So what to do about power? And what about a back light?

« Last Edit: June 03, 2013, 07:16:39 AM by Aerendraca » Logged
Aerendraca
Administrator
Hero Member
*****
Posts: 1057


View Profile
« Reply #1 on: September 22, 2012, 08:30:44 PM »

Answer:
Get a laptop power brick, build a couple of regulator circuits (12V and 5V), also a transistor switch circuit, and possibly a dimming circuit for true integration.
Use 3528 SMD birght white LEDs [120LEDs per metre] in place of the ccfls!

First problem, how much current will I need?
This needed breaking down into areas of power consumption:
1. The tft panel itself
2. The driver board
3. The backlight

Scouring the internet for panel information lead me to a fab website called www.Panelook.com.
#Top Tip# Rather than going to Panelook and using their search bar I suggest going to Google and typing your panel number followed by 'Panelook' and 'Overview'. For some reason their search bar didn't find my model number but Google did!

Under the section 'Signal Interface' they show the input voltage/current/power consumed by the panel. Nice!

1. 5V at 0.74A max - Doing the maths that's 3.7W of power (although it's actually shown on the overview Wink )

Now to work out the power consumption of the driver board is going to require a little bit of investigative work and deductive reasoning (as well as some overestimating!).

In the same overview gathered from Panelook it shows under the 'Backlight System' field that the ccfls run at 640V (740V max) with a max current of 7.5mA. The values shown are per ccfl, I have 4 of them on this monitor so that's a total of 30mA current and *22.2W power. *Note: this is based on maximums, most of the time the backlight will operate at about 18W.

So far then we have a total power consumption of about 18W+4W=22W. The monitor is rated at 30W so we have 8W remaining as a huge over-estimate. The driverboard uses 12V of potential so 8/12 = 0.67A max.

2. Driverboard requires no more than 0.67A (670mA)

3. The diy LED backlight modification - This one's easy!

A reel of 3528 SMD LEDs comes as 600 per 5m, or 120 per m x 5. The total current draw of all 600 is given as 4A.

Right, how many do I need?

5m or 5000mm/600LEDs=8.33mm between LEDs

My ccfls are 306mm wide:
306mm/8.33mm=37LEDs

Let's call this 36LEDs since they are divided into banks of 3 and that gives 12 banks exactly. Not to mention that I wont be using all of the 17" screen so coverage to the corners isn't essential.

I need a row of lights at the top and the bottom, so that's 72LEDs in total.

600LEDs (including the resistors) consume 4A of current so we should calculate the current draw per bank to also include the single resistor which is in series with each bank:

600LEDs/(3LEDs per bank)=200banks

4Amps/200banks=0.02A (20mA per bank)

There's 72 of them so that's 24banks of 3:
24banks x 20mA=480mA (total current consumption)


So the grand total current consumption for the 5V line is 740mA + 20% =~1A
And, 480mA + 670mA = 1150mA + 20% =~1.5A for the 12V line.

Therefore a power supply that provides 12V 4A will be perfect to allow for any further mods that might crop up along the way. Now to build the rectifiers.

**Note: This is all a bit fuzzy. Some assumptions have been made in the calculations. This is not intended to give exact figures, just an educated guess at current requirements.**
« Last Edit: September 24, 2012, 09:29:49 PM by Aerendraca » Logged
GwenLP
Full Member
***
Posts: 89


View Profile
« Reply #2 on: September 22, 2012, 11:59:43 PM »

Hi Aerendraca !

Nice momentum, you seem full of confidence !
One thing striked me in your post, and that's the part about power supply. You seem to say that some type of power supply may create radio interference.
I've just ordered a controller from NJYTouch, and the power supply that comes along is cheap enough to be conspicuous... 8€.
So i guess it may well be the nasty type (and it's small too, so i think it's an hint to the fact it may well be of the switch-mode type).

Now, what do you recommend ? I am not able to
Quote
Get a laptop power brick, build a couple of regulator circuits (12V and 5V), also a transistor switch circuit
.

What sort of power supply would you recommend ?
I may have an old Macintosh LC475 power supply. I am sure it would deliver 12v DC. but maybe not under enough amperage...also the specsheet of the panel gives me maximum values of 4.2W and 1.0W for LED backlight and Logic.
I don't know how much the controller would draw, but with 4A, I am allowed 4×12=48W am i right ?
« Last Edit: September 23, 2012, 12:18:17 PM by GwenLP » Logged
bernard
Administrator
Hero Member
*****
Posts: 2584


pato mania


View Profile
« Reply #3 on: September 23, 2012, 02:09:40 AM »

Welcome!  You are not at your first electronic project, that shows!

Can you give out more information on what you intent to do for the LED "driverboard" -- are you going to do a custom one?  Will you connect this to a "dim" control that's already in the monitor?  (I often see a voltage-based dim or PWM-based dim signals coming out of the logic board)
Logged
Aerendraca
Administrator
Hero Member
*****
Posts: 1057


View Profile
« Reply #4 on: September 23, 2012, 11:53:19 AM »

Hi  GwenLP and bernard,

The power supply you mention GwenLP, is it a power brick that will be situated away from the tablet? If so no worries. Just to clarify a little, I believe that not all switch-mode power supplies will produce RFI at frequencies that will interfere with the tablet, and some have some quite sophisticated circuitry designed especially to minimize this, however they tend to be a bit more expensive. Plus there's always the application of shielding for those who have circuit board power supplies which will be placed under the tablet, as many others have done on the forums. I think that the NJYTouch LVDS boards have a DC power jack input which suggests that you will be using a brick, so all good!

As for the LED driver, well this is where I am at currently. I'm designing the circuitry at the moment which will interface with the lcd driverboard to switch on the LEDs and apply dimming from the OSD. Bernard this is where you hit the nail on the head, although I would have preferred the output for dimming to be voltage based, my controller has a PWM output for dimming to control the inverter circuit meaning that I have some more research to do before I can give any more details of my solution. If in the mean time I stumble across some commercial method I will be sure to try it (so long as it's cheap enough) and post my findings and the source.
It is the fact that I'm not currently sure how to use the PWM signal that excites me about this idea, and I will post everything I discover along the way!
Logged
Aerendraca
Administrator
Hero Member
*****
Posts: 1057


View Profile
« Reply #5 on: September 23, 2012, 02:21:39 PM »

Thought I would post a quick tip for anyone that intends on playing about the with power supply board that comes inside a monitor (also applies to most other power boards too).

So you don't injure yourself by getting zapped from the board I recommend a little trick I picked up while browsing an old TV repair website. Pictured below is a closeup of the filter capacitor on the power supply that came with my monitor:



The capacitor in question is the large black cylinder to the bottom left of the image. It's easy to spot a filter capacitor as they're generally quite large and will have a relatively high voltage printed on the side, in my case 400V. These capacitors can hold there voltage for some time after the power supply has been switched off so it is essential that before you start prodding around or handling the board you should discharge it!!

The best and cheapest way that I have seen to discharge these without damaging the rest of the circuitry is to connect the terminals across a mains rated 60-100W filament light bulb for about 5-10 seconds. NOTE THAT ENERGY SAVING BULBS WILL NOT WORK FOR THIS METHOD. The high resistance of the filament allows the capacitor to safely discharge quickly.
Always remember to follow this up with a quick voltage check with a mutlimeter (or if you're feeling bold, short the terminals with a well insulated screw driver). Here's the bit of kit I use to do this (less the multimeter):



By having the plug socket connected to the probe wires I can simply plug in my bedside lamp when I want to use it and wrap the wires around it and store it easily when I don't.
« Last Edit: March 05, 2013, 12:47:12 PM by Aerendraca » Logged
GwenLP
Full Member
***
Posts: 89


View Profile
« Reply #6 on: September 23, 2012, 07:56:53 PM »

Hi Aerendraca.

Thanks for the clarification ! Indeed NJY's power supply is a remote brick. It didn't think you were thinking about buit-in power supplies. Power supplies is an area that I'd like to know more (as is Electronics in general in fact) !

Good luck with your build ! It seems you are not afraid of tackling that PWM problem. I thought about going that route myself for a while, but it really is a step too far for me.
Logged
bernard
Administrator
Hero Member
*****
Posts: 2584


pato mania


View Profile
« Reply #7 on: September 24, 2012, 01:50:56 AM »

One of the ways I often see PWM drive circuits like LEDs is to use it as a hard "gate".  Often Led drivers will output at some very high frequency (for brigthness/power efficiency) and on top of that, the PWM signal just controls when to shut off (and put back on) the main output. The LEDs will blink indeed but at a rate that your eyes is not supposed to notice.

Just take a look at the datasheet of LED driver chipsets - Often you see how the PWM input acts on the output.  If I ever find back one of those, I'll copy the link.
Logged
Aerendraca
Administrator
Hero Member
*****
Posts: 1057


View Profile
« Reply #8 on: September 24, 2012, 09:19:28 AM »

Hi bernard,

Sounds like you've spent some time looking into this yourself, any joy?

I've been looking into possible LED driver chips that fit the criteria and have got most excited by the CAT4101 chip (Datasheet below):
http://www.onsemi.com/pub_link/Collateral/CAT4101-D.PDF

Only thing is - as I think you may have posted elsewhere on the forum - I'm not sure if the frequency and duty cycles of the PWM output for driving the CCFLs is the same or similar to that required by the LEDs, or if this is standardized across the industry. This is my current state of research.

Has anybody else had a play around with driver chips for LEDs?
« Last Edit: September 25, 2012, 05:27:50 PM by Aerendraca » Logged
Aerendraca
Administrator
Hero Member
*****
Posts: 1057


View Profile
« Reply #9 on: September 24, 2012, 09:22:49 PM »

Well today I've learnt alot about how PWM works in TFT screens, the benifits of CCFL over LED when dealing with PWM, and some useful information regarding frequencies.

First it seems that there is no 'standard' regarding CCFL PWM frequencies, but, TFTCentral states the following:
Quote
So how fast is PWM cycling backlights on and off? This seems to depend on the backlight type used, with CCFL-based backlights nearly all cycling at 175Hz or 175 times per second. LED backlights have been reported running from 90Hz up to 420Hz, with those at the lower end flickering much more visibly.

175Hz is a good frequency but not a great one for LEDs although I think still workable for this task, plus it falls in line with the LED driver I like the look of. The reason that this frequency is ok for CCFLs is that they have a turn off time i.e they dim naturally when switched off, this essentially means that the transition from on to off to on again is smoothed. LEDs on the other hand switch of virtually immediately creating more of a strobing effect and are possibly more uncomfortable for the viewer. See http://www.squidoo.com/led-backlight-flicker for more on this.

Not put off by this I thought I would have a go at building a simple PWM circuit to control the brightness of a few LEDs - proof of concept and furthering my knowledge hands-on style. The circuit was acquired from this video: http://www.youtube.com/watch?v=YmPziPfaByw. I would urge anyone wanting to learn a little more about the principle to at least watch the video.

Anyway, following building the circuit and powering it up I decided to take a look at the PWM output on an old oscilloscope I had kicking around to establish what frequency the circuit produced. Here's a little vid of it in action (sorry for poor quality, filmed on my phone):



The frequency that this little circuit produced wavers slightly from about 200Hz at a pulse width of 10% of the period to about 250Hz when the pulse width is virtually 100% of the period. I couldn't consciously perceive the flicker at all. This little experiment inspired me to take a look at the output PWM of my TFTs controller board, so I hacked a little circuit with a simple 5V regulator and once again plugged in my oscilloscope. Here's the vid:



The multimeter on the left is measuring the total current consumed by the regulator and the controller circuit (in Amps), the multimeter on the right is monitoring the voltage of the power supply (in Volts), and the oscilloscope is showing the PWM output of the controller board. Once again I apologize for the poor quality of the video, I really didn't have anything better than my phone to film on at the time. Maybe I'll see if I can find a better camera for future filming.

The video shows the output signal going high, dropping low, and then pulsing. Although not shown all that clearly you can just about make out that the pulse is probably something like 90% of the period. Unfortunately the PWM output disappears too quickly for the oscilloscope to be set, and then the circuit appears to cycle (watch the current bounce from 0.3A to 0.6A) - I imagine either because I didn't have the actual TFT panel attached, because there was no VGA/DVI input, or both. I will repeat this with both of these issues addressed when I get some more time to play.

P.S. I still like the look of the CAT4101 chip, but I can't find anywhere that has one that doesn't charge ridiculous amounts for P&P. Maybe I'll just suck up the cost and go for it...........maybe.
« Last Edit: September 25, 2012, 05:28:24 PM by Aerendraca » Logged
bernard
Administrator
Hero Member
*****
Posts: 2584


pato mania


View Profile
« Reply #10 on: September 25, 2012, 02:22:38 AM »

You couldn't guess at the period (frequency) coming out of your controller?

Can't you just set your scope to "normal" (not "auto") and play with the trig level until it catches the PWM pulse and "halts" to show one complete PWM period?

Blinking a tiny surface (a few LEDs) is totally different than blinking an entire screen that spans a big portion of the peripheral vision. You can perceive the flicker much more on a screen.

In terms of frequency 175Hz I think is fine.  I would say that passed 85Hz is probably good enough for a lot of people (some people might perceive a slight flicker). So anything under 100hz is indeed risky.

Logged
Aerendraca
Administrator
Hero Member
*****
Posts: 1057


View Profile
« Reply #11 on: September 25, 2012, 06:59:46 PM »

Believe me I tried it both ways but I just couldn't hone in on a stable display, at least not stable enough to calculate the frequency.

So today I decided I could spare some time to try again, but this time with the TFT panel plugged in and a VGA source to prevent the screen cycling. Bingo! This was the result:



Oscilloscope setup:

Horizontal   -   10ms per division
Vertical   -   2V per division

Initial pulse seen is when brightness is set lowest on OSD.

DOH!!! This is not the 175Hz I was expecting, in fact this is significantly lower than 175Hz. The frequency of the PWM signal is in fact between 20Hz and 25Hz, yes you heard me right 20-25Hz! - See the correction below.

This spanner in the works is going to slow progress on solving this problem but has not dented my determination. I WILL WORK THIS OUT.
I guess the next step is to write some code for a microcontroller chip - probably a Pic 16f628 since I have one kicking around - use the on chip comparator to interpret the PWM signal coming in from the controller board, and produce a new higher frequency PWM signal which can be understood by the CAT4101 Driver. Hmmm I make that sound easy, one problem, I'm a novice at programming microcontrollers. I feel this may be a steep learning curve.

Despite this not going the way I would have liked there are some benefits that this hiccup has presented.
If I need to produce a new PWM frequency then I can make it a high frequency say 300Hz, and then there's the ability to be able to produce my own dimming curve - no need to rely on linearity.

I'm starting to see now why I can't find any details of anyone that might have done this already.

I thought I should just say I really appreciate your input Bernard, it always good to have someone tether your feet down so you don't end up endlessly floating around looking for answers in the wrong places.  

____________________________________________________________________________________________________________
Correction:

I have recently noticed that I had the Pull of my oscilloscope set to X5; can be seen in the video above (little red light on the right hand side). This subsequently means that me maths for the frequency of the PWM was incorrect, a reworking is shown below:

working backwards:
frequency above = 25Hz
1/(25 x 10x10^-3) = 4divisions

now allowing for the X5 mistake:
4/5 x (10x10^-3) = the correct period = 8x10^-3s

frequency = 1/period = 1/(8x10^-3) = 125Hz Which is alot more like it!!!
« Last Edit: January 16, 2013, 10:44:44 AM by Aerendraca » Logged
bernard
Administrator
Hero Member
*****
Posts: 2584


pato mania


View Profile
« Reply #12 on: September 25, 2012, 07:11:41 PM »

i can assist you with the programming part. yeah that pic has an input capture but it is shared with the pwm output. you would have to do your own pwm ... possible but not as clean output signal. you have nothing else around?
Logged
Aerendraca
Administrator
Hero Member
*****
Posts: 1057


View Profile
« Reply #13 on: September 25, 2012, 09:31:36 PM »

Well they're not too expensive so I could buy another. I'd prefer to use Pic chips as I have a PicKit 2 programmer; how about the 16F876? Do you think it would possible with this?

Logged
bernard
Administrator
Hero Member
*****
Posts: 2584


pato mania


View Profile
« Reply #14 on: September 26, 2012, 12:41:03 AM »

yup that cpu has 2 independent CCP module. 1 for the input and the other for the PWM.
Logged
Pages: [1] 2 3 ... 15
  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!