Jump to content
Search In
  • More options...
Find results that contain...
Find results in...
lucky1

Color Bleeding Of Chinese Rgb Panels

Recommended Posts

GetAttachment.aspx.jpg

 

Some of you experienced color bleeding of the panels we use. Unfortunately the is caused by a design fault of the pcbs the panels

use. Cynic from the german flippermarkt forum analyzed the problem and found a solution for it. I translated his post

 


 

I have analyzed the color bleeding problem, and have found a fix for it (which also fixes the column problem with dead pixels) .

 

Brief background: The displays have 32 rows and 64 columns. Upper and lower halves have their own data lines (RGB0 / 1), which feed exactly one row with data. For this, the 64 columns are controlled by 12 16-bit shift register constant-current LED drivers (ICN 2026) (for each color a set of 4 pieces, each with an optimized current) These are open drain,which means they connect the current to ground. To light up a pixel the current is connect to ground  by the 2026 while at the same time the appropriate line is connect to supply voltage. This is done by the scan signal (4 bits that go to a decoder which then each up and down turns one line) and a total of 16 double-P-channel FETs (FDS 4953), which turn the 5V on or off.

 

Here is the problem that leads to bleeding: If only a few pixels of a column are lit, all the rows are turned on one after the other  (but the columns remain off)  and the conductor tracks of the lines get charged like a capacitor (the charge of course is very low ). The column driver lock this very well, there is virtually no leakage current.

If a pixel is driven now, the whole column is connected to ground and the charge of conductor tracks of the inactive lines can drain via the LEDs (which should be actually off). The pixels glow with light. This also explains why it does not come to a glow, if there is no pixel turned on in a column. Then there is never a connection to ground and the charge does not flow through the LEDs.

If a pixel is defective, the constant current driver of the defective color drains more power in less time. The column of the color lights shorter, but much brighter. With a defective blue diode, the column glows clearly visible blue.

 

Now to the solution: We want to avoid that the charge of inactive lines flows through the LEDs, so we have to let them drain otherwise. For this purpose, we just connect a resistor from each row to ground. I have soldered 1K or 2K2 0603 resistors directly to the drain of the FET's legs and pulled wire to ground on the other side.

In addition, there must be ensured that the scanning / Row signal at least a few microseconds before and after the blanking changes so that lines the display Enable are discharged ..

img_20160115_174808_3wbsui.jpg

 

It is not nice looking but works well and is the only way I can think of to solve the problem.

 

LG,

Jann

 

 

  • Upvote 2

Share this post


Link to post
Share on other sites

That is some nice finds, glad that I don't have that problem.... don't have time for that lolW

 

What will happen to the pindmdv3 when the paid customers/users have the bleeding problem is the fun and interesting part... just saying.

  • Upvote 2

Share this post


Link to post
Share on other sites

Hmm clever fix. I have noticed it but it hasn't bothered me too much. Your example photo is a real extreme example of it. The worst I've seen so far is this

 

kcHWrls.png

 

It's not too bright but it's all over. Mostly noticeable with white colors

Share this post


Link to post
Share on other sites

GetAttachment.aspx.jpg

 

Some of you experienced color bleeding of the panels we use. Unfortunately the is caused by a design fault of the pcbs the panels

use. Cynic from the german flippermarkt forum analyzed the problem and found a solution for it. I translated his post

 

Can we get more info in this fix?

My panels bleed a lot in bright colors.

Share this post


Link to post
Share on other sites

Yes! Also interested in this fix, all though I have not that kind of menstrual problem, I sometimes see some stains with the brighter colors.

A schematic would be neat

Cheers

Share this post


Link to post
Share on other sites

I noticed also some bleeding.Always with bright colors.

So i send a message to the supplier(Angela).

Here's what she say.

dear,you can set the software,to adjust the clock signal lower,please try it.

have any question,please do not heasitate to contact us,we will try my best to assist you.

best regards

angela

But i dont know what she means.

https://www.dropbox.com/s/nyxm25psw2isi1n/Bestand%2003-03-16%2009%2026%2002.jpeg?dl=0

Share this post


Link to post
Share on other sites

I assume she means make the picture less bright but that is not a solution.

Cynic has analyzed the problem deeply and it is definitely a design flaw

in the pcb layout which can only be really fixed by the mod he describes.

Funny thing is that afaik the chinese don´t give any specs about these panels.

You have to find out yourself, if you want to get them working.

Share this post


Link to post
Share on other sites

lucky1 please can you make a tutorial for this fix. I'm experiencing the same problem. Where are you grounding the wire to etc. thanks

Share this post


Link to post
Share on other sites

More details would be nice, but the soldering involved is probably not for novice users.  Here is a pic of the the size of the resisistor you are soldering and here is one of many places to ground the them.  

 

1459696004.png

  • Upvote 1

Share this post


Link to post
Share on other sites

I've looked at them. My soldering skills aren't bad. Some clear details of what needs soldering to what would be great. I'm interested in where the wire is grounded to as well.

 

My bad I know where to ground them lol.

Share this post


Link to post
Share on other sites

+1

Which resistors needed and some pics of a soldered DMD or some schematics would be great :)

Share this post


Link to post
Share on other sites

Pictures of a soldered DMD is in the first post. IMHO that´s pretty clear.

Share this post


Link to post
Share on other sites

I don't think it very clear on the first pic.

There is one wire strand in the middle of the picture and one to the right, but is it just those two wires?

 

Is it 12 chips to be soldered or more?

 

I will take a closer look when i am finished with the electronics of my build.

Share this post


Link to post
Share on other sites

Attached a pic of how i did this mod, it may be of help...

I used 0805 resistors as thats what i had on hand, these are a little bigger than 0603 ones but fit perfectly on the two drain pins of each  FET.. I chose the ground point at random, any point can be used to tie too as long as its ground..

 

 

post-10751-0-84762700-1459861376_thumb.jpg

  • Upvote 1

Share this post


Link to post
Share on other sites

Attached a pic of how i did this mod, it may be of help...

I used 0805 resistors as thats what i had on hand, these are a little bigger than 0603 ones but fit perfectly on the two drain pins of each  FET.. I chose the ground point at random, any point can be used to tie too as long as its ground..

 

 

attachicon.gifDMD bleeding mod.jpg

Thanks, this was really a lot clearer to see, but i wonder: on your pic it seems you solder the resistors on pins 5,6 and 7,8, The picture top show resistors soldered to pin 5 and 8 only.

 

Is it 16 chips total on each board?

And you used 1k resistors i see. (102) marking.

Share this post


Link to post
Share on other sites

Thanks, this was really a lot clearer to see, but i wonder: on your pic it seems you solder the resistors on pins 5,6 and 7,8, The picture top show resistors soldered to pin 5 and 8 only.

 

Is it 16 chips total on each board?

And you used 1k resistors i see. (102) marking.

Pins 5,6 are the same point and 7,8 are the same point ( connected internally inside the IC and also connected via the traces on the PCB) it was just neater to solder the resistor evenly over both pins but you can solder only to one pin like in the top post if you like also..

 

It is 16 ic's you need to do, these ic's contain 2 FET's inside the one package. So 16 ic's really is 32 FET's, one for each of the rows..

 

I used 1K 0805 resistors because thats what i had access too at the time, you can use either 1K or 2.2K as stated in the top post.. The size also can be whatever you like as long as it is going to fit in the limited space available so 0603, 0805 even normal 1/4 or 1/8th watt through hole resistors cut to size could be used if thats all you have. It is only limited to what can fit in the limited space..

 

Thanks for the Picture Dann. :P

 

I ordered some "SMD SMT 0603 Chip Resistors Surface Mount 2K2 2.2Kohm 222 +/-5% 1/10W RoHs". Will they work for this?

These would be fine..

  • Upvote 1

Share this post


Link to post
Share on other sites

Attached a pic of how i did this mod, it may be of help...

I used 0805 resistors as thats what i had on hand, these are a little bigger than 0603 ones but fit perfectly on the two drain pins of each FET.. I chose the ground point at random, any point can be used to tie too as long as its ground..

attachicon.gifDMD bleeding mod.jpg

Hi Dann, but between 1K or 2.2K there should be a difference ? in bright / speed or something else ? or is exactly the same ? If you have to choose now a value, would you choose 1K or 2.2K ? Thanks!

Share this post


Link to post
Share on other sites

Anyone have a good .ppm image in order to force bleed so I can compare with before/after.

Would be neat so everyone could post in this thread for others to see  :rolleyes:

Share this post


Link to post
Share on other sites

I think thin white lines bring it out more.  Try this and report back. I am not near my cab.  You just need to create a 128x32 image.  Photoshop can do it or there are online png to ppm converters also

 

1460747981.png

 

 

 

testimage.zip

 

 

 

Share this post


Link to post
Share on other sites

Thanks gtxjoe!  :lol:

 

Tried your pic with the latest FW 2.09 and with the highest intensity, did not get any bleeding(!) :D

Anyone else feel free to ship in here. Would be great to nail down a good ppm-file to lure it out

 

bleed_lines_test.jpg

Share this post


Link to post
Share on other sites

When I get home I can do some DMD rom screen capture and convert to ppm.

Do you have an example of a rom to try. Maybe when displaying older non DMD rom output...

Share this post


Link to post
Share on other sites

Thanks Joe!

Tested it and now I could see some bleeding, not that much with full intensity I must say. But the test1 image produced to most bleed what I could see.

Well noted if one would proceed with the mod to see the after results, great pictures gtxjoe :D

Here is my screens of the 4 pics, also did a quick bat modded from the 'dead pixels rgb test' to run them easy, also attatched

bleed_test2.jpg

PANEL test - BLEED2.rar

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×
×
  • Create New...