t said:
Thanks, so does the LED backlight use a regular inverter instead of
high voltage inverter?
How do the dual GPUs matter here? I estimate it starts with the Intel
GPU and when it detects intense graphical work(like game playing), it
would switch from ATI GPU. Or, do these models have a different way of
choosing which GPU to use?
Are there any signs of faulty connection I need to look out for? There
is no flickering and I know flickering means the backlight is going to
malfunction for a CCFL lamp.
Thanks for your advice and time.
There won't be an inverter, in the same sense as the 1000VAC one used
for CCFLs. There are still power circuits, but their design is
different (and likely, more reliable than an inverter).
There is not even a need for it to flicker. Since LEDs have
no "ignite" or "burn" phase, no plasma or gases, there is nothing
to flicker. Only a cold solder joint would flicker, and since the
panel actually has four LED chains, only a quarter of the
panel would flicker, if the solder joint on one LED was bad.
You can start with Ebay, if regular diagnostic sites aren't helping.
Here, I see a picture of the panel. Jackpot! Got a part number.
http://www.ebay.com/itm/GENUINE-SAM...Laptop_Screens_LCD_Panels&hash=item4ac5db370f
LTN156AT05
That gets me a hit here.
http://www.samsung.com/us/business/oem-solutions/pdfs/PSG2011_web.pdf
Of course, the Samsung site is a waste of time.
I found a datasheet for something close (glare finish rather than
matte), here.
So this should be pretty close.
http://lcd-screen.com.ua/data/pdf/1168.pdf
LTN156AT05 is a color active matrix TFT
15.6"
1366 x 768 pixels
display up to 262,144 colors [means 6 bit dithered, 64*64*64 colors]
LED Back Light with embedded LED Driver
The panel electronics (like, the LVDS interface) runs at 3.3V.
The LED supply is listed as 6V to 20V. That is boosted enough,
to drive eleven LEDs in series (3.2V x 11 = 35.2V). So a switching
power supply of some sort, converts VBL+ into a 35.2V source. The
LED string runs at 20mA. Probably a 20mA current source, intended to handle
a compliance voltage of 35 to 37 volts (as listed on page 12).
3.3V ---> Run LVDS interface
6-20V ---> Powers LED power supply ---> 35.2V, PWM modulated at LED
string
Another page is hinting, the LEDs are arranged in four channels.
Implying a total of 44 LEDs. 35.2V * 0.02mA = 0.7W or about 50 lumens
(estimated) per chain. Roughly the equivalent of a 20W lightbulb, in
terms of total light output from 44 LEDS.
The PWM control signal on the cable, is a 5V logic level. It is
intended to operate between 200Hz and 1000Hz (above flicker perception
range). The control range is listed as 10% to 100%, meaning PWM doesn't
go to zero on its own. There is also an Enable signal.
The pinout is on page 14.
PWM control is on pin 35.
Enable control is on pin 36.
To light the panel, Pin 36 must have a voltage
above 2.0V on it. A logic low (<0.5V), turns off the
backlight completely.
With Pin 36 logic high, the PWM Pin 35 control takes over.
The PWM must also be logic high. And the percentage
of the time the 200Hz signal is high, determines the
intensity of the LEDs. For example, if it did this, the
backlight would run at about 70% or so. As the signal
is high, 70% of the time.
2.0V +----+ +----+ +----+ ~200Hz
| | | | |
<0.5 -+ +-+ +-+
Page 20 of the data sheet, covers sequencing. You're supposed
to apply panel power (3.3V) before LED power (6-20V). And
the pin 36 Enable is used to gate the PWM for some reason.
They mumble something about latchup (panel stops working),
if the sequence is not obeyed. From one perspective, there's
no reason for the LED supply, to have anything to do with
the LVDS end. It implies that perhaps, the same boosted ~35V
internal supply, might also be used to run the LCD panel
horizontal and vertical driver chips.
In any case, you have a delightfully contradictory document
to work with. (I.e. Page 25 warns of "fluorescent lamp", when
the stupid panel is LED based. Hahaha. Copy-paste-twits)
Yes, there is a slight shock hazard, as the LED chain runs
at 35V, so don't run your bathwater, and reach out and touch it
Who knows, the actual LCD panel driver pins, may have more
voltage than that, but they're not required to tell you
about those. As they're not an "interface" issue. Purely
an internal design issue.
You can see in the Ebay photo, there are places to probe with
your multimeter, if you're bored.
Have fun,
Paul