[Kryten]

IIRC, the 'M' signal alternates (with every frame?) to reverse the voltage
across pixels to prevent a DC bias damaging the pixels (by some electrolytic
effect?).

I can see this works for a static image, but what would happen if you had
the screen being inverted at exactly the same rate?

Would this result in a net DC bias and ensuing damage?

Or does M spend 50% of the time in each state for each line of pixels?

A mate of mine reports 'screen burn' on some LCD monitors at work.
I thought that only happened in older CRTs.
Perhaps 'M' wasn't exactly 50% duty cycle for those LCDs.

K.

 

[John Devereux]

IIRC, the 'M' signal alternates (with every frame?) to reverse the voltage
across pixels to prevent a DC bias damaging the pixels (by some electrolytic
effect?).

I can see this works for a static image, but what would happen if you had
the screen being inverted at exactly the same rate?

Would this result in a net DC bias and ensuing damage?

Or does M spend 50% of the time in each state for each line of pixels?

A mate of mine reports 'screen burn' on some LCD monitors at work.
I thought that only happened in older CRTs.
Perhaps 'M' wasn't exactly 50% duty cycle for those LCDs.

K.

Interesting...

But I don't think this is possible because the image "pixels" cannot
be updated faster than the frame rate, while the M signal is typically
faster than this (to avoid flicker).

For example, on some QVGA(320x240) STN LCDs I am working with, the M
signal is the "line sync" (LP) pulse divided by 16.

 

[Kryten]

But I don't think this is possible because the image "pixels" cannot
be updated faster than the frame rate, while the M signal is typically
faster than this (to avoid flicker).

For example, on some QVGA(320x240) STN LCDs I am working with, the M
signal is the "line sync" (LP) pulse divided by 16.

So over 32 lines, M would be 50% high, 50% low.
Hmm, 240/16=15 so M would alternate between frames also.

Would I be right to think this is only going to guarantee 50% duty cycle if
either

a. Each frame

or

b. Each group of 32 lines

is driven twice with the same data.

Each pixel will only experience the alternating M signal over two frames.

I suspect that LCDs drive one line at a time (or two for dual-scan).

Therefore M ought to alternate over the course of each line.

Looking at one LCD data sheet, it shows an internal M generator clocked by
FLM and LP, which suggests M alternates slower than the line rate.


It is possible to flip all pixels at the frame rate (e.g. set
start-of-screen pointer to an all-black image or an all-white image). This
would undo the effect of M.

I'd be curious to see if this can damage a screen, though not so much as to
try it with my expensive new laptop!

 

[Kryten]

But I don't think this is possible because the image "pixels" cannot
be updated faster than the frame rate, while the M signal is typically
faster than this (to avoid flicker).

From:
http://www.cs.ucla.edu/~kohler/class/04f-aos/ref/hardware/vgadoc/CHIPS.TXT

"3d6h index 5Eh (R/W): ACDCLK Control Register (82c455/6/7, 655x0
only)
bit 0-6 ACDCLK Count. Number of Hsync pulses between changes in ACDCLK.
7 If set the ACDCLK phase inverts every frame, if clear the ACDCLK
changes phase when the number of Hsync pulses specified in
bits 0-6 have elapsed."

This indicates that M (aka ACDCLK) alternates at half the frame rate at
most, and one 64th the frame rate at least.

For example, on some QVGA(320x240) STN LCDs I am working with, the M
signal is the "line sync" (LP) pulse divided by 16.

This only has the effect of alternating between odd and even frames.

They only used 16 as a divisor because the number of lines divides to an odd
number,
and it only needs a 4-bit counter.

240 divided by 2, 4 or 8 yields an even number, so they could not use those.

All they really needed was to halve the FLM rate.