Once upon a time on usenet Paul wrote:
~misfit~ wrote:
Once upon a time on usenet ~misfit~ wrote:
Once upon a time on usenet Paul wrote:
~misfit~ wrote:
Has anyone done this? My beloved Dell 2007FP (20" 4:3 UXGA
[1600x1200] IPS) monitor is getting dim. I've been having to use
it on full
brightness for a while now and even then it's not bright and
takes a while to reach
full brightness, maybe 15 seconds.

I'm not rich but could swing a few hundred for a new monitor if I
absolutely had to. However I've been spoiled with this monitor and
nothing that
I can afford even comes close to being as good. I've got a
multimeter, soldering iron and more than half a brain and would
much prefer to keep this
montior out of a landfill.

I've seen a few kits for sale to do the job (this is a cheap one
http://www.aliexpress.com/item/15-23...252220243.html
) but would feel better knowing that someone else here had
successfully done this mod. Also I'm concerned that, as IPS
monitors don't transmit as
much light as TN monitors I might need a brighter than usual
backlight. Thoughts?
I'd stick with replacing with "like" technology.
The trouble with that is CCFL is pretty ancient tech which hasn't
been used in new monitors for a while now. Add to that the fact that
I'm a bit of an 'LED evangelist', an early adopter who has been
running all LED home lighting and have converted my car to all
(except headlights) LEDs. Also CCFLs have a finite lifespan which is
very short when compared with LEDs and it would be nice to fix this
problem once only. I'd like to be able to run these monitors for a
long time yet (I have two of them, my spare isn't quite so dim). So
I figured replace the backlights with LEDs and while I'm in there
check out the condition of the electrolytic capacitors.
But.....

And that's assuming there is actually an inverter or CCFL problem,
and not a problem with the PWM control signal or the power supply.

Odds are, it is an inverter/CCFL problem, with inverters being the
highest probability of failure.

The panel uses multiple CCFL tubes and multiple
inverters. A big panel might use four inverters and
sixteen tubes for example. Now, consider the failure
possibilities. If one CCFL tube turns brown before
the others, you'd notice a non-uniform display. Or,
if one inverter failed, a quadrant of the screen
would go dim. Given such a situation, is the
entire panel going dim ? Would all the inverters
"go on vacation" at the same time ? Unlikely.

Common points of failure, would be the PWM signal
(pulse width modulation, 200Hz, high for 66 cycles,
low for 134 cycles would be 33% intensity setting).
Or, perhaps the 12V power source feeding the inverters
is out of spec. If that was the case, maybe other
circuitry would suffer as well. There have been
monitors with poorly design supplies, power supplies
with leaking electrolytic capacitors. It would be
a shame to waste $100 on LED kits, only to find
the problem was a bad power supply.
Hmmm. The brightness still works, albeit only going from dim to
dimmer so I hope the PWM circuitry is still fine. It was always my
intent to check the PSU when I have it open. I do have the backup
monitor on my second desktop computer but that's used at least
weekly by a friend who visits for gaming session so I'd like to keep
downtime to less than that.
I didn't realise there could be so many tubes. That's what you get
for estimating I suppose. My 15" UXGA IPS laptop screen has a
single tube so, knowing that desktop monitors often have tubes on
two sides of the screen I figured two tubes or perhaps two pairs of
tubes as it's not *that* much bigger than my laptop screen and
they're both UXGA.
I heartily recommend repairing things, where you
can "bound" the repair cost, and see ways of
preventing the repair bill from shooting too high.
Obviously, it's a personal judgment call, as to
how far to push this. If the monitor cost $1000,
I might consider making two repair attempts (CCFL
first, LEDs later if the CCFL didn't work out
for some reason). If the monitor cost $300, I'd
probably just bin it.

Those LED kits don't seem to have a diffuser in
evidence. How will the "ripple" in light intensity
from all those point sources be handled ? Will
the panel look like a "cheese grater" when displaying
an all-white background ?
I wondered about that and thought that, if there's room I'd put two
rows of LEDs in each CCFL 'gully', staggered a few millimeters so
that the light is uniform, even arranging them in a V shape if it's
tight....
The very last thing you want, is a repair job
where you're "reminded" every time you look
at the monitor, that you've fouled things up :-)
I know that feeling (the less than perfect repair).
How very true, I've done that before myself also.

Thanks Paul, you've given me some things to think about. I guess
next step is to swap the brighter one to my desk and open the other
one up and have a look-see. Fingers crossed that opening them isn't
too hard! Then again I could always procrastinate longer, keep turning
the
lights out and drawing the curtain when I want to game. (I only use
my desktops for gaming, the rest of my computing needs are amply met
by my well-specced T60 ThinkPad.)

Well I've found more info on these excellent monitors online since
replying. Apparently there are *six* CCFLs, three at the top and
three at the bottom. Also apparently the most likely cause of what
I'm seeing is a big cap in stage one of the PSU - a 180uF / 450v
electrolytic that's the most common weak point in these monitors.

A guy who replaced the cap in his said it looks like they'd be a good
candidate for an LED conversion but he was unable to isolate the
soft menu OSD brightness signal and needed to get his back together
so didn't have time to do it then. I wonder if I could use three of
those LED kits (staggering emmiters so they overlap) and add a
physical PWM module for brightness and mount it somehow? Something
like http://www.dx.com/p/379683 perhaps?

Unfortunately I couldn't find the lead pitch and available space for
a replacement cap specified anywhere (or pictures to work that out)
so I could order one before pulling the monitor down. :-/ It seems
they're a very tightly joined screwless case that is a chalenge to
anyone opening them. I hope my plastic hasn't got brittle.


Sorry for not replying Paul, this project is on hold for now but I've savd
everything you've contributed, thanks.

That cap with the 450V rating, is probably the primary
side main capacitor. That would be roughly equivalent
to C5 or C6 in this schematic. I've never ever had
a primary cap of the C5/C6 type fail. If you work out
the number of joules some of these capacitors
hold, you'll understand why it's important they
don't fail.

http://www.pavouk.org/hw/en_atxps.html

Yeah, odd I know. However I read a lengthy account of how someone had
increased the brightness of their backlights and they started to flicker on
and off constantly. When the case was opened that capacitor was leaky and
when replaced the problem went away.

Also I had a primary side capacitor blow its top in an Enermax PSU many
years ago. The PSU was used and had been on my shelf for a few months, I
fitted it into a PC case, hooked everything up and flicked the power switch
on the back of the PSU and it went off like a shotgun. It split the top of
the can and blew the black plastic disk that sat there through the slots in
the PSU case and right across the room. My ears were ringing for hours.

I'm more used to output caps failing, such as C30 or C27
on the right-hand edge of the schematic. If those were
failing, there might be more ripple on the output
voltage (not a steady DC).

I also am more familiar with output caps failing.

*******

To make a PWM signal, your monitor already does that,
and one of the signals on each inverter interface
cable, carries PWM. It could be a 5V logic signal
for example, or, it could be driven by an open collector
transistor on a logic board, and be pulled up to
+12V on the inverter end.

The PWM signal might vary between 30% on and 100% on.
The PWM signal might not be adjustable below 30%.
That's because currently, the monitor is using
CCFL tubes, and the CCFL tubes won't stay lit
if the duty cycle of the PWM signal is too low.

If the LEDs you've selected, are way too bright for
the job, now you might need the PWM to adjust below
30%, to 15% or 8%. The LEDs should be able to go
practically all the way to zero, with the right
selection of PWM design (i.e. PWM signal might
be derived from a different master clock
than the one you'd use for a CCFL).

So maybe the PWM signal currently used by the
monitor, looks like this.

+5V +------+ +------+ Approx
| | | | 200Hz square wave
GND ---+ +-------------+ +--------------

Maybe for a LED one, 200Hz would be too irritating.
Maybe 400Hz, at 8% duty cycle, to drop the intensity.

+5V +-+ +-+ +-+ +-+ Approx
| | | | | | | | 400Hz square wave
GND ---+ +--------+ +--------+ +--------+ +---

So what you're making, when generating a PWM
signal, is a pulse generator with adjustable duty
cycle. And if you make a good guess on the frequency,
maybe that part of it never needs to be adjusted.

To give the PWM "resolution", the master clock
must run at a much higher speed. Say you wanted
the adjustment knob to have 256 intensities. Then
the master clock would need to run 256 times the
final clock rate. So 256*400 = 102.4KHz. You might
be able to make a signal like that, with a 555 or
some other form of cheesy oscillator.

To make the waveform, you could use two eight-bit
counters. One counter simply counts to 256
over and over again. That defines the time from
one rising edge of the output waveform to the
next rising edge. The other counter, clocks
at 102.4KHz, but it uses preload terminals.
Say you load it to 253 decimal. It counts
253, 254, 255. When it hits 255, the Carry_Out
on the counter is asserted. You tie the
Carry_out terminal to the Disable input.
That stops the counter from changing state. It
stays at 255, until the other counter, the divide
by 256 gets to its full count, and a signal from
that one, trips the preload on the programmable
counter, and starts it at 253 again. You connect
an eight bit DIP switch to the preload terminals
of the counter, to input a number like 253
(for a low value of PWM), or maybe a value of 25
for high brightness. The DIP switch eight bit value,
then is your "user interface" for setting the
screen intensity.

You build discrete circuits like this, with jelly bean
logic, when retrofitting circuits in old gear.
A teenager building this circuit, would use a PIC
and he would be finished by now :-)

The above isn't very practical, and there are certainly
other ways to do it. For example, you might be
able to use a 4538 monostable, a couple of RC
circuits, and use potentiometers to make the
timimg of the waveform. And then one of the pots
becomes the intensity control. This is simpler
than the counter idea, and has as much resolution
as the potentiometer will allow.

Hobbyists love to build stuff like this :-)

The only way projects like this are practical,
is if you have another monitor to use, until
the duff one is fixed.

I do have two of these monitors that are identical which is part of the
reason I think it's worth working out how to run LED backlights. The second
monitor was on my guest machine and used much less often. However since I
swapped them in preparation to pulling the dim one down (and waiting for the
NZ$ to recover a bit) the brighter one (backlight set at 60%) has developed
'flickery lines' when used in game. That's not a backlight issue and is more
likely to be the low voltage smoothing caps maybe? The graphics card GPU and
RAM isn't O/Ced and was fine previously (although I know that doesn't rule
out it recently developing a problem).

Cheers,
--
Shaun.

"Humans will have advanced a long, long way when religious belief has a cozy
little classification in the DSM*."
David Melville (in r.a.s.f1)
(*Diagnostic and Statistical Manual of Mental Disorders)