2024-09-27 - Replacing the CCFL Tubes in a Monitor With LED Strips
------------------------------------------------------------------

WARNING: The end result of this guide is not very professional and
may not be entirely safe to use. Follow at your own risk.

Many years ago, someone gave me an old LG Flatron W2242T monitor with
the somewhat common problem that the backlight would turn off after
a few seconds. On old monitors like this one, that often means that
at least one of the CCFL tubes is going bad. After leaving it in a
corner for a long time, I decided to try replacing the CCFL tubes
with LED strips. I mainly followed the video [0] since that was one
of the most in-depth guides I could find.

The first step was ordering an LED replacement kit. There are many
options, although most of them probably come from the same factory
anyways. The one I got had the model number GYD-9E[1]. I read later
that the DF6113 chip in use on most of these boards apparently isn't
too great for the purpose[2], but it seems to work well enough so
far. Luckily, mine was packed properly in a hard plastic tube[3],
so I didn't have broken LED strips like the ones shown in the video.

The driver board has four inputs, GND, VIN, ENA, and DIM. VIN is
supposed to work with anything between 10V and 30V, but many people
(including the author of the video) have mentioned that it's probably
not a good idea to run these driver boards with very high voltages
because they get very hot then. Through experimental testing, I
determined that ENA needs to be at least around 2V for the LEDs to turn
on. I'm assuming the maximum is 5V, but I don't know for sure. The DIM
pin can be set between 0V and 5V, where 0V is the brightest and 5V is
the dimmest. During testing, I found that there was a relatively large
region above 0V in which I couldn't determine any lower brightness,
and similarly a region below 5V in which I couldn't determine any
higher brightness. I didn't take any exact measurements, though.

One important point that the video also mentions is that it's
easy to destroy one of these driver boards by connecting it with
the wrong polarity. The problem is that the connectors on the two
ends of the cables are the same, but if you switch them around, the
red and black cables will be at opposite pins. Unfortunately, I had
forgotten this part of the video by the time I got around to playing
with the LED driver, so I promptly plugged it in the wrong way into my
power supply. This was after I had cut the cables apart already and
connected some of them to parts of the display. I wanted to perform
some more tests using my benchtop power supply before connecting
the rest of the cables, so I plugged in the unneeded connector and
was tricked by the switched red and black cables. The magic smoke
immediately started escaping from two resistors on the board, but
surprisingly, there was enough magic smoke left inside that they
still worked afterwards, despite clear burn marks[4]. I did replace
the 1 Ohm resistor, but I didn't have a 0.620 Ohm resistor handy,
so I just left that one as it is, after verifying that it still had
the correct resistance. Maybe I'll swap that one out later.

The most annoying part of the whole project was actually replacing
the CCFL tubes with the LED strips. I forgot to take pictures for
the first part of the disassembly process, but the plastic shell of
the monitor is just held together with clips, so once the front part
(on the side of the screen) has been taken off, the whole monitor
can be flipped over and the back plastic cover taken off, leaving
the display with the two circuit boards inside a metal cage on the
back. The metal cage is held in place by a metal part which can just
be removed, after which the cables going to the CCFL tubes can be
disconnected. While taking the cage off, the ribbon cable going to
the display also needs to be disconnected, then you should be left
with only the display panel[5].

WARNING: As always, be careful when you're near the power supply,
as the capacitors can still hold a charge even after it has been
unplugged. Please read up on safely discharging power supply capacitors
before attempting any repairs. Not heeding this advice may result in
serious injury or death.

Now, the protective cover over the circuit board must be
removed[6]. The metal frame on the outside can then be removed
carefully, making sure not to damage any of the ribbon cables going
to the circuit board[7]. The circuit board can then flipped outwards
very carefully[8] and the back metal plate removed. The metal plate is
held on by a lot of clips[9]. This will expose the backlight[10],
which can then be removed completely, including the diffuser[11][12].
The CCFL tubes are inside metal frames at the top and bottom of the
diffuser[13]. These metal frames were stuck on fairly tightly, but it
was possible to take them off. In order to take out the tubes, it is
easiest to just cut the cables going along the top of the frame[14]. I
managed to break one tube because I used a bit too much force before
I realized that it made more sense to just cut the cables. If you
do manage to break any tubes, make sure to air out the room well and
clean up any glass pieces because CCFL tubes are quite toxic.

The LED strips can be cut if needed, as long as the LEDs are cut off
in groups of three. The driver board also mentions a minimum length,
but I'm not sure what the exact reason is (maybe a minimum load for
the voltage to be stable?). Luckily, my monitor was exactly the right
size for the strips, so I didn't need to cut them at all. I decided
to stick the LED strips in the middle of the frame using double-sided
tape[15]. It probably would have been better to first flatten the
backside of the frame a bit where the cables originally were in order
to have a more secure place for the strips, but I was too lazy for
that (there were originally two tubes on each side of the diffuser,
so one tube in each of the grooves). It is important to make sure that
there aren't any conductive parts on the LED strip that could
accidentally touch the metal frame and short something out. The
resulting assembly didn't fit as well as the original because the
rubber stoppers on the sides were missing[16], but I decided to leave
it like that. I initially worried that the heavy diffuser plate would
damage the LEDs, but that is held in place securely by the frame[17],
so it shouldn't be a problem.

Once all that is done, the entire display can be reassembled. Make sure
to clean everything before putting the diffuser back in so that there
aren't any specks in the picture later. Some of the clips were a bit
difficult to clip back in without using too much force and potentially
damaging the delicate ribbon cables[18]. Finally, everything should
look just like before, but with different cables coming out[19].

Now all that remains is all the wiring together of the different
boards. As mentioned before, it probably isn't a good idea to run
the LED driver board at higher voltages, but the power supply in this
monitor has 22V as the only option[20], so I followed the advice in
the video and got a cheap buck converter[21] (based on the LM2596S)
to reduce that. I set it to output around 12V when connected to the
22V input, but higher voltages should still be fine. I think I saw a
post somewhere in which someone tested at which voltages the LED
driver becomes too hot, but I just stayed with 12V becuase that seemed
to be fine. One thing to note is that the buck converter now gets
somewhat hot, so it might make sense to play around with the voltage a
bit again and/or get a better buck converter at some point. As the
video mentions, it might also be a good idea to replace the cheap
capacitors, but I decided to leave them for now.

The ON pin on the power supply connector gives a voltage that is
sufficient to turn the LED driver on (I think it was between 4V and
5V, but I don't remember exactly), so this can be connected to the
ENA pin on the LED driver board. The cable can be removed completely
from the connector by lifting the little plastic clip while pulling
on the cable[22]. One useful tip mentioned in the video is that it is
a good idea to connect the ON pin on the power supply board directly
to GND after removing the cable to make sure that the inverter always
stays off. Some people have mentioned that it might be good to just
desolder the power MOSFETs or other components so it's impossible
for the inverter to turn on[23], but I left it this way for now. The
final wiring on the power supply can be seen in [24], with the small
white cable connecting ON to GND, and the long red, white, and blue
cables being connected to 22V, GND, and 5V, respectively.

There is also a DIM pin on the power supply. This is probably some sort
of PWM signal, but I decided to just set the brightness at a constant
value using a voltage divider for now. The complete wiring can be
seen in [25]. The buck converter and LED driver board are only stuck
on using double-sided foam tape at the moment. For a more long-term
installation, they should be attached more securely so they can't
fall off as easily (possibly shorting something out in the process and
causing nasty things to happen - the driver board in particular is
slightly worrying because it could fall onto the power supply board).
As can be seen, the 22V and ground cables from the power supply go to
the buck converter, then the 12V and ground connections on the other
side go to the LED driver board. The cable that originally went to the
ON pin on the power supply is now connected directly to the ENA pin on
the driver board. There is another ground cable that just ends in
nothing at the moment. I added that for the voltage divider, but then
I realized that the DIM pin on the driver board already has about
51.5 kOhms resistance to ground, so I just added a 100 kOhm resistor
between 5V and the DIM pin, which leads to about 1.7V on the DIM pin.
This seemed to generate a fairly decent brightness, so I left it like
that. I was too lazy to remove the extra ground cable because the buck
converter was already stuck on at that point, so that's the way it is
now. Lastly, the cables going to the LED strips of course need to be
plugged into the driver board.

UPDATE (2024-09-28): I forgot to mention that one reason to reduce the
brightness a bit (other than it possibly being more comfortable for the
eyes) is that the LEDs might last longer then, or at least that's what
I've heard.

Reassembly isn't particularly difficult. The metal cage needs to be
flipped around again and the ribbon cable reconnected. The metal piece
to hold the cage in place can then be reattached[26]. Then, the back
cover can be placed on top while making sure that the buttons for the
front panel and the cables connecting them to the control panel go in
the right place. After that, the whole monitor can be flipped around
and the front frame reattached[27]. Two buttons on mine are broken,
so I have so use a toothpick or something similar to press them.

Finally, the monitor can be tested[28]. Everything seems to be working
fine, it would just be nice to have working brightness control. It
might be a good idea to just add a potentiometer on the outside instead
of trying to use the PWM signal. One of my biggest gripes with most
monitors is that it takes so many clicks to change the brightness,
so I would actually consider that to be a huge usability upgrade.

Anyways, that concludes my report on replacing the CCFL tubes with LED
strips. Now I just need to figure out what to do with this monitor.
You can get much more modern monitors for free nowadays, yet I decided
to fix this old thing. Oh well.

[0] https://youtube.com/watch?v=nN2NpHR87lk
[1] gopher://lumidify.org/I/phlog/2024-09-27-monitor/led_driver_board.jpg
[2] https://www.eevblog.com/forum/projects/which-smt-white-led-for-ccfl-backlight-replacement/msg1473689/#msg1473689
[3] gopher://lumidify.org/I/phlog/2024-09-27-monitor/led_kit.jpg
[4] gopher://lumidify.org/I/phlog/2024-09-27-monitor/led_driver_board_burned.jpg
[5] gopher://lumidify.org/I/phlog/2024-09-27-monitor/display_back.jpg
[6] gopher://lumidify.org/I/phlog/2024-09-27-monitor/display_board.jpg
[7] gopher://lumidify.org/I/phlog/2024-09-27-monitor/display_frame_clips.jpg
[8] gopher://lumidify.org/I/phlog/2024-09-27-monitor/display_board_flipped.jpg
[9] gopher://lumidify.org/I/phlog/2024-09-27-monitor/display_backplate_clips.jpg
[10] gopher://lumidify.org/I/phlog/2024-09-27-monitor/display_back_removed.jpg
[11] gopher://lumidify.org/I/phlog/2024-09-27-monitor/display_without_backlight.jpg
[12] gopher://lumidify.org/I/phlog/2024-09-27-monitor/display_backlight.jpg
[13] gopher://lumidify.org/I/phlog/2024-09-27-monitor/display_ccfl.jpg
[14] gopher://lumidify.org/I/phlog/2024-09-27-monitor/display_removing_tubes.jpg
[15] gopher://lumidify.org/I/phlog/2024-09-27-monitor/display_led.jpg
[16] gopher://lumidify.org/I/phlog/2024-09-27-monitor/display_led_inserted.jpg
[17] gopher://lumidify.org/I/phlog/2024-09-27-monitor/display_diffuser_position.jpg
[18] gopher://lumidify.org/I/phlog/2024-09-27-monitor/display_reassembly_clips.jpg
[19] gopher://lumidify.org/I/phlog/2024-09-27-monitor/display_reassembled.jpg
[20] gopher://lumidify.org/I/phlog/2024-09-27-monitor/power_supply_connector.jpg
[21] gopher://lumidify.org/I/phlog/2024-09-27-monitor/buck_converter.jpg
[22] gopher://lumidify.org/I/phlog/2024-09-27-monitor/power_supply_on_pin.jpg
[23] https://www.badcaps.net/forum/troubleshooting-hardware-devices-and-electronics-theory/troubleshooting-computer-displays/41160-update-to-ccfl-led-conversion-wiring-question?p=835104#post835104
[24] gopher://lumidify.org/I/phlog/2024-09-27-monitor/power_supply_wiring.jpg
[25] gopher://lumidify.org/I/phlog/2024-09-27-monitor/complete_wiring.jpg
[26] gopher://lumidify.org/I/phlog/2024-09-27-monitor/back_reassembly.jpg
[27] gopher://lumidify.org/I/phlog/2024-09-27-monitor/fully_reassembled.jpg
[28] gopher://lumidify.org/I/phlog/2024-09-27-monitor/final_test.jpg