A couple weeks ago, i won a job lot of 8 *non-working* LCD's on a "local" auction site, for 10 euros (~12-13 US$), and finally this week, i managed to source a ride to go pick'em up
First "victim": a Dell 1908FPc, 19" 5:4 aspect ratio, in really good shape, visually. Only a couple scrapes on the bezel.
Behaviour: 2 seconds to black.
It was one major b*tch to crack open, but i managed to get the back cover off. I was amazed - chock-full of Rubycon YXG and Chemicon KY, with a Chemicon KXG primary; that pretty much ruled out bad caps
Time to break out the test-CCFL (recovered from a smashed laptop LCD ). You can see its pink-and-white wires going off to the side, in the attached pic.
Verdict: one of the two lower CCFL's seems to be... "out of spec". Occasionally, the inverter seems to "pulse" the CCFL's around 40-50Hz i'd reckon, but i've kept the screen on for ~10mins now, and everything's still lit.
First i tried increasing the "start-up" cap for the inverter PWM (OZ9938; Ctimer in the datasheet, on page 7 and onwards), to try to delay the regulation and/or to give more time for the shoddy CCFL to light up or something. I soldered a 3.3uF cap on top of the existing one (0805 ceramics), but no change.
Now, i fixed a 19" HP screen a couple months ago, with a similar behaviour, and the fault ended up actually being the "cold" wire to one of the CCFL's - it had broken right where it was soldered to the tip of the lamp. Oddly enough, that had black insulation, same as this Dell. I think i'm gonna have to take the whole panel apart... Fun, fun, fun...
For now, i'm attaching a top photo of the whole ensemble. It includes a 4-port SMSC-powered USB 2.0 hub - that's what the narrow board on the left is, and it seems to have its own DC/DC converter, which is quite impressive (considering the main PSU provides a 5v rail "too").
First "victim": a Dell 1908FPc, 19" 5:4 aspect ratio, in really good shape, visually. Only a couple scrapes on the bezel.
Behaviour: 2 seconds to black.
It was one major b*tch to crack open, but i managed to get the back cover off. I was amazed - chock-full of Rubycon YXG and Chemicon KY, with a Chemicon KXG primary; that pretty much ruled out bad caps
Time to break out the test-CCFL (recovered from a smashed laptop LCD ). You can see its pink-and-white wires going off to the side, in the attached pic.
Verdict: one of the two lower CCFL's seems to be... "out of spec". Occasionally, the inverter seems to "pulse" the CCFL's around 40-50Hz i'd reckon, but i've kept the screen on for ~10mins now, and everything's still lit.
First i tried increasing the "start-up" cap for the inverter PWM (OZ9938; Ctimer in the datasheet, on page 7 and onwards), to try to delay the regulation and/or to give more time for the shoddy CCFL to light up or something. I soldered a 3.3uF cap on top of the existing one (0805 ceramics), but no change.
Now, i fixed a 19" HP screen a couple months ago, with a similar behaviour, and the fault ended up actually being the "cold" wire to one of the CCFL's - it had broken right where it was soldered to the tip of the lamp. Oddly enough, that had black insulation, same as this Dell. I think i'm gonna have to take the whole panel apart... Fun, fun, fun...
For now, i'm attaching a top photo of the whole ensemble. It includes a 4-port SMSC-powered USB 2.0 hub - that's what the narrow board on the left is, and it seems to have its own DC/DC converter, which is quite impressive (considering the main PSU provides a 5v rail "too").
Comment