Lenovo Thinkpad R400 LED Display backlight fix (#P3)
Yes, it’s running in the “projects” category, as “quick and dirty weekend fixes” isn’t available yet
While I was hosting my brother, he casually asked me when I will fix his display. Umm…what?
Turns out, his Lenovo R400 was showing vertical stripes again. Like these:
(jup, stolen from some Apple customer help page –
I didn’t take a photo at the time. I guess the bug is a watermark, isn’t it?)
Update from 23rd of December in 2020: By pure luck, I found this one from the previous issue while shovelling old photos and other data onto my server. Well, looks the same to me…
I’ve already seen and kind-of fixed that on my visit shortly before Christmas last year, when I basically had no tools available. It was a contact problem somewhere in the display, unlike the Apple problem, which actually is a software induced problem. So some chopped-off ends of toothpicks did the trick and it was all fine and dandy. But my brother failed to tell me that it has gone bad again, and while I had this display on an eBay search alert, I never scored an extra cheap one. So – I had no parts available. But I had a look…and broke it. Classic!
How to not provoke anger due to a back
lightdarked display for two days until the replacement arrives? Well, fixing pesky flat flex cables is beyond reach. But a crude botch will do!
So here’s the LED board from the thing. I only took some photos as I went along, this wasn’t intended for documenting my
botching soldering skills or anything like that, but I found it worth sharing.
It’s running on the bottom side of the display unit in full width and does the entire illumination with the help of some three clever foils behind the light-spreading plastic back plate. It’s quite thin, just shy of entirely flexible, as you can see from the image. The LEDs are lengthy SMD types with 90° emission, so although the PCB is mounted horizontally, the LED will shine upwards nicely.
(Clicking the thumbnails does help in this case!)
So the silkscreen indicates a weird pattern that starts like LED1 – LED9 – LED17 – LED25 – LED33 – LED41 – LED2. Fair enough, after testing each individual LED, this helps a lot with testing the entire chains. As it’s 8 LEDs per row, my 30V lab supply doesn’t cut it, but going shorter chains reveals that everything is working. No dead LEDs that would kill an entire row, no cracks in the PCB.
As the large plane on the right indicates, everything is joined together on one side. The other side however is individually wired to traces on the flat flex, which nicely fits the problem – some of these traces do not have contact with the PCB, so some of the LED chains do not light up. And as they are distributed in said order, the display has some weird yet regular light stripes that fade out from bottom to top.
Here’s the connector in better detail – have a closer look at the end, which shows some missing bits of some traces (the joined ones on the left just ripped off when removing the thing – THAT was me).
The big trace also confirms the large (positive, not GND) plane on the PCB – there are six individual rows and common (+), which has one three pins on the connector, but a much larger trace on the flat flex after that.
So…botch (+) and (-) together on the board, wire to the next board connector and…
Do it again! Because d’oh – the wire has to be bent the other way round when inserting it into the display frame (the flux residue looks messy as hell, but most of the stuff is actually glue from the PCB mount. Didn’t care to clean that up, sorry, temporary fix)
On a side note: This is standard 1.27mm ribbon cable. Four wires cover 5.1mm and that’s a little wider than the 10 contacts (6 + 1 unused + 3) on the board. So that’s 0.5ish mm pitch on the board – nothing impossible, but you damn well should inspect your solder joints once you’re done and then cover them with hot glue for protection.
When writing this article, I was mad at me for a few minutes for not taking photos of the fixed up assembly. Then I remembered that I still have it around, as of course everything is part of the display unit, and therefore was exchanged when the replacement arrived. So this is the display unit with a little bulge at the bottom:
And this is the connector to the controller PCB, which was desoldered with a hot air gun. Nothing for the faint of heart, trust me
There was also a spare controller PCB, so in theory I could have soldered it directly onto it. But why waste a perfectly good unit when you have a working connector…
Aaand of course, hot snot. You wouldn’t risk ripping off any of the legs when handling it. Getting it off the FPC part of the flat flex is enough mechanical and thermal stress already.
The 180° change in soldering angle and the terribly dusty keyboard aside, everything worked fine:
Clip and screw everything else in place, keep the front bezel off, and never even think about closing the damn lid. And you’ll be fine! (for a few days)
It is however still beyond me why you would need six separate chains of 8 LEDs each for the display. I mean, there is no benefit in controlling individual chains as this clearly results in a stripy backlight as shown above. So every leg must be on the same brightness – why split the supply in the first place?
If you insist of having shitty PWM brightness control of your backlight (I hate you, Lenovo), you could, in principle, distribute lowish frequency PWM with phase shift over these 6 chains to lower the visual impact of your penny pinching. But I don’t think they make use of that.
Is it for current limiting purposes, in the rare event of some LED failing and every other LED now taking 1/6 more current? This would clearly require a panel exchange, so the reduced lifetime of the remaining LEDs under full brightness (hello PWM!) would be negligible. It’s probably nice to have, but once your display has failed, you don’t really care much about how that would affect the time until everything will blow up. It will need replacement very soon, so deal with it.
So…yeah. Solder it shut, and everything lights up like a Christmas tree.
Two days later, the replacement arrived and within a few minutes, the damn thing was working again. And moreover, you could close the lid again, without having flimsy cables around that tend to break your front bezel.