Panasonic TX-55AXW634 backlight (short-term) fix (WHL #84)
TVs are apparently still a thing, and so are defective ones. Here’s what I did to a 55″ Panasonic TX-55AXW634 with the red blinking light of death – likely also applicable to similar units of the “LA58 chassis” generation, e.g. the TX-48AX630B 48″ unit or the even smaller 40″ TX-40AX630E. While this fix isn’t perfect nor permanent (I heard it lasted for four months, when presumably another row of LEDs died), it’s good enough to keep the TV going for a bit longer, so one can order spare parts or decide/save up for a new unit.
As per usual for those kinds of repairs: There’s a fully exposed power supply PCB in there, so it’s live at mains voltage when powered on, and the rectified DC smoothing capacitors still hold up to 450V (for days) when fully disconnected. If you have no clue what you’re doing: Just stay the fuck away from it and ask a friend, it’s not worth dying over a TV repair.
Also: I’m sorry for the sub-par photos, this was performed on a Saturday in the office, since I couldn’t be arsed to take the huge thing home and bring it back two days later. Opening it on a large empty bench was pretty convenient (…he said, looking at all the crap on each and every table in his living room)
So, when is this applicable? Apparently there’s a list of blink codes for modern Panasonic TVs, and when it’s “dunno, not this board and not that board, but something else”, it could be the LED backlight. The TV can (doesn’t have to) strobe for a moment (visibly, also audibly) and then default to a half-minute blinking code of the red power LED with the screen permanently off.
eBay and eBay classifieds are actually full of those TVs and their spare parts, because there seem to be two distinct groups of people: Some with a broken Panasonic TV, and some that have fully broken their Panasonic TVs during repair. Of course the right thing to do with defective backlights would be exchanging them, but since that generation requires lifting out the actual LCD panel to access long stripes of PCB-mounted LEDs, and each LED has their own lens that can move freely until everything is perfectly seated, there’s a high chance of breaking the darn panel over a backlight replacement. Another coworker actually tried this and apparently succeeded (with a sub-50″ unit, haven’t seen it in person), but as I said, eBay is full of people that have either given up right away or failed miserably.
(minus the two botched cables on top, I have to say…)
The backlight control circuit is placed on the top quarter of the board, with the PNP driver transistors arranged in groups of three (not in order!) under those heat sinks. The actual backlight is connected via the 15-pin P6 connector on the top right, thankfully in self-explanatory fashion even without the circuit diagram present (but it is, just google it…):
So the top seven wires are the positive supply (common’ed), there’s one blank, and then another seven wires lead back to the transistor array that is controlled by a dedicated chip, switching to ground. Testing voltage of that positive rail would be the first step during troubleshooting, and testing the individual LED strips would be the second. Unfortunately I cannot remember the exact voltage used, but that rail is fed from a 32V supply, so starting from 25V with low current limits is a reasonable approach. Use pairs 1 and 9, 2 and 10, and so on. Note that this likely will be slightly different for smaller models, e.g. the 40″ might only have 5 LED strips on a physically smaller connector, but uses the same overall topology.
In the unlikely case of more than one open circuit, I’d recommend to dump the unit right away – killing multiple lines before the control chip shuts down the entire unit anyway is likely an over-voltage power supply issue. But if just one band of LEDs is open, we’re somewhat in luck.
At first, I thought about just dumping the energy in a big-ass resistor. But that’s not only a lot of power (30ish volts supply at several 100mA) which might be an thermal issue under that plastic back panel, it’s also a tweaking issue: The control chip (visibly!) steps up the drive current and checks for a change in voltage drop. If one matches the resistor for one step, it’s likely out of the allowed operating range for the other – and then the same error code drops. I know because I actually did that and managed to find a suitable value by chance – and then stumbled upon the circuit diagram of the TV. Well, it’s so much easier: Just disable the entire channel, it’s not witchcraft!
This is IC7800, better known as the Rohm BD9479FV. Since that is an off-the-shelf part, unlike those B1ACRL000015 driving transistors, the datasheet tells us everything we need to know – or one could check out the Panasonic schematic where they left us an example of what to do. The BD9479FV is advertised as “White LED Driver for large LCD Panels (DCDC Converter type)” and offers eight channels at up to 500mA. But the entire TV only uses seven…
Since that would shut down a functional TV right away, one channel is disabled. All we need to do is to localize the defective backlight channel, find out the channel number, and disable it as well. Here’s how that is done on channel 8 (BS8, CL8, PWM8):
BSx pins: PNP Tr Base connecting pin x
PWMx pins: Dimming signal input pin x
CLx pins: PNP Tr collector current detection
Do yourself a favour here and discharge the 450V caps before removing the PCB; everything up to here can be done on the top of the board, but the control chip is located on the back side, which requires unplugging all connectors to remove the PCB.
So the dimming signal input PWMx needs to be disconnected and bridged to AGND (pin 24 on the chip) – this is done by removing the corresponding 0Ω link of that individual line, which was kindly provided by Panasonic instead of just feeding one PWM signal to the chip and then bridging all PWM pins right there. They actually route seven individual PWM signals over here, but I don’t think they carry different signals. After that, once a short to pin 24 is placed, the channel is disabled and should not care about the BS and CL pins still being connected to *something* – there will be no current flow to those pins, since the input of the entire attached circuit is connected to the (floating) return pin of the LED strip.
How does it look? Well, like this, of course depending on the channel number of the failed LED strip:
The self-test menu can be accessed by holding down the “Volume-” button on the back of the device, and then pressing the INFO button on the remote a couple times in quick succession (don’t quote me on this!). In there, one can access several test images, which are useful for spotting defective pixels, but also to assess backlight quality. Those pure red/green/blue/white images are basically worst case for image uniformity:
Of course the missing line is pretty bad on its own, but general edge illumination is also pretty poor and there might be some “DSE” (dirty screen effect) added on top – that’s caused by ageing of the remaining LEDs, where some might lose luminosity quicker than others, or change tint ever so slightly. After all, that’s an eight year old TV, probably one of the first 4K ones with all of their downsides, and now it’s time to think about a replacement…