Grundig Fine Arts 32 - strange backlight defect - not fixed, but hacked to make it usable

A friend wanted a 32 inch full HD to use as a PC monitor. Those are not easy to find. Most (defect) 32 models are 720p. I found a Grundig Fine Arts 32 on ebay and my friend got it for 42€.

If we only had investigated the seller a little more. It was a professional TV repair shop! Now, if such a guy gets rid of a TV, it will not be an easy fix. In fact, I wasn't able to locate the error this time, but at least I found a wild hack to make the TV usable.

Grundig is actually a no-name product. Someone in Turkey just slaps the name on the devices and the software on the mainboard gets some customizing. I found an almost identically looking TV under the name Murphy, which seems to be Indian. Apart from that, it is a pleasantly looking device with a very stable aluminum frame, a very heavy back cover and a solid glass foot. No plastic anywhere. Hovewer, you will not find any service documentation on those devices.

We needed luck.

When I switched it on, it was dark. With the help of a torch I was able to see the menu. So, the display worked fine. Always a relief.

Then I turned my attention to the inside and the inverter board in particular (the little board in the middle). The TCON and inverter board are part of the panel and manufactured by Samsung. The rest by god knows who.

This board is a nightmare to analyse. It is not just producing 95V for the LEDs, it also has a complex circuitry with a micro processor in the center, which controls the boost converter voltage and the LED dimming.

The next image shows the driver stage. There are four independent LED channels, each driven by a transistor. The transistors control the ground level of the LEDs. The 95V go straight into the LEDs.

I noticed that one of the transistors had a different voltage on its base than the others. This could not be ok and I traced the problem back to a quad OPA, which controls the base voltage of the transistors. Ok, a standard part for 25 cents. I soldered in a new one and was bitterly disappointed that nothing had changed, even though the transistor now was in line with the other three.

Countless hours of testing and experimenting followed. I tested the LEDs with my high voltage bench power supply and all four channels were perfectly ok. Continuously dimmable with no excess current. The 24V supply to the board was stable and the dimming voltage signal from the mainboard was in sync with the menu. Thus, all the inputs were ok. I couldn't say much about that mysterious "sync" input from the TCON, which is supposed to control the adaptive backlight (my theory) and exposed a regular square wave pattern at 150Hz. Whatever this does...disconnecting it did not provoke anything at all.

It turned out that the backlight had only two states: fully lit or completely off. The backlight intensity control menu goes from 0 to 100. At 41, the light snapped from dead to full, but not above 41, no, below 41! That didn't make any sense at all.

I suspected the microprocessor. It is actually cheap, but changing it was no option, because its built-in EEPROM contains the control software.

I came to the conclusion that this board is doomed and ordered a used one from flattvparts.co.uk. Alas, that board exhibited the same erratic behavior. Well, sometimes used boards are defect, so yet another one was ordered from eBay. No luck! Still not working! Both boards came from professional sellers, so I have the chance to send them back.

Desperation was creeping up.

While poking around on the board with my multi-meter, I noticed that the backlight would come on once I touched a certain trace, which lead to the processor. What the..? The processor input is connected to the dimming signal from the main board through two transistors. It knows only two states: High or low. Aha! Pulling it low for a tiny amount of time - the capacitance of the multi-meter was enough - fired up the dimming logic.

Screw all this. I decided to connect a simple "pull down starter switch" as a hack to make the TV at least usable. No backlight dimming, always 100%, but a cheap semi-fix.

Well, this hack works reliably :-) Switch the TV on, wait until the stand by LED stops blinking, press the button, enjoy a fully lit image.

How the circuit on the inverter board is able to dim the LEDs in a continuous fashion is beyond me. The input circuitry, which handles the dimming voltage from the mainboard, switches hard between high and low. I could not find any other pin on the processor, which is coupled to the dimming signal. The only possible input left is the sync signal from the TCON. But this doesn't change with the dimming and is constant. Maybe it was the TCON all along? But the image is perfectly ok, and I did not want to take any chances there.

My nerves were put to a test this time, and the TV is not actually fixed, but I am still satisfied that I came up with a workable solution.