Hah! Got my LCD TV working again after four straight days of troubleshooting!
Dennis Ferron
Posts: 480
I bought a 7" TFT LCD TV for my suitcase computer, but the case was too bulky to fit completely inside the available space so I removed the boards and laid them out flat without a chassis on the inside of the lid of my suitcase, using their existing interconnecting wires. Unfortunately the TV boards stopped working after I removed them from the case. I'm not sure if it's something I did to it while removing the boards, or if it was just an unfortunate coincidence that it went bad at that time. When you first turned on the power, the screen would come up and it would display correctly, but it would die after just 1 second or so and shut down the video, though the audio continued to work.
At first I thought it was a power supply issue and would take just a few minutes to sort out, but I located the 5 volt regulator on the board and it was putting out the correct voltage. Then I thought maybe there was an input that blanked out the display that wasn't properly set. I couldn't find such an input, so I postulated that perhaps the TV included a "video killer" circuit that would shut off the TV to conserve power and not display snow if there were no video input.
I worked on it every day after work until about midnight. Ordinarily I would have just given up but it just irked me that obviously the TV was WORKING because you could see an image when it first came up, you just couldn't get the image to stay there. Plus, I had not been sure if it was wise to spend the money on a TV when it's not a necessity and I don't have a lot of money, so to have spent money on it and then not even have it working would have been even worse. What kept me going is that while it seemed I would never get to the light at the end of the tunnel, and I was often stuck, I always had a trail to follow, some line of reasoning that led me to something else that led me to the next thing. Ironically, in hindsight I now know that I was way off on some of my theories, yet the act of pursuing them led me to the correct reasoning behind what was wrong.
I spent the next day trying to find this magical "video killer" circuit. I figured there must be a high-pass filter on the board that would only turn on the video electronics if a signal passed through the filter, and that the circuit had failed. If I could locate the transistor that was responsible for doing this, I should be able to bypass it. I looked up the datasheets of all the IC's on the board and none of them included a built-in video killer circuit, so I figured it must be made from discrete components. I never found the actual video killer but in the process I found a place on the board where I could inject a signal and make the screen come back up. This wasn't a total fix because the display would flicker badly. Though I had found a way to make the display temporarily come back up, I couldn't figure out how it was doing it or how to stop the flicker.
By day three I was so frustrated with not having a schematic available for the board that I started reverse-engineering the bloody thing to get a schematic to work off of. I had already downloaded datasheets for all the chips, so, might as well ohm-out where all the components connect. This is not easy to do with a multilayer surface mount board but it can be done. I also found a datasheet for the board itself, and I found out that the place where I was injecting a signal was not the video killer circuit, but amazingly it was the inverter / video power on off input that I had tried to find on day 1 but had thought didn't exist. You would think that the solution then would be to wire the input to ground or +5V, whichever is necessary to force the video power not to go off. But the input was actually working how it was supposed to, and wiring it to "on" doesn't change anything. Wiring it to "off" doesn't, either. It only makes the video driver board work when you turn the inverter on and off really fast. Interesting.
Well, to make a long story short, it was actually a power supply issue. I had had the right idea to begin with but what I didn't know was that the 5 volt line I deemed OK was only one of about SEVEN different power supplies on the tiny board!! It seems as though every time the engineers added a circuit to the design, they decided to power it from a different voltage and make it require its own supply. As impressive as the small size of this board is, most of it is actually a big waste! Only half of the board is devoted to video processing; another complete 50% of the whole board is just a super-complicated regulated switching power supply that generates the 7 different voltages. It's complicated, and probably unnecessary - hey LG-Phillips, ever heard of a "resistor voltage divider"? They added regulated switching power supply taps for reference voltages that only draw 1 or 2 milliamps, for god's sakes. Even if they really needed seven voltages, why not use a resistor ladder with op-amp voltage buffers? Maybe they don't have op-amps at LG-Phillips. Geez. Clearly they knew how to build switching DC-DC converters, and I guess if the only circuit you know how to build is a DC-DC converter, every electronics problem looks like a DC-DC conversion problem...
Anyhow what was going on is that they had one all-in-one PWM switching power supply controller chip that was providing the "heartbeat" for all these DC-DC conversions, the PWM chip drove the coils, and it looks as though they were regulating the supplies by using resistors and transistor crowbars to pull the voltages down into the correct range. Doesn't soaking up the excess energy that way defeat the purpose of not using a linear regulator in the first place?? All the complexity of a switching power supply, with the heat dissipation and energy inefficiency of a linear supply. That's classic. Anyhow the problem was that this heartbeat PWM chip wasn't sustaining an oscillation. It would oscillate for a second or so when power is first applied, which is why the display comes up, and then it would just die.
I have to give the designers of this board credit - you do need an inverter to drive the 600 volt cold cathode backlight, and you need one to drive the negative supply. Apparently the LCD cannot display black without a negative voltage. I guess the crystals are grey by default and you have to twist them one way to get white, and another way to get black.
Rather than try to resurrect the dead PWM controller chip, I tapped into the board and supplied it with external power. The heck with all this switching junk, I'm sticking with linear regulators for everything except the backlight. I went to the electronics surplus store today and picked up a power supply for cold cathode lights for just $5, and that drives my backlight now. I'm using a 7908 for my negative 8 volt supply (I just made that number up and it worked, I hope negative 8 is what it was before!) and I'm dropping 9 volts through 2 diodes to get 7.8 volts which is close to the 7.5 volts it is supposed to have for that supply. All in all I had to supply it with 600vac for the backlight, dc supplies in 7.5v, 9v, 12v, and -8v, and it has some on board linear regulators that produce 5v, 3.3v, 2.5v, 1.5v, and 1v. Pretty crazy huh? But once I got all the correct voltages fed into it, the TV worked again!!
Now I just have to build a "proper" permanent power supply for it. I used a 9 volt battery to get my negative input to my 7908, and a combination of batteries and a 12 volt wall transformer for the positive input to the boards and regulators. I think I'll just replace all this mess with a center-tapped 24 volt transformer to give me +12 and -12 and regulate it from there.
At first I thought it was a power supply issue and would take just a few minutes to sort out, but I located the 5 volt regulator on the board and it was putting out the correct voltage. Then I thought maybe there was an input that blanked out the display that wasn't properly set. I couldn't find such an input, so I postulated that perhaps the TV included a "video killer" circuit that would shut off the TV to conserve power and not display snow if there were no video input.
I worked on it every day after work until about midnight. Ordinarily I would have just given up but it just irked me that obviously the TV was WORKING because you could see an image when it first came up, you just couldn't get the image to stay there. Plus, I had not been sure if it was wise to spend the money on a TV when it's not a necessity and I don't have a lot of money, so to have spent money on it and then not even have it working would have been even worse. What kept me going is that while it seemed I would never get to the light at the end of the tunnel, and I was often stuck, I always had a trail to follow, some line of reasoning that led me to something else that led me to the next thing. Ironically, in hindsight I now know that I was way off on some of my theories, yet the act of pursuing them led me to the correct reasoning behind what was wrong.
I spent the next day trying to find this magical "video killer" circuit. I figured there must be a high-pass filter on the board that would only turn on the video electronics if a signal passed through the filter, and that the circuit had failed. If I could locate the transistor that was responsible for doing this, I should be able to bypass it. I looked up the datasheets of all the IC's on the board and none of them included a built-in video killer circuit, so I figured it must be made from discrete components. I never found the actual video killer but in the process I found a place on the board where I could inject a signal and make the screen come back up. This wasn't a total fix because the display would flicker badly. Though I had found a way to make the display temporarily come back up, I couldn't figure out how it was doing it or how to stop the flicker.
By day three I was so frustrated with not having a schematic available for the board that I started reverse-engineering the bloody thing to get a schematic to work off of. I had already downloaded datasheets for all the chips, so, might as well ohm-out where all the components connect. This is not easy to do with a multilayer surface mount board but it can be done. I also found a datasheet for the board itself, and I found out that the place where I was injecting a signal was not the video killer circuit, but amazingly it was the inverter / video power on off input that I had tried to find on day 1 but had thought didn't exist. You would think that the solution then would be to wire the input to ground or +5V, whichever is necessary to force the video power not to go off. But the input was actually working how it was supposed to, and wiring it to "on" doesn't change anything. Wiring it to "off" doesn't, either. It only makes the video driver board work when you turn the inverter on and off really fast. Interesting.
Well, to make a long story short, it was actually a power supply issue. I had had the right idea to begin with but what I didn't know was that the 5 volt line I deemed OK was only one of about SEVEN different power supplies on the tiny board!! It seems as though every time the engineers added a circuit to the design, they decided to power it from a different voltage and make it require its own supply. As impressive as the small size of this board is, most of it is actually a big waste! Only half of the board is devoted to video processing; another complete 50% of the whole board is just a super-complicated regulated switching power supply that generates the 7 different voltages. It's complicated, and probably unnecessary - hey LG-Phillips, ever heard of a "resistor voltage divider"? They added regulated switching power supply taps for reference voltages that only draw 1 or 2 milliamps, for god's sakes. Even if they really needed seven voltages, why not use a resistor ladder with op-amp voltage buffers? Maybe they don't have op-amps at LG-Phillips. Geez. Clearly they knew how to build switching DC-DC converters, and I guess if the only circuit you know how to build is a DC-DC converter, every electronics problem looks like a DC-DC conversion problem...
Anyhow what was going on is that they had one all-in-one PWM switching power supply controller chip that was providing the "heartbeat" for all these DC-DC conversions, the PWM chip drove the coils, and it looks as though they were regulating the supplies by using resistors and transistor crowbars to pull the voltages down into the correct range. Doesn't soaking up the excess energy that way defeat the purpose of not using a linear regulator in the first place?? All the complexity of a switching power supply, with the heat dissipation and energy inefficiency of a linear supply. That's classic. Anyhow the problem was that this heartbeat PWM chip wasn't sustaining an oscillation. It would oscillate for a second or so when power is first applied, which is why the display comes up, and then it would just die.
I have to give the designers of this board credit - you do need an inverter to drive the 600 volt cold cathode backlight, and you need one to drive the negative supply. Apparently the LCD cannot display black without a negative voltage. I guess the crystals are grey by default and you have to twist them one way to get white, and another way to get black.
Rather than try to resurrect the dead PWM controller chip, I tapped into the board and supplied it with external power. The heck with all this switching junk, I'm sticking with linear regulators for everything except the backlight. I went to the electronics surplus store today and picked up a power supply for cold cathode lights for just $5, and that drives my backlight now. I'm using a 7908 for my negative 8 volt supply (I just made that number up and it worked, I hope negative 8 is what it was before!) and I'm dropping 9 volts through 2 diodes to get 7.8 volts which is close to the 7.5 volts it is supposed to have for that supply. All in all I had to supply it with 600vac for the backlight, dc supplies in 7.5v, 9v, 12v, and -8v, and it has some on board linear regulators that produce 5v, 3.3v, 2.5v, 1.5v, and 1v. Pretty crazy huh? But once I got all the correct voltages fed into it, the TV worked again!!
Now I just have to build a "proper" permanent power supply for it. I used a 9 volt battery to get my negative input to my 7908, and a combination of batteries and a 12 volt wall transformer for the positive input to the boards and regulators. I think I'll just replace all this mess with a center-tapped 24 volt transformer to give me +12 and -12 and regulate it from there.