Last week i replaced a mainboard in a LG plasma. The board was unrepairable as the main CPU went short on the 3.3v rail. Just the next day after i replaced it, i get called that it now does the same thing again (powers up, no picture) and that before it went out there was a pop noise.
I went to inspect the damage, and there was a lot. A surge had gone thru the HDMI input, blew the resistors on that input, burned a hole in the HDMI switch and killed two traces, blew a little transistor that had something to do with switching the HDMI too and i also found the tuner chip and the audio amplifier shorted out after i removed the HDMI switch. Also blown was a 1 ohm resistor which connected the 3.3v regulator to these devices.
Somehow, the CPU's audio section, which was also powered from the same 3.3v rail these things were on, made it alive. It had an inductor before the ceramic filter caps, likely that's why. I was able to fix most of this board using parts from the other board with the dead CPU. Two HDMI inputs and the tuner no longer work, but it is used with a single STB and nothing else anyway, so it's not that big of a deal. The tuner should have worked but for some reason the Intel (yeah, really!) tuner chip decided to blow itself up at some point between pulling it from the old board and soldering it to the new one.
Now, the STB's HDMI output was also shot, likely it died along with the first mainboard, but i didn't think much of it at that time, we tried another STB when i swapped the board and the HDMI worked on that one, so we left it at that. After the second board blew, i checked the ground of the coax going into the STB, and the mains testing screwdriver glowed. Ouch. The STB has no earth connection, it is powered from a wall wart, and the only path for that coax to mains earth is thru... the plasma's HDMI connector. The plasma's power supply has an earth connection and the mains earth in the house was correctly done.
This is one of those cases where it wouldn't have blown if the TV were plugged incorrectly in a socket that was not earthed. I asked the woman's husband to run an earth connection to the TV cable splitter and there should be no more problems with any TV due to this kind of surge. This should be the cable company's responsibility, but even here in my apartment you get a shock if you touch the coax and something earthed at the same time, so they don't care.
Now, looking at this, it is likely that a simple, cheap 3.6v 1W zener would have eaten that surge for breakfast, limiting the damage to maybe one HDMI input. Or, worst case it would have shorted out first instead of having half the ICs on the board blow up.
So, next time you design or troubleshoot a power supply intended to feed expensive stuff, put some damn zeners on the output. Even if you got a linear reg before your MCU - if that regulator shorts out (and they certainly do if the maximum input voltage is exceeded), then the full input voltage will be applied to your little processor, and you can be pretty sure the magic smoke will escape. A $0.1 zener can save a $100 processor - just think about it.
I know that most devices nowadays include on-chip zeners or ESD protection diodes, but relying on them is stupid. If an on-chip protection diode shorts out, the whole device is rended inoperable anyway, so they are absolutely pointless. I can understand the (ab)use of a built-in zener on a $1 mosfet in a cheap one-transistor SMPS, but relying on protection diodes built into a complex digital circuit is very bad engineering practice. *hint* ASUS *hint*
Hope you learned something tonight.
I went to inspect the damage, and there was a lot. A surge had gone thru the HDMI input, blew the resistors on that input, burned a hole in the HDMI switch and killed two traces, blew a little transistor that had something to do with switching the HDMI too and i also found the tuner chip and the audio amplifier shorted out after i removed the HDMI switch. Also blown was a 1 ohm resistor which connected the 3.3v regulator to these devices.
Somehow, the CPU's audio section, which was also powered from the same 3.3v rail these things were on, made it alive. It had an inductor before the ceramic filter caps, likely that's why. I was able to fix most of this board using parts from the other board with the dead CPU. Two HDMI inputs and the tuner no longer work, but it is used with a single STB and nothing else anyway, so it's not that big of a deal. The tuner should have worked but for some reason the Intel (yeah, really!) tuner chip decided to blow itself up at some point between pulling it from the old board and soldering it to the new one.
Now, the STB's HDMI output was also shot, likely it died along with the first mainboard, but i didn't think much of it at that time, we tried another STB when i swapped the board and the HDMI worked on that one, so we left it at that. After the second board blew, i checked the ground of the coax going into the STB, and the mains testing screwdriver glowed. Ouch. The STB has no earth connection, it is powered from a wall wart, and the only path for that coax to mains earth is thru... the plasma's HDMI connector. The plasma's power supply has an earth connection and the mains earth in the house was correctly done.
This is one of those cases where it wouldn't have blown if the TV were plugged incorrectly in a socket that was not earthed. I asked the woman's husband to run an earth connection to the TV cable splitter and there should be no more problems with any TV due to this kind of surge. This should be the cable company's responsibility, but even here in my apartment you get a shock if you touch the coax and something earthed at the same time, so they don't care.
Now, looking at this, it is likely that a simple, cheap 3.6v 1W zener would have eaten that surge for breakfast, limiting the damage to maybe one HDMI input. Or, worst case it would have shorted out first instead of having half the ICs on the board blow up.
So, next time you design or troubleshoot a power supply intended to feed expensive stuff, put some damn zeners on the output. Even if you got a linear reg before your MCU - if that regulator shorts out (and they certainly do if the maximum input voltage is exceeded), then the full input voltage will be applied to your little processor, and you can be pretty sure the magic smoke will escape. A $0.1 zener can save a $100 processor - just think about it.
I know that most devices nowadays include on-chip zeners or ESD protection diodes, but relying on them is stupid. If an on-chip protection diode shorts out, the whole device is rended inoperable anyway, so they are absolutely pointless. I can understand the (ab)use of a built-in zener on a $1 mosfet in a cheap one-transistor SMPS, but relying on protection diodes built into a complex digital circuit is very bad engineering practice. *hint* ASUS *hint*
Hope you learned something tonight.
Comment