My first proper post, mostly of a 'for information' nature, in that it might be of use/interest to others rather than being a call for help, as the problem encountered in our unit has been largely resolved. (A huge thank you to tom66 for this thread, which got me going in the right direction!)
The unit was sold (for a song) as unworking, having received a suspected large voltage spike during a thunderstorm when it stopped working. On applying power there was no response whatsoever - no relay clicking, no front panel LED, no spinning fans, nothing. Given that there are no schematics available online (well at least not that we could find), it became clear that if we were to find the fault, we would only do so after a good deal of detective work, most likely involving reverse-engineering selected parts of whatever boards as necessary. (There are service manuals available online for both the European model, TH-85PF12E, and the US equivalent, TH-85PF12U, and they do contain half-a-dozen or so block diagrams, but which nevertheless did turn out to be useful. We even contacted Panasonic, but no dice.)
On removing covers and pulling the power board which is fed from the mains inlet (the P(SUB) board), we immediately spotted an electrolytic cap with the vent split open, which gave us a hope of a really simple fix, but alas on replacing it there was no change in behaviour. Initial attempts to troubleshoot the switching supply on the P(SUB) board were thwarted by my lack of experience with such supplies, but once we established there were five windings on the transformer, and not the four shown on the block diagram, some simple tracing out of the circuit clarified things a good deal, and we were able to establish that the switching IC was alive and well and generating at least the +5V rail sent out all over the place to power many other boards, and a more localised +9V 'biasing' rail.
It was then we realised we'd probably have to start trying to establish the correct functioning of some of the many error signals and commands coming in/going out to various other boards, something that was likely to involve a deal of guesswork and supposition, as most of these signals are not described or even mentioned in the service manual. It was Googling that led me to the thread mentioned above, and which sowed the idea that closer scrutiny of the 'AC_DET' circuit might be a worthwhile endeavour. I then traced that part of the circuit, which is shown in the attached schematic - that it bore more than a passing resemblance to tom66's diagram was a confidence booster.
Notes on the schematic:
- the '0V' node is the negative end of the DC output of a bridge rectifier (D506 on the board)
- as neutral is basically ground, the circuit relies on this '0V' point out of the bridge being a diode-drop above the negatively-swinging 'live' mains input during the one half-cycle (well it took me a while to get this straight in my head!)
- the 15V zener at the bottom of the chain is not 'zenering': measurements/calcs show that about half the (peak) voltage of 340V is dropped across the 3x 820k resistors, giving a peak current of around 70uA; this drops about 9V (peak) across the 130k's at the bottom, thus keeping the zener off.
- the 2SK3018s are SOT-323 packaged and marked 'KN'
- the 130ks having no marking - 130k is based on DVM readings
- I've included a few of the actual reference designators on the board on the very small chance someone else really wants to know them
Not immediately bothering to work out the precise functionality of the caps involved, I guessed the two MOSFETs would keep the PNP continually on, thus giving a 'high' 'AC detected' signal. However on applying the scope to the board (powered-up on the bench), we got 50Hz spikes from the PNP, a looks-like-it-is-working dip at the one MOSFET drain, and nothing from the other. This immediately suggested that one of the MOSFETs wasn't working: the input at the gate on the good transistor was peaking at 9V, the 'bad' one was 200mV at most. So we kludged in a SOT23 2N7002 instead, and discovered we now got 100Hz pulsing at the PNP, which looked encouraging, but it was asymmetric, with 7ms/13ms between pulses. This was because the voltage swing at the 2N7002 gate was only peaking at about 3V (and not 9V as the good one) - whether due to further damage in that locale, or because it is not the correct replacement transistor type (they are on order) we have yet to determine.
Thinking there was maybe a 20% chance this would allow the TV to run, we replaced the P(SUB) board into the TV and powered up, immediately to get some clicking, a front panel LED, and she is clearly running! An immense feeling of satisfaction I can tell you, as it had taken several weeks and a good deal of head scratching to arrive at this point!
I have now cobbled together a SPICE simulation of the circuit, using a very idealised 177V zener (!), and found the Rohm 2SK3018 SPICE model on the web, and can now better understand the circuit operation. I have attached a few traces for those that maybe like to see such things (as I do!). I guestimated the value of the caps on the drains at 4.7n, as this visually ties in with what we saw on the scope. Because these charge up quite quickly, this explains the narrow pulses at the PNP collector (feeding the optoisolator LED), which again tie in nicely with the narrow pulses we saw on the scope. (The little glitches in the blue trace are the interaction with the other transistor's switching on the other half-cycle, which I've not plotted.)
If using a correct 2SK3018 transistor significantly changes the situation I will report so later!
Tim
The unit was sold (for a song) as unworking, having received a suspected large voltage spike during a thunderstorm when it stopped working. On applying power there was no response whatsoever - no relay clicking, no front panel LED, no spinning fans, nothing. Given that there are no schematics available online (well at least not that we could find), it became clear that if we were to find the fault, we would only do so after a good deal of detective work, most likely involving reverse-engineering selected parts of whatever boards as necessary. (There are service manuals available online for both the European model, TH-85PF12E, and the US equivalent, TH-85PF12U, and they do contain half-a-dozen or so block diagrams, but which nevertheless did turn out to be useful. We even contacted Panasonic, but no dice.)
On removing covers and pulling the power board which is fed from the mains inlet (the P(SUB) board), we immediately spotted an electrolytic cap with the vent split open, which gave us a hope of a really simple fix, but alas on replacing it there was no change in behaviour. Initial attempts to troubleshoot the switching supply on the P(SUB) board were thwarted by my lack of experience with such supplies, but once we established there were five windings on the transformer, and not the four shown on the block diagram, some simple tracing out of the circuit clarified things a good deal, and we were able to establish that the switching IC was alive and well and generating at least the +5V rail sent out all over the place to power many other boards, and a more localised +9V 'biasing' rail.
It was then we realised we'd probably have to start trying to establish the correct functioning of some of the many error signals and commands coming in/going out to various other boards, something that was likely to involve a deal of guesswork and supposition, as most of these signals are not described or even mentioned in the service manual. It was Googling that led me to the thread mentioned above, and which sowed the idea that closer scrutiny of the 'AC_DET' circuit might be a worthwhile endeavour. I then traced that part of the circuit, which is shown in the attached schematic - that it bore more than a passing resemblance to tom66's diagram was a confidence booster.
Notes on the schematic:
- the '0V' node is the negative end of the DC output of a bridge rectifier (D506 on the board)
- as neutral is basically ground, the circuit relies on this '0V' point out of the bridge being a diode-drop above the negatively-swinging 'live' mains input during the one half-cycle (well it took me a while to get this straight in my head!)
- the 15V zener at the bottom of the chain is not 'zenering': measurements/calcs show that about half the (peak) voltage of 340V is dropped across the 3x 820k resistors, giving a peak current of around 70uA; this drops about 9V (peak) across the 130k's at the bottom, thus keeping the zener off.
- the 2SK3018s are SOT-323 packaged and marked 'KN'
- the 130ks having no marking - 130k is based on DVM readings
- I've included a few of the actual reference designators on the board on the very small chance someone else really wants to know them
Not immediately bothering to work out the precise functionality of the caps involved, I guessed the two MOSFETs would keep the PNP continually on, thus giving a 'high' 'AC detected' signal. However on applying the scope to the board (powered-up on the bench), we got 50Hz spikes from the PNP, a looks-like-it-is-working dip at the one MOSFET drain, and nothing from the other. This immediately suggested that one of the MOSFETs wasn't working: the input at the gate on the good transistor was peaking at 9V, the 'bad' one was 200mV at most. So we kludged in a SOT23 2N7002 instead, and discovered we now got 100Hz pulsing at the PNP, which looked encouraging, but it was asymmetric, with 7ms/13ms between pulses. This was because the voltage swing at the 2N7002 gate was only peaking at about 3V (and not 9V as the good one) - whether due to further damage in that locale, or because it is not the correct replacement transistor type (they are on order) we have yet to determine.
Thinking there was maybe a 20% chance this would allow the TV to run, we replaced the P(SUB) board into the TV and powered up, immediately to get some clicking, a front panel LED, and she is clearly running! An immense feeling of satisfaction I can tell you, as it had taken several weeks and a good deal of head scratching to arrive at this point!
I have now cobbled together a SPICE simulation of the circuit, using a very idealised 177V zener (!), and found the Rohm 2SK3018 SPICE model on the web, and can now better understand the circuit operation. I have attached a few traces for those that maybe like to see such things (as I do!). I guestimated the value of the caps on the drains at 4.7n, as this visually ties in with what we saw on the scope. Because these charge up quite quickly, this explains the narrow pulses at the PNP collector (feeding the optoisolator LED), which again tie in nicely with the narrow pulses we saw on the scope. (The little glitches in the blue trace are the interaction with the other transistor's switching on the other half-cycle, which I've not plotted.)
If using a correct 2SK3018 transistor significantly changes the situation I will report so later!
Tim
Comment