This may be completely off track because it would all depend upon the part layout and ch1 shield connection relative to the rest of the ground-plane and path to real Earth ground.
Agreed it is important to identify the complete current surge path.
Notice the positive 12 volts is used for reference and feedback via the opto isolator. This keeps all the secondary windings nominally regulated. Since the +12 and -12 are regulating to within 100 millivolts on startup and for two minutes thereafter, I kinda ruled this out, but it is always a possibility.
Usually if the feedback fails voltages will soar a bit until shutdown, this would be a good clue if it could be noted.
The cold test is a good idea. If it is really cold outdoors, let the scope sit outside for a bit then see if the run time extends.
Cold by outdoors or spray is a good idea, but should be used carefully since the sprayed area will acquire condensation which could then cause tracking and migration of metals in DC circuits.My best guess at this time, assuming there is not a part shorting beyond the power supply:
If the bias FET V1016 were leaky, it would begin over biasing the drive transistor V1019 preventing full cut off, heating it and causing core saturation, squealing and a sudden drop in output. Scoping the secondary winding waveform would be the best way to identify this, as scoping the primary is not recommended without isolation transformers. A wafer puncture wound of the FET could cause this. Puncture wounds grow more leaky with time and temperature. The excesss current would keep the SCS biased close to turn on at the edge of what should have been a linear ramp across the transformer current sense resistors, but is now just a pedestal close to SCS firing threshold.
The current sense resistors R1023, R1024 which should show a linear ramp, then cutoff of current when the SCS V1014 fires at the prescribed current of around 2.5 amps, are now seeing a slowly rising bias as the steady state conduction builds in the output transistor V1019. When this current builds high enough, the SCS V1014 is not reset or is only reset at the very edge of the remaining ramp, which now rides on top of the current pedestal.
As I said priorly, I have designed these type of circuits using an SCR turnoff (instead of SCS) and a resistor bias in place of the FET. The use of an FET in the location shown is definitely the "Achilles Heel" of this circuit design. They are using an SCS so that they can surely reset for the next cycle, which is often difficult with just an SCR once the holding current is exceeded.Question to Grumage: does the squealing continue or does it stop completely after the two minute warmup period?
If the PS unit could be removed and fired up with dummy loads as outlined in the manual, that would definitely isolate the problem, but that is a real pain to do.
« Last Edit: 2013-11-30, 15:08:50 by ION »
Just because it has a patent application or is patented does not always mean it really works.