Other Parts Discussed in Thread: ISO7710,
Hello:
We finally got the whole circuit working. There were several things to pay attention to
- startup capacitors at S13, P13 zener, TVS at Depletion mode MOSFET,
- For debugging, it is safer to use a diode in the secondary. We learned this after some mishap with MP6908 that blew up EPC2034. We have found fix for that but since we are still in bedug mode, using a diode for now.
- If the controller won't start, we isolate it at S13 &PWML nodes. Insert an external Pulse on a small PCB assy with short interconnect wires. This way we can see all switching in the primary & the secondary waveforms including output behavior, open-loop of course.
- Divider resistor values in-circuit: needs adjusting Vout so that it is not too high to trip OVP( happened during debug, Rv2 appears lower to the feedback)
- There were assembly errors: some values were installed of different value; correct values were put in place.
All that done, we loaded output with 40W( the converter can handle 65W).
All looked as expected: jubilation time!
BUT , someone wanted to see how it would be if we pulled the AC plug & restarted it.
That was complied with:
The circuit DID NOT RESTART. Not even after several attempts & load changes.
Upon setting up "debug" Pulse circuit, we found lower NV6117 appeared to be damaged. It switched but with distorted waveforms, missing cycles, had sharp negative-going spikes...
We do not know if the upper NV6117 also got damaged.
But first, we need to analyze this totally unexpected failure of the lower NV6117.
A review of the waveforms attached indicates that the failure cannot be due to high leakage of other high peaks on lower drain.
SO WHY?
Our hypothesis is as follow
- AC was switched off. But the rectifier circuit has 250 uF capacitor. It has enough energy to keep supplying power to the output.
- While the DC is going down, the controller keeps all functions ON all the way to 9V at VDD. With feedback active, the controller will be forced to keep on increasing Pulse widths to maintain output.
If this is so, at some point we suspect NV6117 can no longer provide adequate VZ, input distorts PWMLs, and finally gets damaged. We know NV6117 is very sensitive to how the input behaves even in normal operation.
- If this hypothesis is valid, then one would suspect the controller to be damaged as well.
- Is it possible the ISO7710, upper NV6117 also were affected?
Possible.
What do you think? What is your hypothesis of this failure? Has this been reported by others?
We clearly need to add some protection: but first what is the root cause?
There is simply not enough room to add this late " turn-Off " protection scheme.
But what is your hypothesis of turn-off failure?
attached: waveforms, schematic, gerbers. These are very similar to TI's eval kit of the same topic.
Clearly, we can build other units only after we W223912ASR37_GNMREQ30-002_COMBO.ziphave addressed and proven a fix .
Right now, time is of the essence.
Appreciate any help in the matter.
