cupoftea
Advanced Member level 5
Must a PCB be fully populated for a hipot test?
If doing a hipot test on an offline flyback SMPS, in an earthed metal enclosure, what would
you expect to be the difference between doing it with
a bare PCB and a fully populated PCB?
Obviously the enclosure would never actually be earthed for the hipot test. As such the enclosure would effectively be a piece of floating metal in the isolation barrier of the flyback SMPS.
The earthed enclosure should be at least 3mm away from the primary and secondary PCB tracks and components, so as to ensure an overall 6mm of clearance between primary and secondary.
Now suppose that a secondary side PCB track that is connected to a Zener diode cathode is unfortunately placed just 1mm away from the earthed enclosure. This is bad, but would it
show up as a HIPOT_FAIL in the case of the unpopulated PCB? -Because in
the unpopulated PCB, the Zener is not fitted, and so the offending track is actually just "floating" (because no components are fitted to connect the offending track to anything)....As such, the offending track cannot be deemed to part of the secondary circuit as such...and so
there is unlikely to be a flashover failure. So would you agree that for a proper hipot test on an offline flyback, the PCB must indeed be fully populated?
Chance of failure of hipot test increased?
Also, supposing you are doing a hipot test on an offline flyback with a fully populated PCB...and
supposing that all of your secondary side components have solder shorts between every pin. Then
would you agree that this flyback would be more likely to fail the hipot test than a fully populated PCB with no
shorts?
_________________________- -------------------------___________________________--
Method for hipot on unpopulated PCB:
So anyway,
Suppose you have a unpopulated offline flyback PCB in an earthed enclosure, and you want to hipot test it.
But you don't have the components to be able to populate it yet.
1...Would you agree that this is possible?
2....Would you agree, the best way to do it, would be to look at the isolation barrier...
then pick the tracks on pri and sec sides which are nearest to the isolation barrier.
Then..on sec side, find the sec track that is closest to the earth (this is the one that's most likely to flash over)..
then..short this track to both the output connector pad, and to the sec track thats nearest the pri tracks
...do the same for the primary...(ie find pri track that's nearest to the earthed enclosure)...then short that track to the input
connector pad, and to the pri track thats nearest the sec tracks.....then do the hipot test...if this PCB passes the hipot test, then the fully populated PCB will definitely
pass the hipot test?
_______________________---------------------------_____________________-----
Checking a hipot tester:
Also, supposing your prototype offline Flyback PCB
comes back from production saying it failed hipot.
But On your hipot instrument, (the same as theirs) it passed.
So you now suspect your hipot tester is not giving 4kV , 50Hz.
So you now have to measure its output. You have a 100:1 probe but that's not enough
attenuation to protect the DS1074 Rigol scope.
So, would you agree its OK to put say forty 100kOhm resistors in series and measure at the half way point of this
resistive ladder?...using the 100:1 probe?
The hipot tester's current scale only goes to 3mA so you cant load it with more than 3mA.
Do you have any better measurement method?
_____________________--------------------
Apologies is this is all as clear as mud.
In brief, whats your favoured way of doing a hipot test on an unpopulated offline flyback PCB (one that will sit in an earthed enclosure)?
If doing a hipot test on an offline flyback SMPS, in an earthed metal enclosure, what would
you expect to be the difference between doing it with
a bare PCB and a fully populated PCB?
Obviously the enclosure would never actually be earthed for the hipot test. As such the enclosure would effectively be a piece of floating metal in the isolation barrier of the flyback SMPS.
The earthed enclosure should be at least 3mm away from the primary and secondary PCB tracks and components, so as to ensure an overall 6mm of clearance between primary and secondary.
Now suppose that a secondary side PCB track that is connected to a Zener diode cathode is unfortunately placed just 1mm away from the earthed enclosure. This is bad, but would it
show up as a HIPOT_FAIL in the case of the unpopulated PCB? -Because in
the unpopulated PCB, the Zener is not fitted, and so the offending track is actually just "floating" (because no components are fitted to connect the offending track to anything)....As such, the offending track cannot be deemed to part of the secondary circuit as such...and so
there is unlikely to be a flashover failure. So would you agree that for a proper hipot test on an offline flyback, the PCB must indeed be fully populated?
Chance of failure of hipot test increased?
Also, supposing you are doing a hipot test on an offline flyback with a fully populated PCB...and
supposing that all of your secondary side components have solder shorts between every pin. Then
would you agree that this flyback would be more likely to fail the hipot test than a fully populated PCB with no
shorts?
_________________________- -------------------------___________________________--
Method for hipot on unpopulated PCB:
So anyway,
Suppose you have a unpopulated offline flyback PCB in an earthed enclosure, and you want to hipot test it.
But you don't have the components to be able to populate it yet.
1...Would you agree that this is possible?
2....Would you agree, the best way to do it, would be to look at the isolation barrier...
then pick the tracks on pri and sec sides which are nearest to the isolation barrier.
Then..on sec side, find the sec track that is closest to the earth (this is the one that's most likely to flash over)..
then..short this track to both the output connector pad, and to the sec track thats nearest the pri tracks
...do the same for the primary...(ie find pri track that's nearest to the earthed enclosure)...then short that track to the input
connector pad, and to the pri track thats nearest the sec tracks.....then do the hipot test...if this PCB passes the hipot test, then the fully populated PCB will definitely
pass the hipot test?
_______________________---------------------------_____________________-----
Checking a hipot tester:
Also, supposing your prototype offline Flyback PCB
comes back from production saying it failed hipot.
But On your hipot instrument, (the same as theirs) it passed.
So you now suspect your hipot tester is not giving 4kV , 50Hz.
So you now have to measure its output. You have a 100:1 probe but that's not enough
attenuation to protect the DS1074 Rigol scope.
So, would you agree its OK to put say forty 100kOhm resistors in series and measure at the half way point of this
resistive ladder?...using the 100:1 probe?
The hipot tester's current scale only goes to 3mA so you cant load it with more than 3mA.
Do you have any better measurement method?
_____________________--------------------
Apologies is this is all as clear as mud.
In brief, whats your favoured way of doing a hipot test on an unpopulated offline flyback PCB (one that will sit in an earthed enclosure)?
Last edited: