I decided to ask the power supply guy at work!
He suggested I check the peak reverse voltage on the diodes especially on start up on the 24V rail - most people report that these power supplies blow up when they are turned on. And it's always a shorted diode on the 24V output with associated primary side damage.
Guess what I discovered. The diodes are rated to 100V, UF5401 type. The peak reverse voltage? 283V. Yes, I kid you not, they are overvolting these parts by nearly 300%. That's one Quality Vestel(TM) Design!
The transients are repetitive, approximately 20 x ~30ns long over 1.8ms, before it starts ramping up the output voltage.
Scope traces attached show the extent of the FAIL in this design. Note that this set had 1kV diodes installed after a repair. If the set had 100V diodes installed the voltage would be clamped to not much more than 100V dumping tons of energy into the diodes. In operation the reverse voltage is respected (60-70V peak), but it's the transient that really matters.
AND, because the diodes are in parallel, there will be one weak link diode which will break down first, dumping all of the transient energy into it.
The forward voltage bias issue still remains and may also contribute to the failure rate, although after running a 26" set for some time, no one diode seemed to get particularly hot. This was after I installed some metal fins on the diodes to balance the temperature between them though.
So, the proper fix for these power supplies is a replacement of all the 100V diodes with MIN 400V rated parts and ideally 1kV parts. I used HER308G parts from Taiwan Semi, 1kV rated 3A fast recovery.
While I am at it, I am not sure why they bothered with the PFC circuit on these power supplies as it does not seem to work properly at all:
https://www.youtube.com/watch?v=gOOG...ature=youtu.be
It's more like a passive PFC circuit, but has active components. Very odd. Very Vestel to use parts for odd uses. They've used FAN7710 before - a small ~15W CFL driver - to drive a 200W transformer. It's not thermally overspec'd as it's driving external transistors... but why on earth would they use that given there are many other available parts?
Blue trace below is line current in arbitrary units.
He suggested I check the peak reverse voltage on the diodes especially on start up on the 24V rail - most people report that these power supplies blow up when they are turned on. And it's always a shorted diode on the 24V output with associated primary side damage.
Guess what I discovered. The diodes are rated to 100V, UF5401 type. The peak reverse voltage? 283V. Yes, I kid you not, they are overvolting these parts by nearly 300%. That's one Quality Vestel(TM) Design!
The transients are repetitive, approximately 20 x ~30ns long over 1.8ms, before it starts ramping up the output voltage.
Scope traces attached show the extent of the FAIL in this design. Note that this set had 1kV diodes installed after a repair. If the set had 100V diodes installed the voltage would be clamped to not much more than 100V dumping tons of energy into the diodes. In operation the reverse voltage is respected (60-70V peak), but it's the transient that really matters.
AND, because the diodes are in parallel, there will be one weak link diode which will break down first, dumping all of the transient energy into it.
The forward voltage bias issue still remains and may also contribute to the failure rate, although after running a 26" set for some time, no one diode seemed to get particularly hot. This was after I installed some metal fins on the diodes to balance the temperature between them though.
So, the proper fix for these power supplies is a replacement of all the 100V diodes with MIN 400V rated parts and ideally 1kV parts. I used HER308G parts from Taiwan Semi, 1kV rated 3A fast recovery.
While I am at it, I am not sure why they bothered with the PFC circuit on these power supplies as it does not seem to work properly at all:
https://www.youtube.com/watch?v=gOOG...ature=youtu.be
It's more like a passive PFC circuit, but has active components. Very odd. Very Vestel to use parts for odd uses. They've used FAN7710 before - a small ~15W CFL driver - to drive a 200W transformer. It's not thermally overspec'd as it's driving external transistors... but why on earth would they use that given there are many other available parts?
Blue trace below is line current in arbitrary units.
Attached Files
Comment