A freebie 32" LCD here. HITACHI brand, but a rebadge Vestel junker by Argos (who are better described by a four letter word beginning with "s".)
17PW25-4 which is dead, no standby, no run power.
Blown mains fuse, primary MOSFET and assoc. components. Also has a shorted diode on the output side (which is what blows the FET in the first place.)
These power supplies are terrible! But the design is so poor, nothing you can really do to fix it, only make it as good as possible.
Because Vestel do not match the output diodes, failure is common.
They probably put all the diodes into a bin, with those pre-turned leads, then pick them randomly!! No lot assignment or matching! This of course does not match them anywhere close to each other... and 7 of them are all in parallel. ARGH.
I measured them, readings were: 443, 444, 443, 460, 433, 438, and 000 (shorted.) Readings mV.
Now, what can be seen here - obviously the shorted one is failed, but it can be also seen that the diode reading 460mV was probably taking close to none of the load (because diodes have an exponential I-V curve, a small change in forward voltage will lead to a large change in current), and the majority of the load was probably taken by the one which shorted, failing that it would have then been split between the 433 and 438 diodes. The rest probably didn't take anywhere near a significant portion.
Terrible design! And don't forget about the -2.1mV/degC variation for diodes. That's negative, so as it gets hotter, the forward voltage drops further causing more current and more heat, a rapidly destructive cycle.
This is the 24V, 5.5A backlight rail, but these diodes experience ripple current of around 2X~3X due to it using a flyback topology. No idea how they snuck a 150W flyback past the EMC guys...!
What can be done? I think the best I can do (as VESTEL love putting crappy UF5402 in parallel) is to match them before install by choosing diodes within +/-2mV and also by using some thermal epoxy to closely match these devices.
This isn't even going into the other issues these supplies have, such as the wildly unstable control loop (1.5Vp-p output ripple on 24V when backlight is at 50%, due to poor compensation and increasing capacitor ESR) and the noisy buzzing they make when operating at certain brightnesses (related to the backlight level.) Plus the odd standby controller fault due to pulling about 15W from the 5V standby rail with no heatsink on the standby IC (when operating...)
/rant over
17PW25-4 which is dead, no standby, no run power.
Blown mains fuse, primary MOSFET and assoc. components. Also has a shorted diode on the output side (which is what blows the FET in the first place.)
These power supplies are terrible! But the design is so poor, nothing you can really do to fix it, only make it as good as possible.
Because Vestel do not match the output diodes, failure is common.
They probably put all the diodes into a bin, with those pre-turned leads, then pick them randomly!! No lot assignment or matching! This of course does not match them anywhere close to each other... and 7 of them are all in parallel. ARGH.
I measured them, readings were: 443, 444, 443, 460, 433, 438, and 000 (shorted.) Readings mV.
Now, what can be seen here - obviously the shorted one is failed, but it can be also seen that the diode reading 460mV was probably taking close to none of the load (because diodes have an exponential I-V curve, a small change in forward voltage will lead to a large change in current), and the majority of the load was probably taken by the one which shorted, failing that it would have then been split between the 433 and 438 diodes. The rest probably didn't take anywhere near a significant portion.
Terrible design! And don't forget about the -2.1mV/degC variation for diodes. That's negative, so as it gets hotter, the forward voltage drops further causing more current and more heat, a rapidly destructive cycle.
This is the 24V, 5.5A backlight rail, but these diodes experience ripple current of around 2X~3X due to it using a flyback topology. No idea how they snuck a 150W flyback past the EMC guys...!
What can be done? I think the best I can do (as VESTEL love putting crappy UF5402 in parallel) is to match them before install by choosing diodes within +/-2mV and also by using some thermal epoxy to closely match these devices.
This isn't even going into the other issues these supplies have, such as the wildly unstable control loop (1.5Vp-p output ripple on 24V when backlight is at 50%, due to poor compensation and increasing capacitor ESR) and the noisy buzzing they make when operating at certain brightnesses (related to the backlight level.) Plus the odd standby controller fault due to pulling about 15W from the 5V standby rail with no heatsink on the standby IC (when operating...)
/rant over
Comment