The 5V Supply
This board's bottom-right area is almost entirely dedicated to the 5V supply. There are no discrete external transistors here. The chip under that thick U-shaped copper sink is driving the 5V transformer directly. Based on the silkscreen, the output can provide up to 3A.
The two larger caps are the ones I replaced during my original repair. They're on the 5V rail, as is the smaller original cap on the left. The bulging small cap to the right is on the 12V output and I did not bother replacing it originally, since it still looked fine back then.
About That Heatsink
If you want to see the chip hidden under that copper heat-sink, you need to wick the solder off and straighten the pins holding it down. Since I still use the W2452V as my primary display, I would rather not risk going that far.
I have looked at a bunch of DIP-7 off-line standby PWMs from various vendors to find what this chip might be, and the closest probable match I found for this wiring configuration was Sanken's STR-A60xxH.
The 12V And 24V Rails
This is where I ran out of spare parts during my first repair effort. The two larger caps in the top-right corner provide the 24V required by the CCFL inverter board. The crustier capacitor on the left had one of the largest mounds of dried-out electrolyte I had ever seen on a cap, but I accidentally knocked it off before taking a picture. The smaller cap below them is on the 12V rail and also looks like it has seen better days.
These two output rails are regulated by the ST L6599D resonant-mode controller on the bottom with help from the two MOSFETs on the heat sink to drive the transformer and its series resonant capacitor.
Since the previous image does not do justice to how crusty that capacitor's top is, here is a different angle. Before I accidentally knocked it off, the pile of dried black electrolyte formed a nearly perfect cone roughly 10mm tall covering the whole metal top.
Those of you who know their bad caps might recognize these from the infamous Samwha WB series, which was allegedly discontinued in 1999 but found its way back into products manufactured in 2007-2008. Somehow, they still appear readily available on the Shenzen market, even today.
The Last Centurion
The third and only pristine-looking original cap on the 5V rail (bottom-left) is connected downstream from the large caps and inductor, isolating it from most switching noise. This goes to show you that even “bad caps” can last quite a while when they're not exposed to harsh current waveforms.
Curiously enough, the footprint tells us that a fatter, presumably beefier capacitor was originally intended to go in this low-stress location.
Meet The Family
One often overlooked capacitor specification is ripple current rating. Based on the Samwha WB specs I managed to find, the 35V 1000µF caps are rated for a respectable 2.7A, the 25V 680µF for a more modest 1.8A and the 10V for a wimpy 1A. I will be replacing them with 35V 1000µF Epcos B41888 rated for 2.4A, 16V 2700µF Panasonic FM rated for 3.6A and 16V 1200µF FM rated for 2.5A.
Will the Epcos' slightly worse specifications on paper bite me in the rear a few years down the line or will higher quality prevail with more graceful aging?
There we go; the caps look much better without those bulgy tops. Is it just me or does the board look much sexier now?
Rigged For Testing
Before I can turn the monitor on to take any measurements, I have to almost completely re-assemble it. Since I cannot access the PCB once the panel is back on, I had to solder wires and bring the rails out. I mention this now because long wires will introduce common-mode noise, making switching transients look much worse than they actually are. In other words, try to ignore the high-frequency ringing and spikes.
Let's Have Some Results
Since the graticule is nearly invisible after scaling oscilloscope screen caps to fit two in a single image with matching scales, I traded the normal “per division” vertical scale for a “Full Scale”, which applies to individual image halves.
Going from 680µF to 1200µF on the last original 5V cap did not have a big impact. The ripples look marginally better, while the fuzzier lines are caused by slightly worse ringing, courtesy of the new cap's lower ESR providing less damping.
Something More Encouraging
It looks like more of the same on the 12V rail. I replaced the cap closest to the rectifiers with a 2700µF and the second cap with a 1200µF, roughly tripling total rail capacitance. Ripples are halved at best, but ringing also increases noticeably from the decreased ESR.
As shocking as it may sound, the two leaky, bulgy Samwhas on the 12V rail were still getting the job done about as well as my new super-sized caps.