I recapped my AGI BPS-2504-4U 250 W AT style PSU this week and ran into an interesting problem with ripple. The first problem was simply my being a bonehead -- in my haste to check out the recapped unit, I measured the waveform on the scope without putting the PCB back into the PSU case. Wow! There was an awful lot of RF ripple -- more than a hundred millivolts on all rails. That was a dumb idea.
OK, having calmed down and put everything back together, I got more reasonable numbers, except the +12 V rail still had about 90 mV of ripple. I had replaced the two CapXon 470 uF 16 V caps, one each on the -5 and -12 V rails, with Panasonic FMs of the same capacitance and voltage rating. (The old caps had about 0.2 Ohms of ESR and the new ones about 0.03 Ohms. Nonetheless, the PSU was working OK.) I managed just barely to squeeze in 10 mm caps in place of the old 8 mm ones. I figured that I had ordered the wrong size. These 10 mm caps were each touching an inductor. To make a long story short, I looked through my order and found that I had also ordered many of the 470 uF 16 V Panasonic FMs of the correct 8 mm diameter. (I have no idea why I ordered two different sizes.) When I replaced the 10 mm caps with the 8 mm caps the ripple came down to a more reasonable 30 mV!
The only explanation for this improvement that I can come up with is that the larger diameter caps, each touching both a straight inductor and the +12 V filter cap, must have managed to couple some ripple from an inductor back into the output, while the smaller diameter caps have a smaller coupling because of the gap.
OK, having calmed down and put everything back together, I got more reasonable numbers, except the +12 V rail still had about 90 mV of ripple. I had replaced the two CapXon 470 uF 16 V caps, one each on the -5 and -12 V rails, with Panasonic FMs of the same capacitance and voltage rating. (The old caps had about 0.2 Ohms of ESR and the new ones about 0.03 Ohms. Nonetheless, the PSU was working OK.) I managed just barely to squeeze in 10 mm caps in place of the old 8 mm ones. I figured that I had ordered the wrong size. These 10 mm caps were each touching an inductor. To make a long story short, I looked through my order and found that I had also ordered many of the 470 uF 16 V Panasonic FMs of the correct 8 mm diameter. (I have no idea why I ordered two different sizes.) When I replaced the 10 mm caps with the 8 mm caps the ripple came down to a more reasonable 30 mV!
The only explanation for this improvement that I can come up with is that the larger diameter caps, each touching both a straight inductor and the +12 V filter cap, must have managed to couple some ripple from an inductor back into the output, while the smaller diameter caps have a smaller coupling because of the gap.
Comment