Capacitor Sizing for Smoothing DC Output on Phase Angle Fired 3kVA Transformer

Hi, after a year of procrastination ive resurected my power supply project as another smps has died. just to refresh your memories, ive got a 3kva transformer controlled on the primary by phase angle firing. it seemsto work OK, but i am struggling to get an accurate voltage reading from the secondary output which i thin is down to the high ripple.
I guess i should put some smoothing on the dc output and wondering what the rule of thumb for capacitors should be. ive got a figure if 10,000uf per amp on a standard rectified out put, but whatshould this be for a zero crossing phase angled firing output ? sorry if this is a dumb question, but i'm stumped.

Are you trying to filter the output to get rid of the ripple or just measure the output?. If you just want to measure it a large resistor in series with a small capacitor could give you a filtered output. the resistor and cap sel As to the particulars, you need to supply some details, as to voltage and current.

Filter capacitors are sized based on the the acceptable ripple, not just the amperage, so selecting the caps can't be done with just the current. As the duty cycle changes waveform to be filtered changes so the filter requirements will change as well. Filters can use inductors and resistors as well.

I'm just trying to measure the output, ive got a panel volt meter and ammeter which read different to my fluke multimeter. he voltage is 32v ( calculated rms after recification) on max setting, but when reducing input the voltage readings become erratic. I am pulling approx 50 amps.

well, the panel meter and fluke will have a different response time and so will read differently as the input is not a sin wave. If you are regulating the output level by switching the input then the output won't be even a little like a sin wave.

I assume the transformer input is AC with a triac varying the duty cycle. As the duty cycle reduces the waveform becomes less sin like so the meter and fluke response both will get odd and will be less like each other.

If you don't see the meter needle vibrating, it has a significant filter. Since the fluke is a digital meter it samples the voltage. I have no idea whether it just makes a quick sample at the display update rate or has some more sophisticated measuring technique. If it samples at the screen update rate it will really have the bunnies.

If you connect a large resistor, say 100k, to voltage and connect a small capacitor, say 0.1 uF, in series to ground, it will filter (sort of average) the voltage measured from the junction of the resistor and capacitor to ground. I'm just picking these values out of a hat, I probably could figure out the size, but it requires way too much thought this late in the day. The larger the capacitor, the slower the output will change as the average voltage changes. If the cap is too small the output will be unstable and if it's too large the measured out will lag the actual output.

You can experiment with the capacitor size so the jitter is almost eliminated. If you eliminate it completely you may have way to slow a response.

As for RMS calculations these assume a sin wave. Calculating the average voltage for some other waveform involves sampling the voltages at a high rate and summing them or knowing the formula for the waveform and integrating it with calculus.

thats where my lack of knowledge comes to play, my idea of putting 100,000uf on the output seems legit, but ive heard horror stories of high ripple causing caps to explode, is this cured by adding more caps ?

100,000 mf would certainly work, but it's probably serious overkill (and expensive). The best quick comparison I can make is that a 3A switching power supply commonly has less than 100 mF for filtering. (I recently designed a 3A switching supply with 22 mf for filtering). It's hard to imagine that less than 7 times the current would require 1000 times the capacitance, even if it's a very different application.

I have only heard of or seen caps explode when they were the wrong (that is too low) voltage or they were polarized and installed backwards. When I was in high school there was a goofy kid who'd plug DC electrolytic caps into the AC outlets at school for the show.

It could be that big caps designed to filter AC could be over worked by other waveforms. I'd suspect that over voltage from inductive kickback would be more likely to cause problems. Certainly large ripple would make the caps work harder. The DC resistance in the caps would generate heat. On the other hand, I'd suspect that the inductance of the transformer would smooth things a bit. It'd be fun to look at it with a scope.

The amount of filtration will also depend on the amount of ripple you can tolerate. I would be surprised if any device drawing 50A would require real good filtering.