What is the best way to figure the capacitor voltage? I have two cases: the output from the rectifier (half wave or full wave, sqrt(2)*RMS). I have typically used the peak voltage, multiplied it by 1.5, and chosen the first stock voltage rating that exceeds this value. Is there a better way? Is my choice of 1.5 reasonable? The second case is for the output from the regulator, I use the same formula. However, I suspect that since I have a regulator, perhaps 1.5 is too large a factor. Suggestions?

 

I usually go for a safety margin of at least 20%, taking worst case (highest) line and load (no load). 30% would be better for commercial products (different companies have varying derating requirements.

 

For 230V countries the maximum mains voltage is 253V.
If you look at your transformer specifications, you will be given either the off-load voltage or the regulation (or both). The regulation is given as a percentage. If you have the rated voltage, then you can work out the off-load voltage by adding the regulation. i.e if regulation is 15%, then off-load voltage = rated voltage * 1.15
Multiply the off-load voltage by 253/230. Then multiply by√2.
That is the minimum rated voltage for your capacitor.

 

For 230V countries the maximum mains voltage is 253V.
If you look at your transformer specifications, you will be given either the off-load voltage or the regulation (or both). The regulation is given as a percentage. If you have the rated voltage, then you can work out the off-load voltage by adding the regulation. i.e if regulation is 15%, then off-load voltage = rated voltage * 1.15
Multiply the off-load voltage by 253/230. Then multiply by√2.
That is the minimum rated voltage for your capacitor.

Actually, the line voltage has little to do with the problem, since I would be looking at the output of a transformer. 4V RMS to 48V RMS are the design limits for my program (I set the spin control for selecting the output voltage to run 4-48 in 1-volt increments and have selections for 6.3 and 12.6V). And I am only looking at the off-load voltage, since that would be the highest voltage.

 

Capacitors rated for 6.3V are used on 5V supplies. That's a derating of about 20%. I think derating by 20% is a common.

Technically, you can operate an electrolytic capacitor at 100% of its rated voltage because it is formed at a voltage higher than that. But reducing the operating voltage will increase lifetime:

From Panasonic:

 

Actually, the line voltage has little to do with the problem, since I would be looking at the output of a transformer.

Don't fool yourself - the output of the transformer is directly proportional to the line voltage.
If the line voltage goes up by 10%, then the transformer output goes up by 10%.

 

That's a derating of about 20%. I think derating by 20% is a common.

There are a lot of "rules of thumb" in electronics. Many have their basis in fact and some don't.
As far as I can see, +20% for smoothing capacitors comes from 10% for the maximum possible variation in supply voltage, and 10% for transformer regulation.
10% for transformer regulation is about right for transformers in the 50VA to 500VA region, so I'm guessing that the rule of thumb might have originated in the audio field. However, it's way too small for small transformers, which can have regulation figures of 25% or even 40%.

By the way "Electric Double Layer Capacitors (Gold Capacitors)" are those with capacitances of about 1F for memory backup purposes, but the argument applies to normal electrolytics, and the reduction of life by half for every 10°C increase in temperature comes from the Arrhenius equation, and applies to almost everything in electronics.

 

Don't forget to consider the "Ripple-Current-Rating" of the Capacitor(s) You choose.
They must be able to supply at least the Full-Load-Amps of the Load,
preferably twice the FLA of the Load.
.
.
.