Which type of 100nF capacitor is best for 78xx regulator output?

I use 78L12 or similar voltage regulators in a few power supplies. I use them in their basic setup, see the picture:

This diagram is common for all 78xx and 78Lxx regulators. It says that output capacitor should be 100nF. I am able to obtain these 100nF capacitors either ceramic, or tantalum, or film for almost the same price (near to nothing) so I'd like to know which type of capacitor is the best for this particular case.

Note: I know that a tantalum cap is recommended for SMPS output, but 78xx isn't switched-mode.

• The correct capacitor in this situation is what you have on hand/is cheapest. The 78** series regulators are very tolerant of input and output capacitors. Jan 21, 2012 at 12:10

There is little point to ever use a tantalum when a ceramic is available for the same cost.

Actually, I'd use a 1µF ceramic on the output and a 10µF ceramic on the input. The 7805 will be plenty stable with those, and that will help with transcient response.

The reason you see 100nF (let's not use ".1µF" baby talk) a lot is for historical reasons. Back in neolithic times when a caveman first carved a 7805 from a chunk of silicon, that was about the limit of cheap and available ceramic caps.

• @avakar: It isn't. I'll edit my answer to make this more clear. As I said, baby talk is ".1uF", "10000pF", and the like. In engineering we use the proper units prefix so that there are always 1-3 digits left of the decimal point. Jan 20, 2012 at 13:57
• @avakar I'm guessing some people use ".1µF" instead of "100nF" because they're not confident with their unit conversion skills.. Jan 20, 2012 at 13:59
• @Olin: Not to nitpick, but shouldn't it be written 100 nF (with a space) rather than 100nF? en.wikipedia.org/wiki/ISO_31-0 Jan 20, 2012 at 18:38
• @Jonny: You are nitpicking, but you're also right. There is supposed to be a space there. Sometimes it seems clearer to leave it off, but you're right, that's technically incorrect. Jan 20, 2012 at 18:41
• @Fake: Spacing the prefixes at multiples of 1000 is a convention. It's the standard for how it's done. The reason for not using 10^6 is that this causes too many digits floating around that may not be significant. And, it's no work at all to shift "convert" since it's just shifting a decimal point around. You'd have to "convert" a 5 digit value to something else just as likely. Jan 21, 2012 at 12:44

I use ceramic capacitors in both positions, as do most people, and don't have any problems.