Dave,
it's a lot of fun to reply to you! You challenge me, and move me to measuring more things! So now I have to dissect your answer;
Never........NEVER.......rely on an electrolytic as a bypass for RF
energy!!!!!!
WHY? I accept such generalized statements only when they are delivered with generally valid fundamentation. The only reason I can see is that the ESR of an electrolytic capacitor is typically higher than that of a ceramic one. Thus, when bypassing is required only over a frequency range where a sufficiently small ceramic capacitor also provides low reactance, the ceramic is fine. But when good bypassing is needed, say, from 30kHz to 300MHz, what value of ceramic capacitor would you use? In such a case an electrolytic capacitor is required. Common wisdom says that it would be wise to parallel it with a small ceramic cap, but if you look at my measurement below, it turns out that an electrolytic alone can be plenty good enough - specially if it's an SMD.
Use series resonance to your advantage.
This is common practice in the microwave range, where the inductance of any part of practical size is too high for effective bypassing without resonance. But it's always narrow banded.
It's a well known fact among (most) design
engineers that tantalum capacitors are not much good above 500 kHz to a
couple of MHz. Most electrolytics are even worse. Go measure them on
your NANOs.
I did. The graphs are below. A plain run-of-the-mill SMD tantalum cap, 15?F, turns out having an impedance below 0.2? from 50kHz to 300MHz. What's bad about that? If indeed engineers know what you say, then they know wrong. They probably never measured it, and only ever believed in hearsay. Or took data measured long ago, on long-obsolete, giant size capacitors, that simply isn't applicable to modern, small size parts. Look at my RXZ curves below for that tantalum SMD cap.
I just grabbed a 1000 ?F / 25 VDC cap from the parts bin and measured it on
a calibrated HP 8753C using the Smith Chart. Even at 1 MHz it measures 78
m¦¸ with a series reactance 78 nH. Sure, the DC portion is fine, but what
is 78 nH at 50 MHz?
78nH suggests that it's a large, long, axial-lead capacitor. Of course you should NOT use that for RF bypassing! I just measured a modern radial-lead 1000?F 25V electrolytic, and got 6.8nH, more than 10 times better than yours! At that level, it's not much worse at 50MHz than any typical leaded ceramic cap. You just can't beat physics, and lead length costs inductance. Only SMDs get much better.
Take a capacitor better suited as a bypass at HF, a 450 pF dip mica.
WHAT?????? HF is 3 to 30MHz. 450pF has a reactance of 118? at 3MHz! Good luck with using that as a bypass!
Same setup at 10 MHz: 0.1 ohms at 462 pF.
That's its resistance, but its reactance at 10MHz is several tens of ohm, and thus it's pretty useless as a bypass. Except in high impedance circuitry.
I think that you need to get the basics straight...
Here are some measurements on candidate bypass capacitors, from 50kHz to 300MHz. First a typical small ceramic 100nF bypass cap, then a typical cheap 47?F 25V aluminium electrolytic, then a tantalum cap of the same rating, and then a 15?F 20V SMD tantalum cap. I kept the same scaling for all four graphs, for easy comparison. Judge yourself what's best, and do away with long-standing, unfounded prejudice!
Of course these graphs only show resistance, reactance, and impedance, not their ability to handle high currents, run in hot environments, and so on. In high power circuits the need to handle high current or heat might dictate the use of a capacitor that doesn't have the lowest impedance.
Very clearly these measurements run straight against the myth that electrolytic capacitors are useless at RF.
Jim Shorney