Good points. Especially how too low of an ESR can cause oscillations. (While having a high ESR can be bad in terms of decreased efficiency and greater self heating, it certain situations it can be advantageous due of its higher damping factor. In other words, it is less likely to oscillate with the inductance of the power rails.) But I disagree that “higher voltage capacitors often have lower ESR due to a larger plate area.” A higher voltage capacitor has a thicker dielectric, all else being equal. Perhaps you meant that a higher capacitance capacitor has a lower ESR (all else being equal), which would be correct. Lowering the dielectric’s dissipation factor will also lower the ESR.
ESR isn’t due to the dielectric (which ideally has infinite resistance), it depends on the thickness and size of the plates, and in electrolytics, the conductivity of the electrolyte (which acts as one of the plates).
As an example, here is a datasheet (PDF) typical of motherboard capacitors; ESR generally decreases with voltage and larger case size for the same values, for example, 1200 uF at 6.3 v has twice the ESR as the same value at 16 v, and 1500 uf has three case sizes at 6.3 v, all with different ESR.
One exception you will find is capacitors (of the same series) which span a very wide voltage range; high voltage capacitors (as found on primary side filters, 200-450 v) usually have higher ESR because they need a different electrolyte.
Also, I didn’t mention ripple current in my previous post, which is largely a function of ESR and temperature rise.
ok, thanks everyone. I will go check it. Hopefully it will be obvious upon inspection. I’m not entirely confident of that tho, lol.
No, this is not correct. ESR is an AC phenomenon, not a DC phenomenon, and thus depends on the dissipation factor (D.F.) of the dielectric in addition to plate geometry. (Here is a paper by Illinois Capacitor which talks about it.) ESR is also affected by the conductivity of the electrolyte.
Yes, on electrolytic capacitors with working voltages below 100 V, the maximum D.F. specification for the dielectric becomes lower as the working voltage spec is increased. So all else being equal, the maximum D.F. specification for a capacitor with a working voltage of 25 V will be lower than the maximum D.F. specification for a capacitor with a working voltage of 10 V. Because ESR is a function of D.F., the ESR of the former will be less than the ESR of the latter.
No, those are not ESR values. They are overall impedance values at 100 kHz. The impedance is comprised of X[sub]C[/sub], ESR, ESL, and R[sub]P[/sub]. For a graphical representation of this, check out Fig. 1-11 in Nichicon’s technical document on aluminum electrolytic capacitors. Note that ESR is the dashed line and impedance is the solid line.
Add in the fact that he was on his way home from Jim Williams’ memorial service when he crashed, and it was a really bad month for this sparky’s heros. At least Pease went out while driving his beloved beetle, and got extra irony points for having written a book about how to avoid car crashes.
Yea, Kevbo, we lost a couple good guys there. 
I own books by both of them.
Just wanted to clarify that, technically speaking, ESR is still “there” (due to plate geometry and electrolyte conductivity, but not D.F.) if a steady-state DC voltage is applied across a capacitor. But it doesn’t matter, because the leakage current is usually so small that the ESR is dissipating almost zero energy.
I know that ESR isn’t the same as AC impedance, but I often see the terms used interchangeably, especially since only one or the other is usually specified (e.g. a datasheet for a regulator may state that ESR should be less than 1 ohm, with no mention of impedance*).
*Which can be modeled, but in the SPICE program I use (LTSPICE IV) the built-in capacitors (a random selection from different manufacturers) have only ESR specified, although ESL and parallel resistance can be defined (I have defined my own by putting impedance where ESR goes and haven’t had discrepancies from the actual circuits). That’s at typical usage frequencies, of course.
Back around 1980 I had a job testing PCBs with electrolytic capacitors in them and if one had been wired in wrong way round you soon knew about it. Oil and paper contents blasted all over the place, and your nerves shredded too. It got so I would turn one on and take a turn round the office. Those tantalums burnt up quickly and you had to act fast if you didn’t want the board to burn (which meant they couldn’t sell it and it got retained for internal use).
I’ve seen tantalum caps blast holes in PCBs when the fail due to excessive self heating.
The (newer) polymer tantalum caps are supposedly “better behaved” when they fail, since they don’t contain MnO[sub]2[/sub].