on-chip bypass caps

I do that sort of thing with TDR.

And here's the TDR of a PCB power plane, against the adjacent ground plane, in a multilayer PCB.

The flat pulse is the 50 ohm hardline. It slams down to about 1 ohm when the TDR step hits the PCB plane.

For all practical purposes, the power plane is a perfect capacitor, and adding bypass caps, pretty much anywhere, increases its capacitance. So there's no big incentive to put the bypass caps very close to ICs.

On a sunny day (Thu, 10 Apr 2014 21:25:35 +0200) it happened Gerhard Hoffmann wrote in :

played around with a spectrum analyzer & tracking gen. and

Thank you, that is good info, I can do something with that. But tantals are not that bad? Just too big?

Tants are to be used in conjunction with smaller caps, yes. And smaller caps in conjunction with a ground plane, if possible. The ground plane resonates with the ceramic cap at the highest frequency, and the ceramic caps between each other at middle frequencies, but the tant's ESR dampens them all out. If you're allergic to tant, you might use a fat ceramic with a series resistor (which will have better manufacturing tolerances too), but you need to be aware of voltage coefficient and aging if the value is at all critical (e.g., dominant pole on one of those crummy switchers or LDOs).

I'm disappointed by the last test in that link: what should've been done is, two short male pins soldered to the board, and the socket plugged onto it, face down -- not standing proud on more lead length than a wire wrapped assembly!

Tim

Switch 100 Amps in 50ps for a 1V uP, then figure how much bond wire inductance would be allowed if there was no interleaved capacitance with the logic on chip:-)

Kevin Aylward

(snip)

Well, most now have separate power and ground for IO and core, usually at a different voltage, but yes.

Though there is likely enough clock skew that they don't all switch in the same 50ps.

-- glen