caps are caps, aren't they?

[Martin Czech]

Just comming back from the lab. I thought I'll have some HP4200 fun,
took some caps and measured their frequency response from 100Hz to 40MHz.
Generall: Nearly all caps can be very good modeled with a series RLC
circuit. Some caps however refuse to fit into such an easy circuit.
The fitted RLC substitutes are displayed, as well as resonance frequency
estimation. The resonance can be sharp (not good damped), or flat
(damped). In the sharp case some ringing may be observed in circuits. I
prefer good damping that eats up hf energy, instead of sharp ringing.
I tried to push the caps as far as possible into the test clamps, to
avoid lead inductance. All electrolytics got 9V bias.

10   nF WIMA MKT          9.15 nF, 163  mOhm, 3.37 nH, 28  MHz (sharp)
68   nF WIMA MKT          53.7 nF, 217  mOhm, 4.09 nH, 11  MHz (sharp)
0.68 uF polyester         689  nF,  27  mOhm, 14.0 nH, 1.6 MHz (sharp)
10   nF ceramic           7.99 nF, 790  mOhm, 6.61 nH, 22  MHz (sharp)
100  nF ceramic           90.8 nF, 52.2 mOhm, 2.54 nH, 11  MHz (sharp)
6.8  uF Tantal            6.81 uF, 387  mOhm, 5.71 nH, 1.0 MHz (flat)
1000 uF electrolyt radial 1.15 mF, 56.6 mOhm, 10.7 nH, 50  kHz (flat) 
1000 uF electrolyt axial  1.14 mF, 425  mOhm, 232  nH, 8.8 kHz (flat)
2200 uF electrolyt axial  **
10000uF electrolyt axial  8.95 mF, 26.3 mOhm, 37.1 nH, 9.4 kHz (flat)

The axial electrolytics had of course variing results, depending on
the leg bending. Sometimes I could get very strange resonance, phase
dropping from -90 to -180 deg, then jump to +180 and fall to +90 deg.
I think, this is the extra air capacitance between the legs.  No fitting
was possible in this case. The 2200uF wouldn't behave well under any
conditions, so no fitting of RLC was possible.  The axial 1000uF had the
best behaviour. I always thought, that an array of smaller electrolytic
caps works better, then a giant electrolytic.  If you simulate with
these guys, you will see, that a parallel configuration of the 6.8uF
and 100nF will NOT show capacitive behaviour at higher frequencys.
The 100nF seems to be too far away. You get something like 0 deg,
ie. ohmsch resistive. This is not bad, it eats up energy.

m.c.