[time-nuts] Building a DMTD/phase noise set in the 21st century

[ed breya]

I think there may be some confusion over the "super-capacitor" term. 
Over the years, I've seen two types.

The most commonly encountered are the ones in consumer gear, for storing 
charge to keep CMOS RAM alive during power outage and such, for a 
reasonable amount of time. These may be from 47 mF to a Farad or so, and 
have high ESR - they're not expected to dump huge charges, rather steady 
uA range flows for the CMOS.

The other, more exotic kind are for bigger energy storage and power 
conversion - these have very low ESR, like batteries, and may have lots 
of Farads, but generally low working voltages, similar to a battery 
cell. Higher working voltages are attained by stacking, with the 
expected reduction of capacitance, of course.

I've only played with the former, since the latter were unusual and 
expensive. One thing I can say is that the CMOS backup types are pretty 
crappy capacitors for any use beyond their normal role, and they don't 
last very long, in terms of service life - perhaps a decade or so. Most 
that I've salvaged were found to be nearly totally open, like regular 
old electrolytics after their juices have evaporated.

There may be more types nowadays, that overlap a number of applications.

In the situation I mentioned previously, I was planning on paralleling a 
bunch of references to average out the noise somewhat, with combining 
resistances around a hundred ohms. I had hoped to put a bunch of big 
OSCONs in for filtering at the summing junction, but was wary of their 
possible leakage currents lugging it down. I figured I could use 
reasonable-valued, low leakage Ta caps instead, with enough low noise 
amplifier gain to boost their effective values.

Ed