In sci.electronics.misc AB said:
Yes indeed, that would be marvelous Ian!
How does one build an ultralow power inverter? Circuit complexity and
expense (within limits) isn't a big factor, but efficiency and light
weight are. I can tollerate high ripple as long as the oscillation
frequency is relatively stable and is >10 kHz.
I need 80 to 120 vdc variable and regulated (or stiff enough so that
regulation is not necessary). Note that my load will draw 100
microamps (worst case) and will be as low as 100 nA (or less).
My idea was to use a small stack of batteries, which would not have
inefficiencies at low load currents.
Ok, that's an utterly reasonable, and totally impractical solution
(in many cases) reliability will suck, as a hundred cells in series
only needs one to fail.
Why do you want this?
Some numbers on your proposed solution.
Call it 90 1/2N cells.
That's 10mm or so diameter by 14mm * 90, or about 10mm diameter by
a meter.
Call it 100cc, and I'll guess 800g, probably a kilo of battery pack,
and it'll store maybe 10Wh.
You say you want efficiancy down to picoamps, which is under a microwatt.
The batteries will typically self-discharge by at least a percent a week,
or 100mWh/week, or about .59mw, so around 5ua.
There generally isn't much point in going below 2-3ua.
You need 100ua *150V, or say 15mw.
Not a really big number, you might concievably get it from one cell.
If weight is a big concern, I'd probably be looking hard at using a
Li-ion cell (mobile phone battery) that are light, availble nearly free,
and have very good self-discharge rates.
They also have a reasonable voltage, whereas Ni* get a bit low on one
cell at the low end.
I'd look at maxims low power chips that have low battery cutoff (li-ion
hate being deep discharged) with a transformer on the output inductor.
You probably want a very low frequency, perhaps 10Khz.
Also perhaps an external FET rather than an internal one.
