On Thu, 24 Nov 2011 18:13:15 -0800 (PST),
wrote:

I found this, which calculates and measures caliper battery life:
http://www.davehylands.com/Machinist/Caliper-Batteries/

Small, cheap and simple are the main factors here. The r.c.m. guys
aren't going to be building switching regulators, and switching
regulators generally aren't more efficient at these power levels
anyhow--their quiescent current draw's too high.

True. However, switching regulators usually have some manner of load
shedding when the supply voltage is insufficient. Below that
threshold, the current drain is usually in nanoamps.

(I've made a study of designing microwatt switchers, from scratch.
It's possible, but wholly inappropriate here.)

You're ahead of me. I've never designed anything in that low power
class. Different world. Can you point me to a suitable (or close to
suitable) regulator chip?

Not so fast... The advantage of the thin-film PV panels is that
(appropriate) panels excel at producing power even in dim light.
Polycrystalline silicon panels don't.
The array I suggested for experimentation is thin-film for that
reason--so it can work in indoor light levels.

Decisions, decisions, and more decisions. Polycrystaline has a cost
advantage and is more efficient than single layer thin-film. Well, if
I wanted to go cheap, I would use amorphous cells and mold them into
the plastic case. For small solar cells, the cost of monocrystaline
isn't all that much more (i.e. most of the cost is in packaging and
handling) but won't work well with indoor lighting. So, I guess
thin-film is the least disgusting.
http://en.wikipedia.org/wiki/Solar_powered_calculator
"Solar calculators may not work well in indoor
conditions under ambient lighting as sufficient lighting
is not available."

The next question is for how long will it run? *Assuming the calipers
can handle 3.0VDC without damage, how long will a junk 100UF
electrolytic cap run the calipers?

a) How long will it run? Not nearly long enough, and b) 3.0VDC is
waayyy too risky for my blood. 20uA will discharge 100uF from 2.0V to
1.35V in 3.25 seconds.

I used 1000uF elsewhere in my calcs, but slipped here and used 100uF
instead. Sorry.

I think you might be a bit too conservative. 5ua leakage is high.
Most of the spec sheets I've skimmed show 1-2ua for a typical 1F 5.5V
super-cap.

Of the setup I suggested, the most marginal part is the itty bitty PV
panel (its output is on the low side). Dark leakage on my much-larger
10x55mm calculator panel is about 8uA @ 1.7V bias.

The alternative is to lose approximately 0.3V in a series Schottky
diode. That's about 20% of the power budget, which is probably too
much.

The supercap works wonderfully well. Charge 0.6F to 1.8V, and you've
got 4 hours' runtime until you reach the 1.35V battery-low display-
starts-blinking level. (Assuming 20uA total draw, to allow for some
leakage.)

Ok. You've sold me. I was trying to see what could be done with
commodity electrolytic caps. Also, super-caps fail to appreciate high
humidity, which may become a problem.

http://www.kpsec.freeuk.com/capacit.htm
From 1.37V is roughly 50% of full 3.0VDC charge. *That's about 80% of
1RC time constant. *1RC is:
* *0.8 * 100K * 1000uF = 80 seconds
That's probably enough to make a few measurements. *Any longer and a
super-cap will probably be needed. *Picking 50% of full charge out of
the hat is rather convenient, as it makes the time to charge from zero
to the dropout point the same 80 seconds (yes, I'm lazy). *Whether the
user really wants to wait 1.5 minutes under a desk lamp for the
calipers to be usable is dubious. *Of course, a longer run time, means
a longer charge time. *For example, a 1F 5V 1ua leakage super-cap,
will run the calipers for 80,000 seconds, but will also take 80,000
seconds to charge.

Not 80,000s. Expose the PV to sunlight (or directly to a lamp), and
it'll charge (initially) 50x faster. You'd only have to do that
once. Indoors, the PV would keep it topped off, that's the idea.

Yep. However, I screwed up. The discharge load is:
1.5VDC / 15uA = 100K ohms
However, the charging ESR is much less.
3.0VDC / 2ma = 1.5K
It will certainly be higher a lower illumination levels. Checking my
junk cell under random room lighting conditions, and again scaling for
size, I get:
0.333 * 0.55v / 0.02mA = 9.2K
I don't have a small thin film panel to test. (I have 90watt panel,
but that's a bit much for scaling to caliper size).

Alternatively, an electrolytic works, but gives a caliper that quickly
quits if you accidentally shadow it.

Not if you do exactly like it's done with a calculator. When the cell
is shaded, it runs on battery. A silver-oxide battery holds:
1.5v * 150 mA-Hr = 22.5 milliwatt-Hrs
and will deliver most of that before the voltage drops to unusable
levels.

The super cap will deliver (very roughly):
1.5v * 15uA * 4Hr = 90 microwatt-Hrs

There are much smaller supercaps--0.02F--used in cellphones. That's
another option / compromise. Leakage should be better too.

Overview of CDE super-caps:
http://www.cde.com/catalogs/EDL.pdf
Some interesting notes on charge time and lifetime near the bottom.

In my never humble opinion, what makes more sense is to do it exactly
like the typical solar powered calculator. *They all have one or two
LR44 batteries inside. *However, the solar cell does NOT charge the
battery. *When you turn the calculator on, and there's enough light to
run from the solar cell, the battery is essentially disconnected. When
there's not enough light to run the calculator, it runs off the
battery. *No waiting to charge a capacitor from the solar cell.

That uses the PV as, basically, a battery-extender. That's fine, but
complex--you need a micro-power switch to disconnect the battery, etc.
(A diode drops waayyy too much voltage.) That puts it out of the
realm of a simple project that can fit into the existing caliper.

There has to be a chip in the calipers anyway to count pulses, run the
display, and deal with the push buttons. Adding a power management
feature does not add much real estate or complexity. However, if
you're thinking of a retrofit, I suspect something could be done with
a separate switcher chip.

If you're into high tech, there are various energy scavenging devices
that can also power the calipers.
http://en.wikipedia.org/wiki/Energy_harvesting
With only 22.5 microwatts required, it might be possible to power the
device with a wind up key, piezo pressure, body heat, kinetic magnetic
generator, etc. *I kinda like the idea of a wind up caliper.

Windup would be fun--steampunk.

In the late 1960's, I designed and built a paging receiver, that
produced the message output on a 1/4" wide roll of paper tape. Battery
power to the mechanics for such a portable device was impossible. So,
I went to a wind up coil spring mechanism. I've been somewhat of a
fan of spring power ever since.

The "real" solution is to design the caliper to draw less current in
the first place, like Mitutoyo and Starrett. If you've done that,
solar-powering is a snap, but then, if the battery lasts years, you
don't need solar power, do you?

Agreed. It would be like a digital watch, which typically has a 10
year battery life. However, the solar cell is still a problem because
of the dark current (reverse leakage). An isolating Schottky diode
can reduce that, but then the solar cell would need to be about 20%
larger to compensate for the added loss.

Another problem is that it would be no fun. Windup calipers offer a
far more entertaining problem to solve.


--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558