On Thu, 21 Sep 2017 14:37:35 -0700 (PDT), wrote:

snip

Why not a zener or precision voltage IC?

which one?

That was my question. ;-)

with 0.1% resistors everywhere it matters.

As long as the final result calibrates correctly, does it matter what
the tolerances are / were?

There are 2 ways to get accuracy.
1. adjust/calibrate

But adjustability sometimes brings with it lower precision (noise).

2. build it accurately

Ok.

I don't know which suits you better.

Whatever works best on the day. ;-)

Long ago I used to use selected 5% resistors where 1% were wanted, but they don't stay as accurate as 1%ers.

Ok.


Avoiding anything getting hot always helps keep the errors down,

With 250uA FSD I can't anything much getting warm?

no

;-)

so anything that gets warm heatsink it, and operate parts way under their limits. Not that that will be a problem with this meter.

Quite.

So, precision reference, zener, resistor bridge?

With a precision reference and zener my meter wouldn't start doing
anything till the measured voltage started to exceed 10V, even if I
was ramping from 0 to 15V (and that's fine for a 12V lead acid).

If you had a perfect-knee perfectly accurate voltage zener then sure, but where will you find one of those?

Pass, hence the questions.

With a resistor bridge with a 1/3 split I should be able to get FSD at
my 15V but anything less wouldn't work (would it?) ... so it's going
to have to be either of the first two ... or something else?

or resistors on the input of an opamp so it only turns on once 10v is reached

Ok, but you still need to hold the -Ve end of the meter *at* 10V as
the source goes between 10 and 15V? A LM4040DIZ-10.0 works like a
zener so that would probably do.

If I've read the datasheet properly the (10V) device requires a
minimum of 100uA and a maximum of 15mA.

http://www.ti.com/lit/ds/symlink/lm4040-n-q1.pdf

So, if we ignore the meter and assume a worst case battery voltage of
10.5V then we have .5V to get 100uA so that's 4k7 and giving us 1mA at
15V. We would also have 20500 ohms (20k series resistor and 500 ohms
of the meter) in parallel with the chip feed to help ensure it stays
over 100uA.

Basically, as long as I hold the negative of the meter at the same
voltage as the positive at anything 10V and below, the meter won't do
anything. Then as the +ve of the meter exceeds 10V and goes to 15V, a
meter set to read 0-5 volts should follow ok, giving me reasonable /
best possible accuracy across the working range of the battery.

The only prerequisites are to ensure the -ve of the meter is held
exactly at +10V (from a supply ranging from 10.5 to 14.4V) and that
the meter current limiting resistor is correctly matched (to give the
250uA required for FSD at 5V PD).

As long as any reference would be suitably driving at anything above
10V and not overdriven at 15V then we should be good to go?

Cheers, T i m

There's always protection to consider.

I think about all that would work (without affecting the accuracy /
functionality) is a very small series fuse and reverse biased diode
across the input to the meter circuit ?

Cheers, T i m