On 9/26/2017 5:40 PM, T i m wrote:
On Tue, 26 Sep 2017 14:13:20 -0700, mike wrote:

snip

Yeah, ok ... but what about the question re using a lower value zener
to get it away from the 'knee', especially as it's no more difficult
to do?


Since we're just trying to be clever,

I'm not mike, I'm just trying to refine the design of any solution to
be as good as possible with the fewest possible components. ;-)

Try this:

https://i.imgur.com/irLisN7.jpg

The transistor and zener make a temperature compensated ~7V
reference (must use 6.2V zener with standard tempco).

Ok.

Two pots set the min and max voltages.

Noted.

Rk allows you to add
current to the zener to get it past the most nonlinear portion
of it's V-I curve.

Understood.

This circuit has the ability to calibrate out
almost any component variation.

That sounds like the sort of thing I was hoping to hear. ;-)

It's relatively temperature
independent.

As long as it will cope with the -10 to +10 DegC range it should be
fine here in the UK. ;-)

You can use any meter that will make full scale at
less than about 7V.

And that's most small panel meters I'm sure.

Downside is that the pots are vibration sensitive
and subject to moisture/corrosion.

Sure, but in my case they won't be subject to any real vibration and
will be in a sealed enclosure. Also I would typically use a quality
multiturn pot of the smallest value possible to give me sufficient
'trimming' around the end points and quality resistors to do the main
bridge / work.
Use a small value pot with a fixed resistor on each end.
Easier to set, more stable.

Are we having fun yet?

Well, it's certainly bringing some interesting and varied solutions
out of the woodwork. ;-)

It might also help me use up my stock of BC107's. ;-)

Cheers, T i m