On Mar 13, 5:51*am, Andy wrote:

[about a 5 mV signal at a distance from the measuring
node]
The signal is DC from a instrument, a pyranometer.

That isn't the RIGHT information; what is the output impedance?
If it's like a thermocouple (one kind of pyrometer), coaxial
cable is CONTRAINDICATED because the outer shield
and inner conductor will have different temperature lags
in case of a chill wind on the apparatus.

If it's a thermocouple, with low impedance (under 10 ohms),
then unshielded twisted pair will probably work fine; don't
ground ANYTHING (the high temperature insulation is
failure-prone anyhow) except next to the meter.
Use Cat5 wire (it's cheap) and consider dedicating
an extra pair, for preamp power, at each node.
Either a splice, or a preamplifier, requires a small
box for protection. Small meaning the size of
a connector (print server computers inside a
DB-25 connector shroud are not uncommon).

If it's a temperature measurement, with slow response,
the measurement end can take several readings and
average (or even integrate over an exact interval) to get
rid of AC errors; or you can apply an overall grounded
shield to reduce AC pickup.

Signal sources have four important characteristics, GIVE ALL
FOUR YOUR ATTENTION. Signal amplitude (5 mV);
source impedance (totally unknown); bandwidth (DC);
resolution (totally unknown).

Of the four, we answer-ers only got one (the 5 mV), and
a half (if it were REALLY DC, you could measure once in
January and just refer to that number until June).