Ideas to get a controlled temperature of around -40C (also -40F) in a
test sample?

"Robustly" by physics:
100C - boiling water
0C - ice
-17C - brine-ice slurry (?)
??? -40C ???
-80C - "cardice" - solid CO2
-196C - liquid nitrogen

Reason is, found a way to tensile-test fillet welds, and so far always
seeing breaking strength come out at around 560MPa, when you do the
maths relating breaking force to the fracture area.

The 355MPa yield of the Rectangular Hollow Sections (RHS) isn't seen -
and I know they have exactly that yield stress from beam bending
measurements.

Here's the tests - "Alladin's Cave" of misdemeanours and skulduggery ?
;-)

http://www.weldsmith.co.uk/tech/struct/struct.html
"Steel Structural Performance index-page"

Various tests.

Specifically the fillet weld tensile tests

http://www.weldsmith.co.uk/tech/stru...4_fwbeamt.html
"Fillet welds tensile tested in beam test"

http://www.weldsmith.co.uk/tech/stru...t_testrig.html
"Tensile-test rig for beam-configuration fillet-weld samples"

Movie of - 10 seconds - shared on "Dropbox"
https://www.dropbox.com/s/esgwfk5jen..._fwtr.mp4?dl=0

Yup, I know that as a PhD level scientist I have to work as a welder
because everyone already ensconced in "office engineering" jobs
manicured in their white shirts sees me as a "Dennis Hopper like"
(think eg. "Blue Velvet" (1986) film) character of the science and
engineering world ;-)
If you keep the office door closed to all but your own manicured kind,
you can keep reality out. Who can blame them if no-one comes and
hurls them out into the cold hard world their scheme avoids? :-)

It would be helpful to see whether that "no yield event - straight to
local fracture at high(er) stress" is associated with a low
temperature brittleness charactistic.

I could "dam-off" the RHS close to the weld and at the far end, and
fill it with a cooling fluid.
Ice-brine looks good for -17C.

Throw a blanked over the entire sample for a while for all parts of
the sample to be at that temperature, then slide in the hydraulic
cylinder and "pump it up" and see what the temperature causes or does
not cause.

What about for -40C
???

Regards,
Rich Smith