"Richard Smith" wrote in message ...
....
Or cool to lower temperature and pump the cylinder when weld area has is
showing -40C by thermocouple.

-------------------

That's the first method I would try, to check your assumptions and procedure
and shrapnel containment.

This suggests that impurity levels might be a significant uncontrolled
variable in your experiments:
https://www.tf.uni-kiel.de/matwis/am...tr/s9_1_1.html

Is -40C the lowest you expect in European service?
https://en.wikipedia.org/wiki/Operating_temperature

Some industrial temperature controllers have a cooling option that can
operate a solenoid valve to regulate the temperature with evaporating CO2
from a tank of room-temperature liquid.
https://www.linde-gas.com/en/process...ing/index.html
"Liquid carbon dioxide converts to solid carbon dioxide snow at €“79ºC
(€“109ºF)."

If the test chamber is nearly sealed the escaping gas should reduce or
prevent frost covering the sample. US regulations required an oxygen level
sensor and alarm at the test station. The body's breathing mechanism
responds to CO2 level (pH) in the blood, not oxygen level, so you may not
realize you are about to pass out from anoxia, especially with liquid
nitrogen.