As close to the saturation temperature for the max operating pressure as you can get the PID to not inadvertently trip. This will be a cut and try adjustment to get the machine with a close set point. If you have a sight glass this will work much better. There is an optical sensor that passes an IR beam through the sight glass and the refraction of a full glass at the beam point deflects it away from the receiver, when the glass is empty the beam reaches the receiver. They will fit the usual diameter sight glass. I looked them up once but it will take some time to find them again. These are basically fool proof and reliable as it is water level you are concerned about as when relying on increasing temperature you are already in trouble.

Interesting idea on the sightglass. There is one on my Conti Prestina. I guess it's economics too. The additional relay for the PID controller costs $9 before shipping. An optical sensor could be much more.

Optical sensor was within the realm of reasonable when considering the repair cost for an out of water event. I will look for my note later as I had prices. I think a sensor on the dip tube will be a reasonable low cost solution though.

I believe if I make the cutoff temperature 15f = 9.5c above my highest brew temperature that should be protective enough. Does that seem about right to you? Here are photos showing my thermocouple placement. The first shows the location of the dipper tube compared to the heating element. This was taken during disassembly.



The second shows where I tied the sensor to the dipper tube.

I would go with that temp delta and position the TC so it gets max exposure to the element, run parallel and above element as best you can make it. You want to set the TC so it sees the most element surface as the element will get red hot promptly so this is a radiant energy transfer if the element is dry. The conduction/convection mode will be slower and in part remember steam is a lousy heat transfer medium.

That thermocouple only senses heat at the tip of the wire. Hmmm. If it is very close to the element then steam might not get hot enough to trigger a shutdown in time. Maybe the delta would need to be closer and then it's still a best guess. The way the PID is tuned I don't get more than about 3 degree F variation. So perhaps 6F above brew temp should shut it down. I will be interested to see the optical sensor setup since that would be measuring the critical situation accurately. Otherwise a thermocouple attached to the heating element would offer an adequate safeguard.

The element does not heat up until it is in vapor instead of liquid so there is some conduction/convection but usually the get up to 900F pretty quick when in just vapor. The radiant energy will have an almost instant impact on the TC. as close to normal as possible without false tripping is the desired set point.

If I were designing heating element protection, I'd look at using a current sensor and then cut off power if there is a drop in current flow (there is more resistance/less current as the element overheats). Easy to do once you have a micro-controller and a SSR in a design.

That is true if the element is intact and not in the process of shorting to adjacent turns for other reasons. Not to mention it will increase the initial cost of the unit and as complexity increases the failure rate increases.