and there is also this:
http://www.analog.com/en/interface-isolation/protection-products/products/index.htmla bit more pricy then a zener diode... I like your creativity, the chinese are creaming their pants right now.
this sounds like a interesting experiment at the very least, can you hook up a TVS to a lm317 current source and forget about it for a few months?
I thought that TVS were primarily damaged by over heating (my rudimentary knowledge of diodes), so if its cool you are fine. The only thing I would worry about is spot heating, its possible (but i would imagine unlikely) that there is little attention paid to spot heating problems during the design of these diodes, so a low continuous current could damage it... i wonder if you can decapsulate it and look at it with a thermal imaging microscope (i wish right?
)
There might be some design benefit towards making it in a way that leads to dangerous spot heating, but hell if i know what that could be. (fear mongering)
or that it being hot could cause over voltage conditions to damage the material more then if it is cold..
perhaps you should see how the reliability of the TVS changes over temperature conditions, picking a high temperature version of your part might be a good idea.
the problem is that, even if you measure case temperature, the junction temperature might be much higher, so if yoou measure the case at 40c, you might actually need to look at the temperature reliability of a diode running at 80c (just example numbers here).. this is where decapsulation and thermal imaging would be helpful..
because if the ambient is 40c, the junction will be 40c (the whole part is uniform temperature). under a load the junction might be heated to 60c while the case is 40c,
trap for young players. , you should look up the reliability at your actual
highest junction temperature, not case temperature. I would imagine the whole diode structure would be of a similar temperature, but I don't know jack about doped junctions lol
i think that determining the reliability of this is not trivial. cost savings like this might cost you alot more money (and reputation) if failure happens.
on a personal note I have a beef with shitty input protection schemes, like PTC's in shitty switch mode power supplies burning out (i have seen this many many times). I think that many designers share the same sentiment of not being cheap with protection circuitry.