I think the voltage issue is unrelated and the difference seen in temperature at the point of your electrode with the changes in voltages you could produce would be negligible.

Isn't your GTAW arc voltage also used for controlling your Z axis/AVC ?   Changes in arc voltage are going to have a bigger effect in standoff distance (between the electrode and work.) 

Those are some crazy strict tollerances..   

What on earth can a 0.005 tungsten inclusion do to the mechanicals anyhow?

A few answers

We have done quite a bit of work evaluating tungsten inclusions on SEM's, I'll say it's not outside of the realm of possibility to see missing alloy content. In tungsten it's possible for metal vapors to condensate on the electrode and "drip" into the puddle. Some alloys will form low melt point eutectics  on the electrode and seem to preferentially attack the tungsten. Thus the "drip" will not necessarily be 98-2 W-TH  I've seen one or two odd SEM results in which the tungsten inclusion had a very high concentration of a base metal alloying agent or no thorium.  Mind you this was with automated welding where dipping was never a likely possibility.  I would put it in the round of possibility however.  Sounds like the spec is very tight, I no nothing of the service conditions, but usually we go a step further and characterize flaw major/minor dimensions, I.E a drip in a fill pass is okay because it won't have sharp corners. An electrode tip is rejectable as it creates a big notch. Different code though.

Voltage  does not correspond 1:1 with electrode surface temperature. Current density has an upper limit, after a certain point the plasma column will simply rise up the electrode and cover a greater area.  I believe this is a function of electrode potential and temperature, as the temperature of the electrode increases the resistance goes up at a certain break even point it's easier for the electrons to flow from an area further up the electrode as opposed to trying to come off the very hot tip.  If it wasn't for this theory than electrode size would not have much to do with current carrying capacity, the thermal sink is relatively small between different sized electrodes but the surface area increases exponentially.

Hi Brett

Have you had some answers from further study? Just something from my side:

The SEM analysis is not capable of finding light elements, so if your particle was Tungsten carbide (e.g. from abrasive) then the analysis would not show that, it would suggest it is W. Just see if you are able to rule out this posibility or not, as it may send you off on a needless tangent. The doped W electrodes (e.g. Th) mostly start "splitting" if the current density gets too high, so if the electrode tip looks good, you should not be loosing significant amounts of it into the weld. I assume you are using a water cooled torch to give your electrode the best chance of staying intact.

Let us know if you have further info.

Regards
Niekie