Well, it works but needs some tweaking. Burned up a full gallon of fuel getting through two rods, without much wasted run time.
Here's the whole scketchy set up now that for the first time I'm able to fool with it during daylight hours.
The test: Butt weld of rusty sawn end to rusty torch cut end, typical of actual conditions where nothing is cleaned up before use. The 6010 5p+ rods really work well for this.
Here's the result, the arc length is really short and very difficult to strike, though it has plenty of heat. That means low voltage to me, and probably too many amps for the voltage. When I got a good run going and kept the rod shoved hard into the puddle, I could keep an inch of the pipe glowing dull red in the sunlight. If the least bit too much flux built up, it would snuff the arc just like welding deep with 7018 rods and using too little amperage.
After cleaning the flux. In some places if I could keep the arc lit it had no trouble getting full penetration into the joint of 5/16"-wall pipe, other places when I was fighting the short arc and not able to put steady heat into the puddle the penetration was more like 1/3.
So it looks like I need to maintain higher engine rpm (had it going over 2500 engine rpm to get the best welds and only getting 46 open-circuit volts when things got hot, thats 10K weldernator rpm) in order to get the welding voltage needed, and then install a rhreostat to turn down the field voltage a bit and get the amps down a little. Also, the diode bridge will definitely need a heat sink as it gets blistering hot after only one rod. The reactor seems to be doing ok (ark is a clean hiss when going good, the amps seems stable, and the startup voltage isn't blowing the tip off of the rod) but is probably half the millihenries that I need.
Here in a bit I'm going to give it another go and get my wife to watch the voltmeter while I'm welding to see how bad the voltage is dropping off. If it won't stay up around 24-28 volts under load I might have to re-think this whole thing.