[oil pan 4]

Other power factor corrected 240 volt transformer welding machines run 40 to 60uf of capacitors connected line to line. So the amout of capacitance will be some where around there.

The main transformer runs 3.77 amps no load.
First try was with the 10uf capacitor since I already had it sitting there, result: it dropped the power consumption to 3 amps even.
Next try was with a 25uf capacitor, that dropped no load current to 2 amps even and quieted the angry hum the transformer makes.
Still thinking there is more to be had I connected the 10uf capacitor to the 25uf, making a 35uf capacitor, that reduced power to 1.71 amps.
Then swapped out the 35uf assembly for a single 40uf capacitor, that increased power consumption to 1.76 amps.

So 35uf of capacitance saves around 535 voltamps at no load.

For testing I am going to connect twin 440volt 40uf caps line to neutral, since arc welders have a way of killing 440volt motor run capactors connected line to line.

Weld output appears unchanged.

Connecting the capacitors in parallel, running each capacitor L to N seems to lessen their capacitance, causing the machine to draw 2.2 amps at no load as opposed to 1.76 amps during the quick and dirty L to L single capacitor test.

Twin 55uf capacitors connected L to N seemed to do the trick.
Twin L to N 55uf capacitors plus a 5uf for the fan is almost exactly what welding machines running PFC use.

It would seem that when you run them at lower voltage they have less capacitance. Or has the effect of serriesing them up.

I ordered a set of 80uf motor run capacitors to install L to N on each line. The L to N connected 55s are causing the machine to draw around 2 amps. The 80s should get that down to around 1.75 amps.