I also have a 8,7 µF 300V AC dry-type capacitor which came from a shorted out H38 ballast. It would probably work.

This will deliver ~0.4A into short, i guess not good enough...

In this case what could I do to have a 120V 5A ballast?

What i wrote:
- Or use high power (800 or 1000W halogen) incandescent lamp as resistive ballast,
- Or the MOT's primary with the secondary:
  - Open for an inductive ballast (i have no clue, what current it will yield)
  - Connect one MO capacitor (0.86uF/2200V) parallel to the secondary (you get 10A minus above; usable, when the inductive current is about 3..5A)
  - Connect serial combination of both of your MO capacitors, you get ~5A minus the inductive current is very small (below 2A)

Other idea might be to supply your lamp by DC, it would yield much more stable arc:
Take a 10A rectifier bridge (with appropriate heatsink; expect about 10W losses here)
Connect it's DC output in series with the MOT primary to the arc lamp (the MOT serve there as a filter maintaining the lamp ionisation during current zero-cross) and for first trials add a DC A-metter
Connect it's AC input in series with the high power lamp (~1..1.5kW; will be underdriven) to the mains.

OK thanks for your help! I will connect the 0.86µF capacitor to the secondary and use the primary in series with the load (kind of like a regulated lag ballast?). Should I expect an overheat?

It is not regulated. It is simply "serial reactor lead ballast"...
But try few combinations, not only one...

I have been playing with MOTs and resonant capacitors lately, and I found an interesting behavior - when you connect a capacitor in parallel with the secondary winding, the following happens:
When you start increasing the primary voltage from zero, the current rises quickly because of the series resonant action of the capacitor on the secondary (actually it's a bit off-resonance with the shunts so the resonant rise is not that huge).
At some voltage on the primary, the secondary output reaches the saturation limit, the "Q" of the circuit starts dropping and on further increasing the primary voltage, the current actually DECREASES!
With full mains voltage on the primary (230V), the current is only around 2A, the same as if there was no capacitor!

So I'd say this method is not really a good way to ballast a high pressure lamp...

I have 2 things to add:

- Bluelight: If you get 2A at 230V, then I should have more current at 120V, like 5A.

- What I want to ballast is a carbon arc, not a high-pressure lamp.

Otherwise I'm ready to test the serial reactor lead ballast. It features a 0.80 µF capacitor connected to the secondary of the MOT. The primary is connected in series with the load. The whole thing is cooled by a 12V DC fan supplied separately for the moment (The ballast is contained in an empty computer power supply housing.

By this test I will check 3 things:

- The average carbon rods life
- The electrical characteristics of the ballast
- The average temperature of the ballast while working.

I'll reply to this topic tonight when the test is done. The final result and the points to improve (if necessary) will be given.

I have 2 things to add:

- Bluelight: If you get 2A at 230V, then I should have more current at 120V, like 5A.

- What I want to ballast is a carbon arc, not a high-pressure lamp.

Well, you can try it, nothing can go wrong with carbon arc  but I think the arc will be unstable.
(sorry I have mixed it with another thread with lamp ballast experiments)

But I think the arc will be unstable.

Shouldn't the capacitor stabilize the current going to the load? (Like a regulated lag would do)

Shouldn't the capacitor stabilize the current going to the load? (Like a regulated lag would do)

Yes, that will happen on the secondary side, but not the primary. Remember the transformer in this circuit does not transform neither voltage nor current  (because of its shunts and "resonant effects" with the cap)

Well, just today I learned a LOT about transformers, capacitors, power factor etc.

I thought the capacitor would give a smoother current, but apparently it won't. It'll just reduce the current, I'd like at 5A.

OK here's the final result:

With one of the two microwave capacitors I cannot start any arc. With the H38 capacitor I make damn long arcs! With both microwave capacitor in parallel, I get a bright and short enough arc. However I have issues with MOT overheat. So I'll install three heatsinks on the MOT with their individual fan. I'll also need a bigger housing.

With the H38 capacitor I make damn long arcs! With both microwave capacitor in parallel, I get a bright and short enough arc. However I have issues with MOT overheat. So I'll install three heatsinks on the MOT with their individual fan. I'll also need a bigger housing.

Where did you put the capacitor and whats it capacitance?

I guess you put it on the secondary and its capacitance is much larger than the microwave cap? The effect of this would be almost the same as if you shorted the secondary directly... Well you can try.

The total capacitance of the capacitors is 1.66uF. The capacitors are in parallel and directly connected to the secondary.

The total capacitance of the capacitors is 1.66uF. The capacitors are in parallel and directly connected to the secondary.

OK that could do something.
I hope the voltage rating of the H38 cap is at least 3000V, otherwise its now shorted and act as a piece of wire on the secondary