So you bypassed the governor? Then no wonder, your engine overrun (and kill capacitors) by load disconnect...
Because the governor should normally steer the pump control rod: Spring action toward Full Open (maximum dose), centrifugal force in the governor mechanism should push (or pull) it back toward Full Close (no fuel injected). The equilibrum should then lead to max fuel dose at rpm slightly below "SetPoint" and slower, full fuel cut-off at speed slightly above the "SetPoint" and faster, in the range between the fuel dose is gradually reduced, then with any load the speed should settle within this small range and should stay there, regardless how the load change (unless it goes above the maximum output power).
So if the lamp burn (= reduced load), the speed should then change only a little, as the governor should, as the response to increased speed, reduce the fuel dose, what should then stop the speed rising. The whole speed difference should be then noticeable only by instruments (so the voltage should have not to rise so, it kill those capacitors)...

Thankyou for that, I`ll have to have another look at it , I`ll post pics if all works well this coming weekend.

Ok, so my 3 x 500watt MBFT project didn`t go to plan.
Having now sorted the governor, I managed to only light 2 of the 3, when switching the 3rd on, it sapped the power from the other 2, causing them to exstinguish!

Then I tried lighting all at once, and this just sucked all the life out of the generator , before they lit.
I dont think I have big enough capacitors on, (3x440volt 8uf), however, if I put a larger size on, it means altering all the pully sizes to suit engine speed.

Glad to see the governor was fixed... :-)

For slight (+/-25% at least) speed changes you do not have to change the pulleys, it should be well within the range of the governor adjustments...
If you operate the generator in it's saturation, increasing capacitance would not increase the voltage as much, but definitely will increase the power capability of the generator.
For capacitors you mention the compensation is abut 3x400VA (at 50Hz), so the maximum power (assume resistive load) would not exceed this on most machines.
Moreover the SBMV consume ~30% more power during warmup then at steady state, what mean for 3x500W you should have setup capable of 3x650W (and you only have 3x400W and with the assumption you reach the 50Hz).
Try connecting lamps first and only then start the engine. Don't forge to set the governor first (without the load)...
But 2x500W shuld be OK for the show too, you do not have to keep the load really balanced, it is not needed. The generator itself is well capable to keep all voltages similar...
By the way many induction generators use two windings with 90deg phase shift: One for excitation capacitor and second for load connection.
It allow the excitation capacitor to be phase-cut controlled using a riac, what then allow to regulate the output voltage, as the phase-cut regulator create quite severe voltage surges to the winding, but when this is 90deg shifted, these surges do not couple to the main output winding.

Try connecting lamps first and only then start the engine. Don't forge to set the governor first (without the load)...


Ah, now heres the thing on that, this type of converted motor will not start to generate under load, you have to, get the voltage first, then add the load.
Another display I have thought about is running MV, from 50watt to 400watt, i.e, 50, 80, 125watt, on phase1, 250watt on phase2, and 400watt on phase3.
I have run all these wattages on their own, but not all together, and I do not get the voltage drop off with an inductive load as much, it also seems to bring the Hz up too.

Ah, now heres the thing on that, this type of converted motor will not start to generate under load, you have to, get the voltage first, then add the load.

It would not start with resistive load (and with uncompensated inductive it would even not run), but SBMV do not form any load unless they ignite (the circuit is not complete without the burning arc), so till the vlotage rise to ~100..200V.
So with discharge lamps the generator would start always unloaded (no arc => no load).
When reaching critical speed, the rotor field and so the output voltage start to build up (exponentially) and when it rise to the ignition level of the lamp, the rotor field would be already quite high to keep running.
And when this voltage is present, it is able to continue with the load too...

The same would be with regular MV lamps: Initially it would see only capacitors and only when the voltage build up, it will ignite.
Note, then reactor ballast have to be power factor compensated, as the inductive load type would reduce the rotor field, so the output voltage.

Interesting, I hadn`t thought about it like that , I was looking at putting the PFCs on as so.....
1st Ph, 50,80,125watt = 20uf
2nd Ph, 250watt = 13uf
3rd Ph, 400watt = 20uf

It is not really necessary to exactly compensate each phase separately, but overall compensation VArs should be correct. Any mismatch you will see as wrong voltage/frequency combination.