Actually 1 ( i have 2 pieces, but i think i can repair) ruined ballast for the voltage pulses. Poor filtration, and a bit too high voltage (420-430 with some spikes around 40V sinus, terrible until i took my oscilloscope)
But blown mosfets and capacitors were only at power inverter.
Actually it is not designed well, it must be upgraded. What i did to inverter - i replaced mosfets with some IRL series 3705, not a big choice of FETs in my country.
Snabber RC connected to the drain's ( first ~ of a transformer, push-pull design)
Snabber parameters- 2 ohm resistor and 30nf capacitors(USSR works better that chinese, measured with the oscilloscope voltage spikes)
The problem is, when the upper mosfets closing it creates voltage spikes, inductivity of the transformer. And they were very high, and some high frequency ringing . Now the spikes are only a little bit higher than main voltage and absolutely no ringing. Also no ringing or spikes at 320 V powering mosfets, filtered with the coil and film capacitors. That reduced heating of the fet's greatly, but some heating still there, especially on the first low-esr silver-tantalum capacitor ( ЭТО-2 15 V 400uf). But no blazing hot surfaces. I added good radiator to the ballast radiator and installed 2 fans creating enough air blow to cool down components.
The only problem is to find very high quality bulb with the electrodes pointed exactly to each other and 0.8 true arc gap, precision portion of halogens(bromide-mercury) Extremely clear quartz glass.
And for the first run with the high power setting should be at least 30 min to reduce risk of the arc jumps and blackening.
For what happened to my bulbs is:
Only a few 2 min runs, due to no high-capacity battery setup - failed tungsteen halogen cycle.
Several attemps to strike the arc when bulb is hot, but no stable arc - off within 3-4 seconds, blackening suddenly appeared.
Due to low quality halogens and tungsten with a poor glass - overheating started in the arc chamber, quartz crystallized, consuming tungsten, mercury, whatever.
Now the tungsten-halogen cycle cannot clear the arc chamber, even after a long burst - the blackening only grows, until the bulb will fail/boom( unwanted due to very good reflector, don't want to ruin,)
The electrode shape - was flat surface, instead of original osram/philips shape. I installed new bulb and after 15 of 150W run, suddenly arc jumps appeared. I think if there will be no blackening and quartz crystallizing, i will wait until the right shape will be formed. But i'm afraid i will ruin reflector with the blown bulb. The 25$ and 6$ silver not the big deal compared with a very high quality reflector/base.
IF you want some help for the boosting your ballast... I think it can be done with the reading datasheet of controlling IC, which resistor of feedback loop to change.
But the actuall improvement must be done with the transformer, as they are almost at their critical current setting. I thought of buying some of "100W" ballasts, but i realised that i must build my own ferrit-core transformer and replace mosfets + may be IC levels can be reached so it cannot drive mosfets any higher. And the current shape is not like in philips, so flickering will appear ( it just matter of time), if 150W+ setting will be done. With your 80W setup, i think no significant arc flickering will be, due not enough current in arc to reform the electrodes. I think you found the bulb with the good portion of mercury-argon-halogens, and electrode initial shape is good so there is no flickering even if the form of automotive ballasts pulses is not perfect for our bulbs. BTW, even if the lifetime will be reduced to 300+ hours it is acceptable for us, i believe. Because each strike vaporizes tungsten ((((