Figured out i parted out the wrong Philips SL18. I had two: one worked fine, the other produced a horrible burnt electronics stench.

I had screwed my 'good' one into a socket and after a while... stench. Ugh.

I've now opened up the other one, but i can't really see anything wrong with it. It just runs very hot.

I kept the ballast of the other SL18 so i could conceivably swap them out. But it's hard to get access to it, without damaging the E27 socket. Either way, it's now running without the glass prismatic cover, so i can get an impression of what is running so hot - the ballast or the tube itself. The glass cover would get so hot, that i couldn't hold it for longer than like 5 seconds. That's not normal for an SL...

Might have to build a ring tester to see if the stinky one has a shorted out winding in the ballast. I hope not. One thing's for sure - i can see that the capacitor is undamaged.

There is also another problem with "reducing CWA power by changing the capacitor":
Normally these ballasts are designed with the inductive component in series with the capacitor to be current dependent (shunt saturates, so reduces the inductance when the current goes higher) and by that detune the series LC so the current stays where designed. This current is dictated by the point on which current the shunt saturates. This forms very strong current stabilization mechanism, which suppresses very broad mains voltage variation, as well as to big extent even the capacitor tolerance.
If you just reduce the capacitor, you decrease the lamp power, but the shunt won't be saturating anymore, so the current stabilization effect will be gone.
Does not have to be that much of an issue when the mains voltage and all components stay within tight tolerance or when it is used just to dim a higher power lamp where the tolerance for the exact operating power for that settings is very wide.

HX ballasts are quite common, depending on the application.

 I killed my Sylvania 800w Retrofit lamp by running it from a CWA ballast . It ran for about 10 min. before it started cycing .

Thank you all for your replies! I am new here and I am just discovering how helpful the community here is.

Thanks for letting me know! I was ready to try it but now I won't. If only HX ballasts were more common... 

Thank you so much for letting me know! I didn't know that the type of impedance mattered.

Update: i tried fixing it anyway. Replaced the emitter resistors, the fuse and the transistors. It still doesn't work. It does oscillate - i hear the interference whine on the radio -  but nothing appears to be happening with the tube. Need to do more measurements...

Erythemal radiation and germicidal radiation are two different things. Germicidal radiation (254nm mostly associated to low pressure lamps) just causes strong skin/eye burns and does not induce suntan. It belongs to shortwave UV-C range. Erythemal radiation belongs mostly to UVB and some short end of UV-A, it causes milder sumburn in large doses and induces suntan and vitamin D synthesis.

Yes, regular glass envelope filters out most of shortwave radiation of a mercury lamp, UVA part generally passes but is not considered harmful, though it induces yellowing in many materials.

Built a focussing system for the Eye H50 at work, that's used for hexane or hexene and bromium to show subtitution reactions.
The Narva NF80-in-a-can is a bit more powerful and also potentially overdriven a bit (though no blackening of the arc tube is visible despite a good 15 years of occasional use), but has no system to concentrate the light or use any light radiated from the back - so with a bit of luck, the reaction happens as fast with the tiny 50w lamp, as it does with the 80w one.