Thanks for advices, I will get contact cleaner soon, and take care of those switches.
I fixed channel 2 issue - it was just that bloody R900 that burned(input resistor, 10 ohms, it was between BNC and AC/DC/GND switch).
It's resistance was arround 40k, so just not enought to make obvious difference at DC or lower freuquencies.
I still got no idea how someone could burn that little guy.
Timebases that don't work are:
500ms, 5ms and 50us.
Those timebases correspond to positions 3,9, and 15 on time/div switch.
At first I suspected that 2M37 resistor (R1, fig. 4 in manual), but then, 20us should not work either, but that one seems to work fine.
Still, I will have to measure it, becuase some timebases, including previously noted timebases seem to be...non linear, with signal periods geting shorter towards right end of the screen. Not an huge issue,since i don't plan to use this scope for any kind of precise measurements, but still pretty annoying.
Expect there to be subtle (and not so subtle) problems with electrolytic capacitors. They "dry out" with age. I had a Tek 485 which would take 1 minute to turn on, but if turned off and on again it started instantly. Difficult to fault-find a problem that only occurs twice a day, but I traced it to an electrolytic that took in the SMPS startup circuit.
There's lots of help available on the Yahoo TekScopes group.
Similar case here - if I turn off scope, and then turn it on few minutes after that (obviously, I'm avoiding doing that), all timebases work instanteniously. I guess that I will have to check capacitors too, but still, bad supply capacitors would probbaly cause all timebases to be dead.
I think that somehow most electrolytes are still just fine.
Another issue I ran at is somewhat bad triggering of squarewaves at frequencys higher than 30-ish kHz.
I added picture in attachments. Traces seem to have this "afterglow". Not shure if this is normal or not.
I'm using Arduino uno as squarewave generator, since I don't have propre func. gen
I will have to hack together some sine oscillator to see if similar issues appear with other waveforms.