I found the above circuit at:
with a slight variant at:
and finally with the capacitor polarity swapped (and different R1 resistance, Q1/Q2 BJT type) at:
According to my undersanding and the simulation I've done, the polarity of C1 should be reversed according to the last circuit version above. Otherwise C1 is reverse biased when Q2 is open pulling C1's minus pin close to 7V, while C1's positive pin is close to 0.7V. If this isn't true, could someone explain how the curcuit is supposed to work?
Also if the above is true and C1's polarity needs to be swapped, C1 would still be reverse biased when Q2 is closed, albeit at a smaller absolute voltage (~0.7V vs. ~6.3V). If so, is this 0.7V reverse bias acceptable considering that this is the normal way of operation, that is it's not an occasional event? If so, is there an ideal type of capacitor for this purpose?