Hi Micky,
in fact I was in doubt if you were talking of plans or of your prototype. As you mentioned the Transil, that I also found in your schematic, I supposed the second.
Where do you get 3.5V from, don't you use batteries?
Decreasing supply voltage will affect the quartz frequency considerably.
1.3 mA is really heavy for an electric table clock. I knew that the electro-magnetic system has a poor efficiency, but did not expect that poor. Then you are not forced to save current in that extent.
I am focussed on ATOs mainly, they take about 40uA average, so I am not really satisfied with the 35uA that my device consumes. Here a watch quartz is a must. And you can do a lot between two contact actions if you use assembler (control the rate, measure and control the amplitude of swing, provide a rate tester etc.).
Now to your question. Problem is the reverse polarity of the surge.
My idea: a transistor switch on high and on low side (like your U1) of the clock terminals (coil). They must open at exactly the same moment to free the coil. A diode from O/P_0V to Batt+ and a diode from 0V to O/P_+ will feed the surge to C4 and charge it.
If the result is worth the effort must be investigated. In any case I am interested in your results.
Regards,
Frank