With the help of a DVM and the circuit diagram I sorted out few things on the Saba Compact Clock A.
One trace is hidden under the display support frame, and probably it even contains a diode. I cannot unplug the custom IC to look what is underneath it, so I must trust my DVM readings.
The IC does work, since it outputs voltages compatible with a blinking display (typical of "time not set" state), but there's something wrong on the path of the common cathode I couldn't sort out. On the positive side I could light up segments on the display and confirm they are plain red LEDs with a common cathode, so I can re-wire my own cock circuitry if I want to.
I also spotted a curious optimization: segments a-d-g of tens of hour digit are connected together. That makes sense since they always light up together in numbers 2 and 3, and don't display anything higher than 23 (hours) or 31 (day of the month).
Now I might try to replace a couple electrolytics badly corroded (externally) and then give up on fixing this device, for now.