Wight, I like the way you are taking this.
Here are some suggestions for small tweaks to the layout:
1) Z1 could also be stacked on top of C1, if space is tight, maybe the layout could be shuffled a bit to make room for it alongside, with correct footprint.
2) Value of Z1 = 4V3 sounds a bit low to me. With a high voltage cell it could be conducting. Suggest 4V7 instead.
3) Footprint for 7135 could be re-done to include correct pad beneath tab, e.g. see the SOT089 footprint drawing at http://www.infineon.com/cms/packages/SMD\_-\_Surface_Mounted_Devices/SOT/SOT089_xSC62x.html
This would help improve thermal management. This part dissipates 646 mW with led Vf 2.5 volts, 380 mA, 4.2V cell. With a zener mod, running at 8.4V, and e.g. two LEDs in series, this could increase to 1.3 watts, which is a LOT.
It would also improve reflow soldering as the part could “swim” properly in the solder and align itself better.
4) The hole in the middle for spring bypass is nice, but I see that there is nowhere on the top of the PCB to attach the LED positive lead, if a bypass is not used. Also the only tracks connecting to it are narrow, not suitable for carrying such current.
I would like to use a solid brass post, rather than a spring, so it would be very awkward to attach the LED +V wire.
Suggest expand copper around the centre hole as much as possible, to provide some area that could be used for soldering a wire onto the top, with hole in solder resist to define this point.
5) Also add some more vias to connect this enlarged area of copper to the +V contact on the bottom, instead of just relying on the plating through the hole. Make at least one big enough to solder the LED +V wire through it, if desired, for lowest resistance. Maybe near the top left corner of R4.
6) Also consider adding a through-hole in the middle of the LED -V wire pad, so the wire could be soldered all the way through. This would require an isolated area on the bottom layer. Maybe keep it covered with solder resist (which could be scraped off) to avoid shorting against large springs, or the plated hole could be drilled out to isolate, if required.
7) If these two holes were available they could also be used to fit a large leaded Schottky flywheel wheel diode across the LED, if this turns out to be a good idea.
Regards,
Sharpie.
PS: I think, if the Zener plus C1, plus R3 work as I hope, that C1 = 100 nF decoupler would be sufficient, in fact better than 1 uF. R3 should not be shorted, some resistance here will reduce the current into C1 and Zener caused by the spike, reducing “pump-up” of C1. I.e. the combination will form a filtered supply rail for the MCU.
PPS: I’d suggest a starting point for the “semi defined” OTC might be 10 uF, with R5 at 100K.
I think it should be possible to increase the values of R1 and R2 to reduce parasitic drain, maybe from about 140 uA at present to maybe one tenth or even less. Just multiply the values by ten, or even more.