OpenRPNCalc

• The case is relatively thick (12.5mm total thickness). This is limited by the chosen CR2032 battery holder which is itself 5.5mm thick, and of course it should be placed on the back of the PCB, so one should add the thickness of LCD/switches to this value. A better solution would be to make the cutout in the PCB for the battery and find a thinner holder, but I failed so far to find the good one. Alternatively, I could use two smaller 1.5V batteries placed on the front side of the PCB. To be investigated for the next iteration of PCB. 

Here are few issues related to electronics: 

• The current PCB lacks a few filter capacitors. Normally, STM32 manual recommends a 100 nF capacitor for each VDD-ground pair entering the chip. I had to solder two capacitors in an "ad-hoc" way, PCB needs to be fixed to add them. Also, they could be placed closer to the STM pins. 
• 5V booster for the LCD is controller by the 5V_EN signal. The idea was to switch the 5V power to LCD completely when the calculator is OFF. However, it does not work like this due to my wrong understanding of TPS61222 logic. As it is implemented currently, when the calculator goes OFF, LCD receives 3V from the battery instead of 5V (the booster is switched off, but still supplies battery power). This has to be fixed by adding a transistor to decouple the TPS61222 and LCD from the battery. 
• By the way, it appears that Sharp memory LCD works nicely from 3V instead of nominal 5V as indicated in datasheet. Maybe 5V booster can be removed completely? 
• The circuitry around TPS61222 does not follow exactly the PCB design pattern from the datasheet (conductors are not thick enough, inductor and capacitors are not close enough to the chip). While it works stably, the PCB should still be improved. 
• A couple of filter capacitors are missing near the LCD connector as well. To be fixed.