greatfet-hardware / jasmine Goto Github PK

It is marked 6 V on the silkscreen and in the schematic.

There are parts on the board (and on Azalea) that do not support voltages higher than 5.5 V. Azalea's max input rating is 5.5 V.

VBAT (the pin on the LPC4330 that supplies the RTC) has an absolute maximum rating of 3.6 V, but we are connecting it to the LiPo cell. The datasheet recommends keeping VBAT at least 0.2 V below VCC, so I think we should add a small 3.0 V regulator.

The PCA9674's "quasi-bidirectional" outputs can drive pins low but not high. We either need a different I/O expander or need to wire it up differently.

The registry says Jasmine uses 0x3D, but it seems to actually be using 0x25. I'm thinking the registry should be updated, but perhaps we should check the design to make sure 0x25 was intended.

The blue LEDs in #23 have a forward voltage of 3.9 V, but they are supplied by VBUS which can be as low as 3.6 V when operating on battery power. They fade out as the battery is drained (which might be a cool indicator of battery charge except that we can't count on them to be consistent across all LEDs). The LEDs should have a forward voltage less than 3.6 V.

It is powered by VBUS, but the I2C inputs are on VCC. If VBUS is on the high side, the I2C inputs may not be driven high enough.

This can be resolved by changing to a dual-supply I/O expander such as PCAL6416A which would also give us another 8 I/O pins that we can put to use.

JST connectors J4 and J5 are difficult for insertion and removal. The connectors used in https://github.com/vdbxio/USB-BD are easier to use but are vertical. We should look for a replacement part.

It is difficult to install hex screws/standoffs. Moving J8 (input screw terminal) closer to J3 (USB) would make it a lot easier.

Copy silkscreen pin numbers from Azalea.

D4 is dimly lit when Jasmine is charging with load switch SW1 turned off. This happens when charging from Azalea's USB port but not when charging from Jasmine's USB port. I assume this is due to reverse leakage through D2, but that would be good to test.

It should be powered by VBAT.

It would be nice to have VBAT always present as long as there is battery charge so that the RTC keeps running even when the load swicth is turned off. Move VBAT regulator (see #15) from LOAD to BATT.

R3 (MEASURE_E pull-up) and R15 (RATE_R pull-up) connect to VCC but should connect to VBUS.

It would be nice to have a method to turn on/off X1 and to have it default to off. I'm thinking a flip-flop such as NC7SZ175 that can be powered by VBAT and has 5.5V-tolerant inputs. The flip-flop can be controlled by a bigger I/O expander (see #17). X1 draws such a tiny current that it can easily be powered by the flip-flop output.

It seems silly to take up a GPIO pin when all other similar functions are controlled through the more neighborly I/O expander. We would just need a bigger I/O expander to fix this. See #17.

I suggest increasing series resistance to tone them down.

This isn't a critical issue because U5's inputs are 5.5 V tolerant, but it would make more sense and would probably be safer to have U5 supplied by VBUS which supplies its inputs.