Zappy Module

Zappy Module is a board-level module that takes control input via some kind of protocol and outputs signals to zap someone like shock collars do.

Project Status

2015-04-29: ARM signal is now obeyed, triggering is only possible when ARM is high. Started working on I2C, not working yet.
2015-04-24: Assembled logic side of board and threw together some test firmware, basic functionality is all working. Worked on/sorted out pulse shaping with some extra components on the secondary side of the transformer. Next step is to implement I²C.
2015-04-22: Assembled analog side of first board, +3.3v and +10v rail are both working fine, as is the cap charging circuitry. started testing HV triggering but really need the logic on the board to test it properly. Next time will assemble logic and throw some test firmware together to sequence charging/triggering.
2015-04-21: Boards arrived, ordered parts.
2015-03-26: Sent to OSH Park for PCB manufacturing. Still need to order main transformer from Coilcraft and parts from Digikey.
2015-03-15: Completed routing board, silkscreen and checklist run done. Waiting 24h for final review and submission to OSH Park.
2015-03-08: Finished schematic and BOM, checklisted.
2015-03-07: Started putting schematic and BOM together.
2015-02-02: Started putting together overall circuit idea.
2015-02-01: SPICEd out some ideas.
2015-01-29: Captured waveforms of all output levels.
2015-01-05: Arranged to borrow a shock collar to characterize the output of it.
2013-05: Started throwing ideas around.

Main control input via I2C
ARM signal must be high before zapping is allowed
Two trigger pin inputs, can program levels for each via I2C then zap with simple "pull pin high" I/O

Input voltage from 3-9V is stepped up to 10V using a boost converter (AZ34063U) to get the intermediary voltage rail
MCU controls a FET that charges a cap from that rail via a series resistor to control charge rate
- Divider on the cap feeds into an analog pin on the MCU, once the analog pin reaches the desired level the MCU shuts the FET off again
Once cap is charged, MCU pulses a FET that dumps the charge on the cap through a 1:100 (or so) transformer and to the output
- 0.1-1kV should be achievable via timing the charge FET
May need a dump resistor if the cap is charged and needs to be discharged without zapping?
- Could use the series charging resistor w/ a different FET to dump the cap through it?
- Don't think this will be required, the charge cycle will only take ~1mS so it will be done right before the trigger, so the cap will never remain charged for long

I2C, default address 0x5F
All commands follow the basic format of:
- Three magic bytes to ensure no zapping from random probing or bus noise (0x5A, 0x41, 0x50 / Z A P)
- One byte command
- Variable number of one byte arguments
0x00 - NOOP
0x01 - ZAP
- Arg 1 - Level
  - 0-255, volts/2 to output
- Arg 2 - Pulse length
  - 0-255, milliseconds to pulse for
0x02 - SET
- Arg 1 - Channel to set
  - 0=A, 1=B
- Arg 2 - Level
  - 0-255, volts/2 to output
- Arg 3 - Pulse length
  - 0-255, milliseconds to pulse for
0x03 - ADDR
- Arg 1 - New address to take on

All collar output waveforms are through a 1/3 divider, so multiply by 3 to get actual amplitude
Lowest level (1L) seems to be ~90Vpp, highest seems to be ~450Vpp
- I thought the highest was ~2x that? Need to measure the higher ones again to double check

D1-D3 on BOM only has qty 2
0402 pads are too big
No pins on BOM
Need pulldowns on the FET gates
- Otherwise when programming both trigger/charge can be on at once which pulls lots of current and can blow up parts
Need 33pF (or so) and a diode across the transformer secondary to shape pulses
+10V rail is ~12V since it's mostly unloaded.
Input pins only register as HIGH at 5v, resistors need to be changed to ~800k to make it work at 3.3
No pullups on the I2C pins (errata clarifies that the docs are wrong when it says they have pullups)