Autostart module for EVBox G2/G3 chargers
Written 2026 by Maarten Tromp <maarten@geekabit.nl>
Website: https://www.geekabit.nl/projects/evbox-charger-autostart-module/
License: Public Domain (CC0)
For more info see website


This document details the hardware design choices for the autostart module.


System overview
---------------
The system consists of a single controller module (CP), and 1..20 EV charger modules (CB). All nodes are connected over a RS-485 bus.
The CP is typically placed inside the first charger enclosure.
The autostart module is designed as electrical / mechanical drop-in relacement for the CP, offering partial functionality, without depending on an internet connection, and a backend.
The autostart module is intended as a replacement for CP and is not designed to coexist with a CP.
The main reason for choosing the autostart module over CP is to be independent from internet / backend.

Functional similarities between CP and autostart module:
	Provide RS-485 master functionality.
	Respond to CB messages.
	Initiate charging via protocol.
	Works with both G2 and G3 CBs.

Functional differences between CP and autostart module:
	The autostart module provides a subset of the CP functionality.
	CP can handle up to 20 CBs, autostart module handles 1 CB.
	CP depends on backend, autostart module functions stand-alone.
	CP offers authentication, autostart module does not.



Functional requirements
-----------------------
Easy to install, easy to operate.
When you connect an EV to the CB, charging should start automatically.
No need for manual intervention / user interaction.



Environmental requirements
--------------------------
The autostart module is typically located inside a type-3 EV charger enclosure.
Temperature: Charger enclosures are typically black / antracite and installed on parking lots. Temperatures may rise to ~50 °C, but also sink below freezing. Charger operating temperature range is -25°C..+60°C. The autostart module must meet (and preferably surpass) that.
Humidity: The enclosures are IP54. Chargers are reated Max. 95% relative humididy. The autostart module should meet/surpass that. I have seen little pools of water on the bottom of enclosures.
Note: being installed literally on top of power electronics is an electrically noisy environment.



Power supply requirements
-------------------------
Power comes from a 12 V DIN rail module, and is shared with the charger electronics, including locking servo. There are no contactors, relays, or solenoids on this power rails.
DIN rail PSU: Typically Phoenix Contact, Uno Power, 12 V, 30 W, or Omron, S8VK-G03012, 12 V, 2.5 A. Both are protected against overload (short-circuit-proof).
Phoenix Contact power supply is stable within 3%, according to datasheet. I cannot find stability figures for the Omron PSU, but assume it to be similar to the Phoenx Power one.
The autostart module power supply must provide stable output on suppy voltage 12 V ± 20% = 9.6..14.4 V, and transients of ±30 V for ≤ 100 ms.
Survivability (no damage, operation may stop): up to +36 V
Output voltage: 3.3 V ±5% under all operating conditions
Ripple/noise: < 100 mVpp (non-critical, informational)
TODO: read Vreg datasheet
TODO: reverse CP power supply
TODO: How should negative voltage / wrong polarity be handled?

Reverse polarity protection: Not explicitly protected. The autostart module is placed in existing systems.
Rationale:
	- Controlled installation environment
	- Upstream PSU is fixed and keyed
	- TVS provides optimal surge suppression

Failure mode:
	Reverse polarity may destroy TVS and create short circuit.



RS-485 interface requirements
-----------------------------
BER < 10^-6 under test conditions
Multi-master, single CP, 1..20 CBs, length is up to 1200m, 34.4kbps. 
CP always terminates, but does not provide bias. The bus is terminated on other end only only when >= 6 CBs.
Autostart module terminates and does provides bias, only a single CB, bus is only a few cm long. Termination and bias are always enabled. This should work for all deployment scenarios.
Standard: Compliant with TIA-485-A
Baud rate: 34.4 kbps
Topology: Single master, single slave (autostart configuration)
Termination: 120 Ω at module
Biasing: Provided by module (4.7kΩ pull-up/down)
Cable length: < 0.1 m (internal wiring)



UPDI requirements
-----------------
The UPDI interface complies with Microchip UPDI specifications.
physical: 1 testpin/testpad, use GND from charger interface.
electrical: starts with +12 V/150 µs pulse, the rest is 3.3 V serial communication.
pull-up resistor on UPDI pin to Vcc to have a defined level at all times, and stop noise from potentially triggering programming mode.
High-voltage entry: 12 V pulse, ~150 µs
TODO: verify from Microchip documentation
Usage: Programming only, in controlled environment
No additional robustness requirements.



ESD considerations
------------------
The RS-485 transceiver provides ±15 kV ESD protection (HBM).

Additional protection:
	TVS diode on RS-485 lines

Target:
	Robustness against handling and installation ESD events

No formal IEC 61000-4-2 compliance is claimed.



Mechanical requirements
-----------------------
CP dimensions: 88.0 x 70.0 x 1.4 mm
Autostart module v1.0 dimensions: 22.4 x 22.4 x 3.5 mm
Autostart module dimensions <= CP dimensions
Target dimensions: < 110% of v1.0 hardware
CP weight: 53 g
Autostart module v1.0 weight: 5.6 g
Connector (new) weight: 3.00 g
Target weight: < 110% of v1.0 hardware
The only mechanical connection is provided by the connector to the CB module.
There are 2 additional stands, but these lie far outside autostart module dimensions.
Withdraw strength per pos. approx. 6 N = 24 N
Fixed installation


Interfaces
----------
Charger interface:
	The CP module plugs directly onto a connector on the CB module.
	physical: Phoenix Contact MSTBVA / ICV 1 row, 4 contacts, 2.5 mm2, 5.08 mm spacing, 320 V, 12 A
	Electrical: +12 V/GND, and RS-485 A/B.
	protocol, simple retry
	There is a single GND, connected to PE, shared between power and RS-485.

UPDI programming interface



Failure modes
-------------
If power fails open, then the module will stop working, and the user will notice.
If RS-485 fails open, then same.
If RS-485 bus fails short, then current is limited by the transceiver, transmit time is limited by protocol, this is within safe limits of Vreg current/dissipation
If power fails short before Vreg, then upstream PSU is protected against overload.
If power fails short after Vreg, then Vreg has current-limit and temperature-limit.
If RS-485 gets shorted to power, then all nodes on the bus should be protected against this.
Can the module break the CB? Are there failure modes that result in permanent damage to CB and/or autostart module?
Are thee failure modes that can result in intermittend working/notworking?
	Vreg overheating could oscillate
	RS-485 bus termination/bias could be missing, but it would still mostly work.

Failure mode: RS-485 short to supply
Effect:
	Bus disruption
	No permanent damage expected
Justification:
	Transceiver protection + current limiting

No, module cannot damage CB under defined fault conditions

	Reverse polarity → TVS conducts → destructive short



Safety considerations
---------------------
TODO
12 V system
No direct connection to mains
Dependence on charger safety



Priorities
----------
My priorities are, in order of descending importance:
	technical correctness
	no exotic or expensive parts
	compact size, mostly SMD
	limited number of different parts on BOM



Production / assembly
---------------------
Leaded Pb63/Sn37 solder paste and solder, not RoHS complient
Ambient air environment, no N2
SMD parts:
	footprint limited to 0603
	hand stencilled
	hand placed
	reflow soldered
Through-hole parts:
	hand soldered
No conformal coating



Design decisions
----------------
Power input filtering:
A ferrite bead is recommended between connector and bulk capacitance to reduce high-frequency switching noise from charger power electronics.

Adding bias:
Justification:
	Ensures defined idle state for standalone operation
	Improves noise immunity in electrically noisy enclosure
	This is a best practice.
Risks:
	Conflicts with original EVBox bus design.
Mitigation:
	Could mark bias resistors DMP, or remove from PCB.

Bend connector pins:
	bend flat after assembly, before solderng, for additional stability, and to offload solder joints
	Is this needed?



Component choise
----------------
Capacitors:
	CP uses 105 °C electrolytic capacitors
	V rating >= TBD % headroom.
Capacitor temperature rating:
	≥105°C, based on:
	- Original CP design
	- Elevated ambient temperatures inside enclosure
	- Expected self-heating and lifetime considerations
	Use 125°C where cost allows → increases lifetime margin


Connector:
	requirements:
		compatible connector
	part:
		off-brand part, validated for electrical and mechanical compatibility
	footprint:
		same as original
	notes:
		good / tight fitting footprint
		TBD: verify connector electrical and mechanical suitability

RS-485 input filtering:
note:
	Keep RS-485 pair tightly coupled.

TVS:
	requirements:
		typical RS-485 TVS
	part:
		CDSOT23-SM712 / SM712
	footprint:
		SOT-23
	notes:
		close to connector
		stitch to ground plane

HF filter caps:
	requirements:
		not needed for such a short bus
	part:
		none
	footprint:
		none
	notes:

common mode choke:
	requirements:
		not needed for such a short bus
	part:
		none
	footprint:
		none
	notes:

termination and bias
	requirements:
		typical bias and termination for short bus
	part:
		2x 4k7 pull-up/down (~0.35 mA)
		120 Ohm termination
	footprint:
		3x 0603
	notes:

series resistors:
	requirements:
	part:
		2x 10 Ohm thick film
	footprint:
		2x 0603
	notes:
		close to transceiver

RS-485 transceiver:
	requirements:
	part:
		ADM3485EARZ
	footprint:
		SOIC-8
	notes:
		The "E" version is an upgrade of the original and includes ±15 kV ESD-Protection
		standard temperature operating range -40..+85 C is good enough, industrial range not needed
		local decoupling described elsewhere
		non-"E" transceiver is used in CP


Power filtering:
fuse
	requirements:
		polyfuse
		100mA hold
		200mA trip
		Rely on upstream PSU over-current protecton and Vreg over-current and over-temperature protection
	part:
		none
	footprint:
		none
	notes:

TVS:
	requirements:
		typical clamping for 12 V supply
		min clamping > supply
		max clamping < Vreg Vin(max)
		preferably smaller than SMAJ
	part:
		SMFJ15A
		max clamping = 24.4 V
	footprint:
		SOD-123FL
	notes:
		keep close to connector
		stitch to ground plane
		typical failure mode is short from over-voltage or over-current. This happens during reverse polarity

polarity diode:
	requirements:
		This diode is not needed, TVS already provides reverse polarity protection
	part:
		none
	footprint:
		none
	notes:

ferrite bead / series R:
	requirements:
		Form L/C or R/C filter together with bulk cap.
	part:
		10 Ohm thick film
	footprint:
		0603
	notes:
		same part as in RS-485 protection


Power supply:

Cin:
	requirements:
		3.3 µF recommended for Vreg -> 10 µF
		type: ceramic
		V rating > TVS max clamping voltage (24.4 V) -> 35 V
		temp rating: 105 °C
		X8R\L or X7R
	part:
		10 µF, ceramic, >= 50 V
	footprint:
		1210
	notes:
		close to regulator


regulator:
	requirements:
		linear, for simplicity and tiny current.
		large voltage drop (~8 V), but low average current (4 mA), therefore low power
		largest possible current flows when RS-485 A and B are shortened. Transceiver Imax = 250 mA, but only during transmitting "1". However very little will transmitted without incoming data, therefore average current will still be low. TODO: quantify
		Vin(max) > TVS max clamping voltage
		Imax > 10 mA
	part:
		MCP1792-3.3/CB
		Vout = 3.3 V +/-2%
		Imax = 100 mA
		Vin(max) = 55 V
		protection: thermal, short cirsuit, under voltage (2.7 V)
		safe for transients < 70 V
	footprint:
		sot-23
	notes:
		place on ground plane
		requires ceramic output cap >= 2.2 µF, recommended 3.3 µF
		recommended ceramic input cap 2.2..10 µF
		Ptypical = (Vin − Vout) ⋅ I = (12 - 3.3) ⋅ 4e-3 = 34.8 mW

Cout:
	requirements:
		2.2 µF recommended for Vreg -> 10 µF
		type: ceramic
		temp rating: 105 °C
		V >= Vout (3.3 V)
		X8R\L or X7R
	part:
		10 µF, ceramic, >= 10 V
	footprint:
		0805
	notes:
		close to regulator

local decoupling:
	requirements:
		2x (transceiver + AVR)
		ceramic
		V rating > Vcc
	part:
		100 nF
		6.3 V
	footprint:
		0603
	notes:
		close to component
		could reuse Vreg input decoupling part


MCU:
	requirements:
		standard temperature operating range: -40..+85 C, good enough, industrial range not needed
	part:
		ATtiny402SSN
	footprint:
		SOIC-8
	notes:
		local decoupling described elsewhere

UPDI HV programming:
	requirements:
		Problem: AVR PA0 is used for both HV UPDE and USART XDIR, RS-485 transceiver DE/RE input needs to be protected from the 12 V start pulse.
		Transceiver absolute maximum ratings on digital input: +6 V
		Solution: clamping circuit to Vcc on transceiver input. This is the standard way to handle overvoltage on logic pins.
		Typical value is 10 kOhm.
		With 10 kOhm current is limited to (Vpulse - Vcc - Vf) / R = (12 - 3.3 - 0.3) / 10 k = .84 mA, this is safe.
		Higher values are more susceptible to noise
		Can the power rail handle this? 0.8 mA for ~150 µs pulse, negligible.
		When I addicentally used a HV UPDI programmer without clamping circuit on the transceiver, the transceiver survived.
	part:
		10 kOhm resistor
		BAT54 Schottky diode BAT54, fast, small, cheap
	footprint:
		0603
		SOT-23
	notes:

	requirements:
		Problem: Noise after reset could accidentally trigger UPDI start
		Solution: pull-up resistor on UPDI pin to Vcc, so input has a defined level at all times.
		Strong enough to suppress noise, weak enough for programmer to override
		Standard value
	part:
		10 kOhm resistor
	footprint:
		0603
	notes:


current limiting between AVR PA6 and transceiver DI/RO:
	requirements:
		To limit current in case both transceiver RO and PA6 are accidentally active at the same time.
		This situation should never happen during normal operation or programming.
		AVR is current limited to 40 mA
		Transceiver is current limited to 60 mA
		This part is not needed
	part:
		none
	footprint:
		none
	notes:


PCB
---
dimensions ≈ 22 x 22 mm
PCB plug directly onto the charger module. Both mechanical and electrical connection come from single connector
Bottom side of the board is a ground plane, top side has a ground pour.
All traces are as short and wide as possible, especially around TVS and decoupling.
RS-485 controlled impedance is not required at cm scale, but symmetry matters
grounding strategy: solid ground (not split), shared between power and data. TVSs should be stitched to both ground planes.
TODO: mechanical fastening, mechanical stress


Testpads
--------
Vreg output
UPDI



Validation plan
Design validation tests
-----------------------
Hard short A–B
	Measure:
		Supply current
		Regulator temperature, how?

Supply noise
	Bench supply + long leads
	Relay switching load
	Motor / inductive load nearby
	Disconnect supply quickly (spark)
	Reconnect → creates transients

RS-485 noise
	Long cable
	not twisted, not shielded
	Leave un-terminated
	Inject switching noise nearby

TODO:
	look for noise sources, like noisy switching psu
	quantify. No clear pass/fail cireteria


	ESD testing

	Power cycling endurance

	Brown-out behavior

	Thermal soak test (50°C ambient)

	Fault injection:

		A/B to GND

		A/B to Vcc

	Inject 50Hz square wave from function generator into supply. <- does not add anything

	Put an inductive load (relay, coil, motor) on same supply Switch it with: MOSFET or relay

Capacitive injection (controlled spikes) Inject noise directly: Use function generator → series capacitor (e.g. 10–100 nF) Inject into 12 V rail This creates: high-frequency disturbances without large DC shifts
RS-485 disturbance testing (very relevant) Method A: Common-mode noise injection RS-485 is robust mainly because of common-mode rejection. Test it explicitly: Inject noise equally into A and B e.g. via capacitors from signal generator This tests: ground shifts coupling from power electronics
Method B: Differential noise injection Inject signal directly between A and B Simulates: crosstalk bad cabling
Method C: Ground offset Very important in real installations: Introduce small ground voltage difference (e.g. 1–5 V) Between: your module RS-485 source Tests: common-mode range limits of ADM3485


How about doind tests with the baud rate slightly off?
timing stress


Brown-out margin Minimum supply voltage before failure Recovery behavior - quantify


Validate input voltages, including negative.




Noise immunity validation
------------------------
The module is tested under the following disturbances:

- Supply transients generated by switching inductive loads
- Fast supply interruptions with long wiring
- Common-mode noise injection on RS-485 lines
- Operation over extended cable lengths without termination

Pass criteria:
- No unintended resets
- No protocol deadlock
- Successful recovery from communication errors

Look into arbitrary waveform generator + power amplifier
https://www.dmcinfo.com/blog/24248/low-cost-function-generator-amplifier-diy/



Test points
-----------
Which test points do I need?
For design validation I can connect wires anywhere. Test points are only needed for end-of-line testing.
	Vreg out
	UPDI programming



Note on v1.0
------------
25 units of v1.0 hardware, without any EMI filtering, works fine in the field. No hardware problems reported.


ChatGPT
-------
I want honest, direct answers. Please don't tell me that every idea I have is good, tell me what can be improved.
Understood. Direct answers, with corrections where needed.
I want direct feedback, tell me what can be improved.


search for best practices for RS-485 and power input filtering
