Fall 2025 to Jan 2026
An investigation into SARIT parasitic battery drain, leading to a proposed charger-isolation contactor and a current-monitoring "digital fuse box" architecture.
![[sarit battery compartment]](/assets/[sarit-battery-compartment].jpg)
SARIT battery and electrical compartment; the drain problem came from parked vehicles losing charge without being driven.
Quick Read
- Investigated why SARIT vehicles lost charge while parked.
- Compared immediate post-charge range against range after 72 hours parked.
- Found a drop from about 15 km to about 6 km after sitting with the charger connected but unplugged.
- Estimated continuous parasitic draw around 100-200 mA.
- Ruled out BAC2000 inrush current as the drain source.
- Proposed charger isolation with an AEV250-G contactor.
- Proposed an I2C current-monitoring system across the 48V, 12V, and 5V rails.
Problem
The SARIT fleet had a range problem that was not caused by driving.
A vehicle charged for one hour could drive about 15 km immediately. But after sitting for 72 hours with the charger physically connected to the battery and unplugged from the wall, the same charge cycle produced only about 6 km of range.
| Condition | Charge time | Idle time | Resulting range |
|---|
| Baseline | 1 hour | 0 hours | ~15 km |
| Drain test | 1 hour | 72 hours | ~6 km |
That suggested a parked parasitic drain.
Investigation
A clamp meter was the first tool tried, but it was not useful at the current levels involved. The readings fluctuated between about -0.1A and 0.2A, which was too noisy for milliamp-level leakage.
Large 10-30A spikes appeared when reconnecting the battery, but those were a red herring. They matched the ASI BAC2000 motor controller input capacitance charging up, not a continuous drain.
The useful test was disconnecting the onboard charger from the battery terminals.
After that, the measured drain dropped to zero.
The working hypothesis became charger back-feed: the battery was slowly discharging through the charger's internal circuitry when the charger was connected to the battery but unplugged from the wall.
Proposed Fix
The immediate fix was to isolate the charger from the battery when the vehicle was parked.
The proposed part was an AEV250-G contactor:
- SPST normally open.
- 48-72VDC coil.
- 500A rating for capacitive inrush.
- Coil economizer draw around 0.03A at 48V.
Digital Fuse Box
The longer-term proposal was a current and voltage monitoring system across each rail.
| Component | Part | Purpose |
|---|
| I2C power monitor x4 | Adafruit 5832 / INA228 | Monitor battery, 12V, 5V, and 48V rails |
| Traction current sensor | Allegro ACS758-200B | Isolated current sensing on the motor rail |
| DC shunt | FL-2-100A / 75mV | Main battery negative measurement |
| STEMMA QT cables x4 | Adafruit 4401 | I2C daisy-chain wiring |
Estimated BOM cost was $124.02 CAD.
- 0x40 main battery.
- 0x41 12V rail.
- 0x44 5V rail.
- 0x45 48V rail.
Handoff
In January 2026, the project was handed off for KiCad formalization.
The intended schematic structure was:
- Traction 48V.
- Accessories 12V.
- Logic 5V.
- Current sensors and shunt footprints.
What I Would Improve Now
I would repeat the drain test with a proper inline current measurement setup earlier.
The clamp meter was useful for proving what did not work, but the actual problem needed lower-current instrumentation from the start.