GPS causes +8mA power consumption even when disabled on Heltec T114

[mileshkin]

Environment

• Board: Heltec T114 (nRF52840)
• Build flag: -D ENV_INCLUDE_GPS=1
• GPS Module: Connected to Serial1 (PIN_GPS_TX=37, PIN_GPS_RX=39)
• Control pins: PIN_GPS_EN=21, PIN_GPS_STANDBY=34, PIN_GPS_RESET=38

Problem Description

When compiling with -D ENV_INCLUDE_GPS=1, the board consumes ~8mA additional current even when GPS is disabled through the UI settings.

Current Measurements:

• With -D ENV_INCLUDE_GPS=0: 14-16 mA (normal) ✅
• With -D ENV_INCLUDE_GPS=1 and GPS disabled: 22-24 mA ❌
• With -D ENV_INCLUDE_GPS=1 and GPS enabled: 50-56 mA

Expected Behavior:

GPS disabled → 14-16 mA (same as ENV_INCLUDE_GPS=0)

Actual Behavior:

GPS disabled → 22-24 mA (+8mA leak)

Investigation Done

What we checked:

1. GPIO pins (PIN_GPS_EN, PIN_GPS_STANDBY, PIN_GPS_RESET)

   • ✅ All properly initialized to OFF/SLEEP state in UITask::begin()
   • ✅ setGPSHardwareState() correctly manages all pins
   • ✅ Verified with oscilloscope - pins are in correct state

2. Serial1 (UART)

   • ❌ Adding Serial1.end() in stop_gps() causes issues:
     • Either no effect on power consumption, OR
     • Board hangs/doesn't boot, OR
     • GPS can't be toggled on/off again
   • Serial1 is initialized in EnvironmentSensorManager::initBasicGPS() but never stopped

3. I2C Wire1

   • ✅ Tested by commenting out -D ENV_PIN_SDA and -D ENV_PIN_SCL
   • No change in power consumption

4. MicroNMEALocationProvider constructor

   • Creates global object before setup()
   • Initializes PIN_GPS_EN but NOT PIN_GPS_STANDBY (we fixed this, but no effect)

Code Analysis:

The issue appears when this line executes in target.cpp:

#if ENV_INCLUDE_GPS
#include <helpers/sensors/MicroNMEALocationProvider.h>
MicroNMEALocationProvider nmea = MicroNMEALocationProvider(Serial1);
EnvironmentSensorManager sensors = EnvironmentSensorManager(nmea);

Comparison

Important: The same code works correctly on Heltec V4 board with no extra power consumption when GPS is disabled.

Questions

1. What peripheral/resource is consuming the extra 8mA when ENV_INCLUDE_GPS=1?
2. Why does Serial1.end() cause instability instead of saving power?
3. Is there something in the global object initialization that keeps a resource active?
4. How is this handled differently on Heltec V4 vs T114?

Attempted Fixes

Fix 1: Add PIN_GPS_STANDBY to MicroNMEALocationProvider

// Added _pin_standby member and initialization in constructor
if (_pin_standby != -1) {
    pinMode(_pin_standby, OUTPUT);
    digitalWrite(_pin_standby, PIN_GPS_STANDBY_ACTIVE);  // Put GPS to sleep
}

Result: No change in power consumption

Fix 2: Stop Serial1 in stop_gps()

void EnvironmentSensorManager::stop_gps() {
    _location->stop();
    Serial1.end();  // Stop UART
}

Result: Board becomes unstable or GPS won't restart

Fix 3: Initialize all GPS pins to OFF in UITask::begin()

digitalWrite(PIN_GPS_EN, !PIN_GPS_EN_ACTIVE);      // Disable power
digitalWrite(PIN_GPS_STANDBY, PIN_GPS_STANDBY_ACTIVE);  // Sleep mode

Result: No change in power consumption

Request

Could someone help identify:

1. What resource/peripheral stays active when ENV_INCLUDE_GPS=1?
2. The correct way to fully power down GPS-related peripherals on nRF52840?
3. Why the same code works on V4 but not on T114?

Additional Context

This is critical for battery-powered applications where +50% idle power consumption is unacceptable.

Any guidance would be greatly appreciated!

[weebl2000]

I've created a PR that might fix this. I don't have T114 so I cannot test. Let me know if it works.

[mileshkin]

I've created a PR that might fix this. I don't have T114 so I cannot test. Let me know if it works.

Woohoo! From 4th try It worked! Now everything is as it should be:

• screen on, GPS on: ~53mA
• screen off, GPS on: ~44mA
• screen on, GPS off: ~28mA
• screen off, GPS off: ~16mA - GPS sleeps as a baby now!

Just what we needed! Brilliant work mate!

[mileshkin]

I checked it on other boards - no unwanted consequences were found!

[IoTThinks]

Love it.
We should consume less power for long days.

Some boards do not have GPS_EN to cut off the power to GPS. Eg RAK 4631 if I'm not wrong.