Skip to content

Troubleshooting

Symptoms, causes and fixes, cheapest check first.

Datasheet sections this page relies on

From Bosch Sensortec BST-BMP581-DS004-13 1.13:

  • Normal mode and the ODR/OSR check: 4.3.4 NORMAL mode; Table 7 maximum nominal ODR setting per OSR settings, p. 17
  • Post-power-up checks: 4.3.9 Post-power-up procedure, p. 18
  • I2C protocol and addresses: 5.6 I2C protocol; 5.6.1 I2C write operation; 5.6.2 I2C read operation, p. 38
  • Identification and status: 7.1 Register (0x01) ASIC identification ID; 7.22 Register (0x28) Status register, p. 51

All of them, with the driver API for each: Datasheet map.

EnsureInitialized returns NoDevice or Bus

The port reported a failure: its EnsureInitialized() returned false, or a transfer did.

  1. Check that your port's EnsureInitialized() returns true on its own (log it).
  2. I2C: check the address against SDO (0x46 with SDO at VSS, 0x47 at VDDIO), the pull-ups, VDD and VDDIO, and that CSB is tied to VDDIO. One NACK during the start-up is expected: the soft reset write (section 7.36).
  3. SPI: check the CSB pin and that the port drives it high between frames.
  4. With a logic analyser, check that the first frame (SPI: a CHIP_ID read; I2C: CMD = 0xB6) appears on the bus.

EnsureInitialized returns DeviceIdMismatch

CHIP_ID did not read 0x50, or INT_STATUS did not read 0x10 after the soft reset and 0x00 after that.

Symptom (ReadRegister() or a logic analyser) Likely cause Fix
CHIP_ID 0x00 SDA or MISO stuck low; on SPI no MISO wire, or a chip still in I2C mode Check continuity; check that CSB toggles
CHIP_ID 0xFF The data line floating high, or the chip unpowered Check VDD, VDDIO and SDO
CHIP_ID 0x51 or another value Another part (a BMP585 reads 0x51) Check the part marking: "C" is a BMP581 (table 29)
CHIP_ID 0x50, INT_STATUS not 0x10 The soft reset did not reach the chip SPI: check the first frames; I2C: check that the port sends a write even when a NACK follows

EnsureInitialized returns Timeout or DeviceFault

Timeout: STATUS.status_nvm_rdy never read 1 within Config::startup_timeout_ms after the soft reset; check the supplies and their ramp (section 4.2). DeviceFault: STATUS reports an NVM error, so the trimming data did not load; a power cycle may clear it, a repeated one means a damaged part.

EnsureInitialized or ApplyConfig returns WriteMismatch

A configuration register read back differently from what was written. A single one points at noise on the bus: retry. Every time, on the same register, points at a port that corrupts writes, or at a chip outside standby (the driver always writes in standby, section 4.3.8).

SetPowerMode or EnsureInitialized returns ConfigRejected

Normal mode with a data rate the chip does not accept for the oversampling (table 7): OSR_EFF.odr_is_valid still read 0 at the deadline (t_startup_deep, 4 ms, polled every 0.5 ms), and the chip would measure with its own defaults. Validate() refuses the combinations table 7 rules out, so this means the chip disagrees with the table: lower the data rate or the oversampling, and read GetLastOsrEff() for what the chip chose.

Reads return InvalidReply

The reply cannot come from the chip: the reserved registers 0x23 to 0x26 did not read 0x00, a reserved bit was set, or the frame was all 0x00. Once now and then: noise; the next call reads again. Every time: a stuck or floating line (pull-ups, SDO, the MISO wire). On SPI the first frame after a power-on reset only selects SPI again and reads as garbage; the driver then checks for the reset itself and returns DeviceReset. A pull-up on MISO makes that frame read 0xFF (Hardware).

Reads return DeviceReset

The chip reset since the driver started it, usually a dip of VDD or VDDIO, confirmed by OSR_CONFIG reading 0x00 twice. The driver is uninitialised: call EnsureInitialized(). If it repeats, check the supplies, their decoupling (100 nF at each pin) and the inrush current (section 6.2.5). A line stuck low reads the same way; EnsureInitialized() then fails with DeviceIdMismatch.

Reads return NoData

The data registers hold their reset value 0x7F7F7F: no measurement since the last reset. Start one: MeasureOnce(), or normal or continuous mode. The driver also checks for a reset first.

Reads return OutOfRange

The value lies outside 30 kPa to 125 kPa or -40 °C to 85 °C (tables 1 and 2). In a pressure chamber or a freezer that is the truth; otherwise compare with a reference before trusting the chip.

MeasureOnce returns Busy or Timeout

Busy: the configured mode is normal or continuous, or the previous forced measurement still runs; call SetPowerMode(PowerMode::Standby) first. Timeout: the chip did not return to standby within ForcedMeasurementMaxUs() (106 ms at x128 and x128). Check that your port's DelayUs() really waits: a delay that returns at once makes every wait far too short.

WaitDataReady returns Unsupported or Timeout

Unsupported: Config::source_data_ready is off (the chip reports drdy only for an enabled source), or the mode is standby. Timeout: no measurement within data_ready_timeout_ms; it must cover one data rate period (8 s at 0.125 Hz).

The pressure jumps when the air moves

That is the air: doors, wind and fans move the pressure by more than the sensor's noise (section 4.4.2). Enable the IIR filter (pressure_iir, iir_to_data_registers) and choose the coefficient from table 10.