BMP280 first read BUG

[acassis]

Everytime bmp280 run for the first time after reset the board it is reporting incorrect value.

Near to the sea level it should return near to 1013 hPa, but:

nsh> bmp280                                                                     
Absolute pressure [hPa] = 609.820007                                            
Temperature [C] = 25.740000                                                     

Running it at second time, everything is fine:

nsh> bmp280                                                                     
Absolute pressure [hPa] = 1008.929993                                           
Temperature [C] = 20.930000                                                     
nsh>

@jerpelea did you see this issue before (git blame show you are author) ?

I'm testing on RaspberryPi Pico: https://github.com/apache/nuttx/pull/12420

[raiden00pl]

You can check if the time between changing the sensor mode and reading the data is correct. Try to add mdelay(10) here: https://github.com/apache/nuttx/blob/3cc65d5d35e779d64acc8a2a0f47fbe65b305c61/drivers/sensors/bmp280_uorb.c#L635

[fdcavalcanti]

That's odd.. I don't remember seeing that. I'll have a look in the datasheet. Maybe a quick fix would be running a read right after driver instance is created. But sure, we should understand the root cause.

[acassis]

@raiden00pl it makes no difference.

Modification:

$ git diff
diff --git a/drivers/sensors/bmp280_uorb.c b/drivers/sensors/bmp280_uorb.c
index 7acdb664d9..62e195b825 100644
--- a/drivers/sensors/bmp280_uorb.c
+++ b/drivers/sensors/bmp280_uorb.c
@@ -632,6 +632,8 @@ static int bmp280_fetch(FAR struct sensor_lowerhalf_s *lower,
       up_mdelay(6);
     }
 
+  up_mdelay(10);
+
   /* Read pressure & data */
 
   ret = bmp280_getregs(priv, BMP280_PRESS_MSB, buf, 6);

Before modification:

NuttShell (NSH) NuttX-12.5.1
 
nsh> dmesg
board_bmp280_initialize: Initializing BMP280!
bmp280_checkid: devid: 0x58
bmp280_initialize: T1 = 27551
bmp280_initialize: T2 = 25860
bmp280_initialize: T3 = 50
bmp280_initialize: P1 = 38045
bmp280_initialize: P2 = -10494
bmp280_initialize: P3 = 3024
bmp280_initialize: P4 = 9513
bmp280_initialize: P5 = -303
bmp280_initialize: P6 = -7
bmp280_initialize: P7 = 8588
bmp280_initialize: P8 = -6299
bmp280_initialize: P9 = -242
sensor_custom_register: Registering /dev/uorb/sensor_baro0
bmp280_register: BMP280 driver loaded successfully!
nsh> bmp280
Absolute pressure [hPa] = 609.820007
Temperature [C] = 25.740000
nsh> bmp280
Absolute pressure [hPa] = 1012.669983
Temperature [C] = 24.139999
nsh>

After modification:

NuttShell (NSH) NuttX-12.5.1
nsh> dmesg
board_bmp280_initialize: Initializing BMP280!
bmp280_checkid: devid: 0x58
bmp280_initialize: T1 = 27551
bmp280_initialize: T2 = 25860
bmp280_initialize: T3 = 50
bmp280_initialize: P1 = 38045
bmp280_initialize: P2 = -10494
bmp280_initialize: P3 = 3024
bmp280_initialize: P4 = 9513
bmp280_initialize: P5 = -303
bmp280_initialize: P6 = -7
bmp280_initialize: P7 = 8588
bmp280_initialize: P8 = -6299
bmp280_initialize: P9 = -242
sensor_custom_register: Registering /dev/uorb/sensor_baro0
bmp280_register: BMP280 driver loaded successfully!
nsh> bmp280
Absolute pressure [hPa] = 609.820007
Temperature [C] = 25.740000
nsh>

[raiden00pl]

I see there is no waiting for any status flag before reading the data. Make sure that measuring bit in status register is 0. Max measurement time fir standard resolution mode according to doc is 13.3 ms, so 10 ms delay may not be sufficient. I'm not sure if bmp280_activate() is called before data fetch in this case