fix hang due to powerdown of hts221 sensor

[Rocketct]

Introducing a delay before the power off of the hts in the end function the seems that allow to fix the issue due a power cycle of the module in loop.

this change seems that allow to power down the sensor without generate the issue

the test sketch is:

#include <SPI.h>
#include <MKRENV.h>
#include <ArduinoLowPower.h>

int counter = 0;

void setup() {
  pinMode(0, OUTPUT);
    pinMode(1, OUTPUT);
}

void loop() {
  digitalWrite(1, LOW);
  if (!ENV.begin()) { 
      digitalWrite(1, HIGH);
  }
  float celsius = ENV.readTemperature();
  ENV.end();

  

  if (celsius > 0) {
    digitalWrite(0, HIGH);
    delay(1000);

    celsius = 0;
  }

  digitalWrite(0, LOW);
  delay(1000);  



//LowPower.sleep(1000);
}

with @facchinm have also check that the delay could be reduce to 50, we have choose 80 to give a higher tollerance.

cc/ @sandeepmistry @facchinm please make test, i have tested on wifi1010 + MKRENV shield

i'll try to take a look to the datasheet to check the register and sensor operation to see if compliant whit this change.

This change should be also ported on https://github.com/arduino-libraries/Arduino_HTS221/tree/master/src

[Rocketct]

UPDATE: seems that also calling two times the read like:

   float celsius = ENV.readTemperature();
   celsius = ENV.readTemperature();

rise the issue, then maybe is better move the delay on the read instead of the end.

[sandeepmistry]

After some testing, I don't think this is the best solution, I found better results by added the following to the start of readTemperature() and readHumidity():

while (i2cRead(HTS221_ADDRESS, HTS221_CTRL2_REG) & 0x01);

This is based on the data sheet, section 7.4:

The ONE_SHOT bit is used to start a new conversion. In this situation a single acquisition of temperature and humidity is started when the ONE_SHOT bit is set to ‘1’. At the end of conversion the new data are available in the output registers, the STATUS_REG[0] and STATUS_REG[1] bits are set to ‘1’ and the ONE_SHOT bit comes back to ‘0’ by hardware.

What do you think?

I think we should also consider to set the CTRL_REG1 BDU bit as you mentioned in another comment.

[Rocketct]

yes as discussed i think that the problem of the sensor is that we make operation while the sensor is involved in conversion operation, this should work, let me test, about the BDU yes i think too we need because:

0: continuous update; 1: output registers not updated until MSB and LSB reading)

[Rocketct]

Ok works !! i'll add also the BDU

[sandeepmistry]

I think this change is unnecessary for the fix.

[sandeepmistry]

I think we should update the comment as well?

[sandeepmistry]

we are waiting here for hardware, to reset the bit to set to zero as stated in datasheet comment.