Atmospheric pressure readings of 600-700 hPa

[tobyaharris]

It has been observed in a number of unrelated settings that sometimes the recorded atmospheric pressure is between 600 and 700 hPa. Typical atmospheric pressure should be around 1000 hPa. The sensors in question are not in any kind of (de)pressurised chamer or otherwise extreme conditions.

The majority still give sensible readings, but there is likely a common cause to the 12 that have settled this range.

Sometimes the readings drop to this level hours after deployment. The change is within one sample cycle. Sometimes there may be blips where the reading jumps between this and normal range.

Possible hardware failure? I2C glitch? The library reading the data from the sensor is https://github.com/pimoroni/bmp280-python

There is a suspected correlation with the sensor hanging by its cable. Laying the sensor on a surface or mounting with the bolt holes may alleviate the issue.

[tobyaharris]

The lab demonstrator has weeks of data but has never exhibited this. Could it be related to mechanical disturbance? A mechanically / electrically upset connection?

[gt401]

We have done the following experiments:

Day 1, 4.30pm

• Action: solution left with sensor hanging down (raibow cable up) from a table. Initial readings 1030 hPa ish (normal amient pressure).
• Observations: by 8.30pm readings went down to 600 ish hPa until -- this was found by looking at data at 9am on Day 2. Day 2, 9.10am
• Action: brought to horizontal position (i.e. with holes facing ceiling and back screw facing table). Initial readings 600 ish. Observations: no change for 2 hours. Day 2, 11.00am
• Action: unplug and plug back the sensor from rainbow wire. Initial readings of normal ambient pressure.
• Observations: by 9.30am of Day 3 ambient pressure still normal ambient pressure. Day 3, 9.30am
• Action: sensor changed back to vertical as in Day 1.
• Observations: by 2.30pm, i.e. after about 5 hours, readings went down to 600 ish hPa. Day 3, 2.30pm
• Action: sensor changed back to horizontal position as in Day 2 but wirthout unplug/plug back rainbow wire.
• Observations: by 9am of Day 4, readings remain on 600 ish hPa.

General comments: solution not stable, needs more understanding on how sensor is reset/read/positioned

[tobyaharris]

Thanks for that data! We will look into this. It seems the issue can be suppressed by a soft reset (unplug / replug, power cycle etc) so we should be able to do this in code.