LPS33HW Temperature absolute accuracy

[PSoko.1]

Temperature absolute accuracy. This parameter was not found in the datasheet.

Of the 15 sensors tested, the temperature ​difference was 5 degrees.

What ​temperature absolute accuracy should be?

[Eleon BORLINI]

Hi @Petr S_Ookolov​ ,

But are you referring to 5 degrees spread among the 15 sensors (for example, a sensor shows ), or to some constant offset ()?

And at which temperature are you finding the shift? 25°C or not ambient temperature? And is a constant shift versus temperature?

In any case, although for the temperature sensor the typical specs are not guaranteed, as reported in the datasheet p.9, 5 degrees is a little an high difference.

Can you please check, in order:

1. LSB to physical units conversion:

float_t lps33hw_from_lsb_to_degc(int16_t lsb)
{
  return ( (float_t)lsb / 100.0f );
}

2. Position of the temperature sensors with respect to the reference temperature: are all 15 the sensors in the same position? And in the same experimental condition of the reference sensor, if any? This in order to isolate the contribution of any heat source such as application processor or similar.

-Eleon

[PSoko.1]

Hi Eleon! Thanks for the help.

All 15 sensors had different spread of values. 0.5 to 5 degrees. All sensors showed a value below the environment. I also have boards with an LPS22 sensor, they measure the temperature correctly, the accuracy is within 1 degree. Temperature conversion program code is the same as for LPS33. All sensors are mounted on PCB and are in the same conditions - in a cardboard box to exclude the influence of air flow. The PCB does not heat up, since the battery powered device operates in a micro-consuming mode (~6μA), the activity is 100mls with a hibernation of 2 minutes. The sensor is remote from the processor.

We plan to use lps33 in the production as a temperature and pressure sensor; we need to solve the problem of the accuracy of the temperature channel.

[Eleon BORLINI]

Hi @Petr S_Ookolov​ 

please note that, in general and referring to the whole temperature range, it can be the case that -in general- the temperature measure is affected by the fact that the LPS33HW is insulated with a specific gel (the one that provides the waterproof feature), and there could be a delay in the heat exchange and transmission. Can you please explain how do you perform the temperature test (how many seconds do you wait for device thermalization (temperature set , and so on...)? in the LPS22 case the pressure sensor has the silicon exposed, so it reaches more rapidly the temperature target.

If you however refer to the only accuracy at t_amb, we cannot exclude a calibration error (pretty surprising since the , so I'm checking internally if it can be the case. Do you maybe have the lot tracking name? 

-Eleon

[PSoko.1]

Stabilization of the temperature of the sensors (ambient 24 degrees) was expected for a very long time (longer than 6 hours). This is precisely the sensor measurement error. The sensor is tested by highly skilled developers, so measurement errors are unlikely.

I will write the name of the party later.

Why is there no absolute precision parameter in the lps33 datasheet, but in the lps22 datasheet the parameter is +/- 1.5 degrees.

[Eleon BORLINI · Accepted solution]

Hi @Petr S_Ookolov​ ,

Thank you for your explanation and details, appreciated.

I'm of course it was not my intention to doubt your professionalism, I was just trying to circumscribe the problem.

Probably the fact that the thermalization is a bit longer in the 33HW sensor compared with the 22HH one, could have caused a poor calibration at production level, where the test time is crucial. For this reason, it is important for me to be able to track the specific lot, and that's why I asked you the id.

Btw, there is no incompatibility with the datasheet, as you have seen.

On the other side (the pressure sensor side), please note the accuracy of the pressure parameter for the 33HW sensor is guaranteed, even if the package and the application purposes are much more demanding than for the 22HH.

-Eleon