LSM6DS032 "doubled" acceleration sensitivity (code matches datasheet but not test)

[scandey]

Hello! I have an Adafruit LSM6DS032 board connected by Stemma QT cable to an Adafruit RP2040 Feather, all purchased about a year ago. I've dug them back out for a repeat of a workshop I ran in June last year, where I noticed that the acceleration values were all about double what they should be (Z acceleration on a flat surface was reported as about 19.5G). It didn't matter much to the workshop (just seeing any data was good enough), but I'm curious about it now.

After a peek through the code in this library:
In the LSM6DS032 class, the acceleration ranges override those from the default LSM6DS values in __init__.py with doubled max values and doubled conversion factors . If I use the original conversion factors, the output is reasonable. So the code matches the relevant datasheet, but my test suggests the conversion factors are "actually" somehow the same for both the ±16G max chips and the ±32G max chip, which just cannot be true.

I'm now starting to wonder if is possible that my actual board has the wrong chip on it. Is there any easy way to tell if that were the case, short of trying to put >16G's while getting data from the sensor and seeing what happens?

[ghost]

Found the same issue with my Adafruit LSM6DS032, appears to have a ±16G part (measures ~2G in z-axis when still and facing up) but the WHO_AM_I register matches with the ±32G part, so who knows. Also I'm not using the Adafruit library; just reading the registers with a logic analyzer.