TPOBJ computation needs a cast to uint32_t to reliably produce correct result about calipile HOT 6 OPEN

Odd but this seems to work fine on 32-bit ARM devices, but it certainly can't hurt to cast as uint32_t.

from calipile.

Weird, yes. Does work correctly on 32-bit Particle photon without cast, but gives lots of erratic (some good, some super high) results on 8-bit UNO using Arduino IDE 1.8.3 without cast

from calipile.

from calipile.

Well, now that I look more closely, Tambient and TPAMB also suffer this, though casting the second argument in TPAMB doesn't seem to help. Also I got confused because the typo of this line:
Serial.print("TPAMP = "); Serial.println(TPAMB);

from calipile.

I've fiddled around some more and when I now change the TPOBJ line as above and then also change:
// TPAMB = ( (uint16_t)(rawData[0] & 0x7F) << 8) | rawData[1] ;
to
TPAMB = (uint16_t)( (uint8_t)(rawData[0] & 0x7F) << 8) | (uint8_t)rawData[1] ;
I seem to now get a good level of stability/reliability/accuracy of measurement on my UNO. Don't know enough to know why this seems to help or whether I'm misguided. What do you think? Bug in the 1.8.3 IDE, bug in my head??? Differences in the hardware?

from calipile.

from calipile.