It may be possible to support 2 (or maybe more) thermocouples without the thermocouple multiplexer shield by using the differential mode ADC, 40x gain. Only 8 bits are usable with 40x, the noisy lower bits help with oversampling though. 488 uV per count works out to 21.5 deg. F resolution with a K type thermocouple. The 2.56 volt internal reference would double the resolution and oversampling could probably quadruple it (16x oversample). ADC0 and ADC1 are the negative side, any other ADC pin may be positive. With a filtering cap (10nF) several thermocouples could share a common ground pin.
The internal ATMEGA temperature sensor needs both offset and gain calibration, a 10 deg-C rise is typical as well, 2 point calibration may be much to expect for users. For this reason another analog pin would need to be used for a CJC sensor (Analog TMP36).
It may be possible to support 2 (or maybe more) thermocouples without the thermocouple multiplexer shield by using the differential mode ADC, 40x gain. Only 8 bits are usable with 40x, the noisy lower bits help with oversampling though. 488 uV per count works out to 21.5 deg. F resolution with a K type thermocouple. The 2.56 volt internal reference would double the resolution and oversampling could probably quadruple it (16x oversample). ADC0 and ADC1 are the negative side, any other ADC pin may be positive. With a filtering cap (10nF) several thermocouples could share a common ground pin.
The internal ATMEGA temperature sensor needs both offset and gain calibration, a 10 deg-C rise is typical as well, 2 point calibration may be much to expect for users. For this reason another analog pin would need to be used for a CJC sensor (Analog TMP36).