My SAE Baja team had a competition this past weekend, and we were using some thermocouples in order to get some temperature data on the drivetrain and engine.
They were hooked up right before the start of the race and I think in the rush were connected to the AD595 (here is the datasheet) amplifiers backwards. i.e. T- to the + input and T+ to the - input.
Is there any way to calculate a proper temperature from the output?
There are 2 problems
... or possible lack thereof (as @Ignacio had mentioned above). AD595 sees the reversed thermocouple as reversed \$\Delta T_{TC}\$. If AD595 has a negative power supply, it can deal with negative temperature. From the datasheet, it's not immediately clear for me how AD595 deals with negative \$\Delta T_{TC}\$.
By the way, negative \$\Delta T_{TC}\$ can occur at positive temperatures, when AD595 is at a higher temperature than working end of the thermocouple (e.g. 40°C for AD595 and 25°C for working end).
As you know, AD595 measures its own temperature and does cold junction compensation (CJC) internally. The output is the difference of 2 measured values, and you don't know what they are individually. If the thermocouple is connected correctly, the output is
\$V_{out}=A(\Delta T_{TC}+T_{CJC})\$
where
\$V_{out}\$ is the output voltage
\$\Delta T_{TC}\$ is the temperature difference between the working junction and the cold junction
\$A\$ is 10 mV/°C
If the thermocouple is reversed, the output is
\$V_{out}*=A(-\Delta T_{TC}+T_{CJC})\$
Correction \$V_{out, corrected} = -V_{out}*\$ would have an error of \$2AT_{CJC}\$. My gut feeling is that that wouldn't be good enough.
If by some luck you have recorded temperature from some other sensor near the AD595, you would have an estimated value \$T_{CJC}*\$. Then you can try
\$V_{out, corrected} = -(V_{out}* - 2AT_{CJC}*)\$
Plot your voltage and post it here. May be, we can spot a better answer.
Table 1 shows the relation between input voltage and output voltage. Simply look at the row for the output voltage seen, negate the given input voltage, and use the temperature at the negated input voltage instead. Note that this assumes that your next input stage can measure and handle negative voltages; if not, reverse the thermocouple's connections post haste!
