Now, in order to calibrate this system, we put in front of the filter an Analog Mux, so that we can take two voltage measurements. One, we can control this actually should select line for this multiplexer should go off to like a GPI on the CPU. I didn't realize that and I would just look at this picture. The idea is, we want to take whatever this Vdd is in and make a sample, and then we want to take the voltage based on whatever the Thermistor resistance is at its particular temperature, we want to sample both of those. So, calculating Rtherm, we measure Vdd. So, I call that MVdd, measure that one. Then we switch the Mux and we measure Vtherm. So, if we got measured Vtherm, and we compute the current IC through that voltage divider is the difference of those Vdd minus Vtherm divided by Rfixed, tells us what that current is going down to the voltage divider. We can rearrange this and solve for Rtherm, and Rtherm it turns out to be the measured Vdd divided by IC minus the Rfixed value. Then Rtherm is looked up in the manufacturer's data sheet, it's a table, and that lookup table used then Rtherm as an index into that table and the tables in firmware and you just look it up and boom out pops a temperature in Fahrenheit or Celsius. So, what about the tolerances on specifically on the voltage reference? So, we measured Vdd, has a tolerance associated with it and we measured Vtherm, has a tolerance associated with it and it turns out that I didn't work through all of the math like I probably should have, but you can see that we're taking measured Vdd and subtracting measured Vtherm and we've got this plus T term on both sides. So, we're adding it here and we're subtracting it there. So, it cancels out, the tolerance cancels out. So, we have just a measured Vdd minus measured Vtherm.