Simulations
The simulations of the current source have been done by using LTSpiceXVII. The source file of the simulation are inside this repository at simulations/CurrentSource.
The schematic is shown in the image below.
NOTE: The maximal current as well as the current measurement branch has to be optimized after the first prototype.
There is a slight dependency of the output current on the load resistance - here the resistance of the PT100 - observable. From the simulation, the inital current is about 14.7 mA at 100 Ohm (0°C), which lowers to 12.7 mA at 150 Ohm (130°C).
Assuming a fully linear relation this results in 2 mA/30°C = 0.067 mA/°C. Nevertheless, this dependency is no reason to worry about, because the current through the PT100 gets measured and thus it can be analysed as well.
In case the current dependency is a problem - or a lower dependency is needed - R7 can be increased but the maximal current has to be optimised again.
The PWM filter simulation has been performed using a pulsed voltage source, and the settings, which would be used inside the STM32.
The files can be found in simulations/PWMFilter
The cutoff frequency of 142Hz should be low enough, when using the PWM output in the O(10kHz) region.
For the measurement of the 3dB point this help has been useful.
During the evaluation of the first prototype, it has been figured out, that the resolution of the measured resistance - and via this the temperature - is okay, but leaves a lot of room for improvements. In revision 2 a switchable gain amplifier is implemented.
The amplification can be changed between 1 (simple buffer) and the factor defined by the resistors.
The simulation shows the behaviour, which is expected.
During operation one measurement will be done with unique gain and another one with a higher one. By this the resolution at small currents and voltages can be improved. Thus the resolution of the measured resistance will be increased as well.