Then, could it be compensated by calibrating periodically?
Sure, if you have an accurate voltage reference or stable Vdd. In fact, Parallax's S2 robot incorporates this capability via analog multiplexing with Vdd and Vss inputs, in addition to all the stuff it has to monitor.
Good, I'm starting to see the light. If the "natural" sigma delta fails maybe I'll try this one.
I'm only needing to calculate the temperature of a thermistor that will slowly change (max 6°/sec) so that could be a winning solution, thanks!
Alex
I am learning along with you. I had never considered that for many slow samplings, less that perfect wiring will do nicely. The reality is that in most cases I don't need to push the Propeller to the limit == there is usually the ability to do things a bit slower and there is usually code space and cogs left unused.
Just to expand on what Phil said about using Vdd and Vss as the reference voltages. For your thermistor, you can include a stable reference resistor in your circuit and tie it to an extra output pin that can be high or low, or input. Then the current at the summing junction is ratiometric between your thermistor (which is connected from the junction to ground) and the reference resistor. Drift in the Prop drops out of the equation. This depends on having resistance values large relative to the output resistance and pullup/down capability of the Prop pins.
The intrinsic drift is mostly due to the change in the threshold voltage of the Prop with temperature, change in the output resistance curves of the pmos and nmos output transistors. In addition to external temperature changes, the self heating of the Prop comes into play. Locally on the chip, the sigma-delta technique runs the nmos and pmos output transistors dead on at their switching point, where they are drawing current and locally creating heat, which then sloshes around within the chip.
For the Prop 2, word from Chip is that it will include bits that allow that calibration connection with an internal resistor. I'd hate losing the direct external connection to the summing junction though.
I'd hate losing the direct external connection to the summing junction though.
I must've missed that tidbit from Chip. I would also hate losing access to the feedback pin and the ability to select my own feedback components, e.g. diodes for log response.
Just to expand on what Phil said about using Vdd and Vss as the reference voltages. For your thermistor, you can include a stable reference resistor in your circuit and tie it to an extra output pin that can be high or low, or input. Then the current at the summing junction is ratiometric between your thermistor (which is connected from the junction to ground) and the reference resistor. Drift in the Prop drops out of the equation. This depends on having resistance values large relative to the output resistance and pullup/down capability of the Prop pins.
The intrinsic drift is mostly due to the change in the threshold voltage of the Prop with temperature, change in the output resistance curves of the pmos and nmos output transistors. In addition to external temperature changes, the self heating of the Prop comes into play. Locally on the chip, the sigma-delta technique runs the nmos and pmos output transistors dead on at their switching point, where they are drawing current and locally creating heat, which then sloshes around within the chip.
For the Prop 2, word from Chip is that it will include bits that allow that calibration connection with an internal resistor. I'd hate losing the direct external connection to the summing junction though.
I'm understanding that for getting the drift out of the equation I have to take two different measures anyway (one with the pin as input and the other as output). Just like the calibration scheme in the app note. right?
Comments
How bad can that be and what does it depend on? Do we have an order of magnitude for it?
Alex
-Phil
Alex
-Phil
I am learning along with you. I had never considered that for many slow samplings, less that perfect wiring will do nicely. The reality is that in most cases I don't need to push the Propeller to the limit == there is usually the ability to do things a bit slower and there is usually code space and cogs left unused.
The intrinsic drift is mostly due to the change in the threshold voltage of the Prop with temperature, change in the output resistance curves of the pmos and nmos output transistors. In addition to external temperature changes, the self heating of the Prop comes into play. Locally on the chip, the sigma-delta technique runs the nmos and pmos output transistors dead on at their switching point, where they are drawing current and locally creating heat, which then sloshes around within the chip.
For the Prop 2, word from Chip is that it will include bits that allow that calibration connection with an internal resistor. I'd hate losing the direct external connection to the summing junction though.
-Phil
Alex
-Phil