I replaced the PIC in the brain that was giving me troubles. It looks like the internal DAC isn't getting hung up anymore. The other thing that I was hoping for as a result of this replacement was better accuracy. Due to the complexity involved in some of the other (Dials and switches) circuitry I've decided and had feedback to use buffering opamps to provide a 'cleaner' signal to the brain's analog inputs. But I didn't think I'd need to for some of the other more simple input channels.
Here's an example of me thinking that I may need buffers on much less-complex circuits:
This is just a simple photosensor circuit. Unfortunately when there is no light, the software sees the voltage as 5.49V. With 2 independant meters, at the pin at the brain and on my interface board, the actual voltage is 4.96V. I did notice that for these simple sensors I connected the ground to the rest of the grounds on the card. This was unnecessary, and I removed that connection to eliminate one possibility now the circuit above is completely isolated and I'm still having the problem. I also removed the current limiting resistor (10K) and replaced it with a jumper. No difference (although I was hopeful, and crossed my fingers tightly).
Another simple circuit I have on the interface board is this:
I used the Humirel HTG3515CH sensor.
I also removed the common ground connection to the remainder of the interface board for similar reasons as above, it did not make a difference.
Voltages are again about 0.2 - 0.3V off (at ~2.00V), which is significant after the signal processing loops have their way with it. Which for humidity is about 20% off and for temp about 10F off.
So do I need a buffer here? Will it help? I have a bad feeling....
I also now have a potentiometer set up replacing a sensor currently in the car. This is the intake sensor circuit that I have shown in other posts in here (controls the AC compressor clutch request, forcing it off if the evaporator freezes). The intake sensor is set up in a voltage divider (not your everyday basic version) that drives a dual comparator. Voltage at the sensor can be as high as ~8.5V during worst case (open circuit), and the circuit is sensative to additional loading, so I tapped that voltage with a 0-12V buffer (unity) opamp. The output of that opamp goes through 2 15K resistors to ground with another 0-5V buffer opamp tapping between the resistors as it's input and providing an output to the brain. I have a .1uF cap on the output, as with other analog channels. Ground is connected at the source, and the 5V opamp has its own 7805 (shared with the DACs and other components on my interface board). 12V opamp is supplied with the same 12V I am supplying to the power plug on the brains. On either brain, the software sees higher than local meter readings, by about 0.2 to 0.3V currently (which is significant for my temperature readings (~10 F).
So even a buffer appears like it wont fix this for me...
I'm probably going to need to redesign my interface board... But I've got to figure out what's going on first. The voltage at the input is just not what I see in the software. I've messed with a lot of things and approached it from a few different angles now, and still the problem persits.