[theoldwizard1]

Originally Posted by jeff5may

How will the unit know what sensor is what? Initially, sensors will need to be assigned. How painless can this process be? Maybe the unit can run a "test mode" and find its own sensors.

Highly unlikely !

In order to "find" an input/output "device" means that device must have some intelligence to answer the question "Who/what are you ?" with "I am a ...". Temperature sensor are about as dumb/stupid as they come.

All of this must be "hard coded" in the controller software. Even once you have coded that fact that low pressure compressor input temperature sensor is on analog channel 3, you are only half way home. An A-to-D converter only returns a number of counts. Those "counts" can be converted to a voltage, but we humans understand is degrees F or C. Doing this is actually pretty easy assuming you have the spec sheet that the resistance versus temperature chart for the thermistor (I have never used any of the "electronic" temperature sensors, so I am unfamiliar with them).

But what happens if a sensor or three goes bad in a few years?

Without re-writing the software (changing the resistance versus temperature curve) you had better have an identical spec thermistor.

How will the plumber hook them back up when the (mystery part) is replaced? Will the unit glitch or hang?

Like all electronics, service should be performed the the power off.

What if we want to swap the controller quickly between separate devices? Then what?

You can only do that on identical systems. Do you think Ford or GM uses an identical PCM/ECU on a 4 cylinder and 8 cylinder engine ? Admittedly, a heat pump is likely to have similar sensors and outputs, regardless of its size (within reason).

Finding sensor addresses and data values is not the problem. Recording, logging, manipulating, spitting out, etc. data is not the problem. Making something that runs reliably and is easy to program is the problem.

What is "easy" for me, may be difficult for you !

I believe for the critical sensors (defrost, ambient) I will use 10k thermistors in the first real version. There are 3 more analog pins left open, so that should be enough for the old-school type of hookup. If they were multiplexed, you could hook up over a dozen sensors with 4 pins. That's gotta be enough sensors.

You are getting to wrapped up in the "details" ! You have to see/comprehend the "forest" before looking at the "bark on the trees" (or looking at "the poop on the butt of the ant on the bark of the trees" !)

Designing a "universal" controller that can be [I]re-programmed in the field /I] for different sensors/actuators should be a long term goal. A lot can be done with circuit board switches/jumpers.


The "world" I came from had extremely complex (well, not asteroid landing complex) control algorithms that were constantly being "improved"/modified. The sensor data was obfuscated by input routines that handle converting from counts to an "engineering unit" and then storing the value in a variable (RAM location) that had a meaningful name (i.e. Compressor_Low_Side_Pressure). An algorithm that controlled one "subsystem" (say, condenser fan speed) could be ported from one system with one set off inputs and outputs to a different system with different inputs and outs by changing a few #defines in the source code and then re-compiling.

This is probably not what you want to hear, but this is reasonable.


KISS !