Add Support for MCP3002

[brentru]

Support for the MCP3002: 2.7V Dual Channel 10-Bit A/D Converter with SPI Serial Interface

Datasheet: http://ww1.microchip.com/downloads/en/DeviceDoc/21294E.pdf

[2bndy5]

I have limited access to a couple MCP3002, so I plan on addressing this issue when I snag them from my robotics club's supply. I have started a branch to preempt testing

[ladyada]

awesome, thank you!

[2bndy5]

@brentru I've had to alter the AnalogIn.__init__() a little to allow for the new MCP3002 class... I have a few questions:

1. Does the negative_pin parameter actually need to be saved to the object instance (self._negative_pin)? It seems like it is only used during instantiation to modify self._pin_setting accordingly, but then it is never used again.
2. What's up with the __getitem__()? I understand the concept of the function, but in AnalogIn it returns an element of self._channels. I see that self._channels is declared as a blank list when discerning the differential pin mapping, but then nothing is added to the that list. So wouldn't __getitem__() throw an IndexError exception? The way it currently written, __getitem__ only applies for differential reads, and I see no chan[<pin>] in any of the examples. Is the __getitem__() meant to help mimic another type of analog input object like analogio.AnalogIn?

I don't mean to pick apart the code, I'm just trying to understand some design decisions.

[brentru]

@2bndy5 Sorry for my delay in responding, it's exam-season for me.

I don't mean to pick apart the code, I'm just trying to understand some design decisions.

You're not picking apart the code, you're improving it while adding a new chip!

1. _negative_pin can be removed here (https://github.com/adafruit/Adafruit_CircuitPython_MCP3xxx/blob/master/adafruit_mcp3xxx/analog_in.py#L42). Instead, use negative_pin here: https://github.com/adafruit/Adafruit_CircuitPython_MCP3xxx/blob/master/adafruit_mcp3xxx/analog_in.py#L51

2. You can remove the __getitem__ method there since AnalogIn's API doesn't mention it (see: https://circuitpython.readthedocs.io/en/4.x/shared-bindings/analogio/AnalogIn.html).

[2bndy5]

I was hoping you'd say something like that. I've actually made quite a few "improvements" to all the MCP classes. Some for reduced computation, and others just avoid excessive Python garbage collection. My branch has all the commits. It is working for single ended reads on both MCP3008 & MCP3002, but I want to double check the "psuedo differential" reads work properly for both before creating a PR. Please continue focus on your exams 🙂

[brentru]

Fixed by #20. Merged into master and released as 1.2.0.

Closing out...