Add additional ADC chips

[lurch]

A little "reminder" issue, as requested #41 (comment)

As I mentioned at #156 (comment) it'll hopefully be fairly straightforward to also add classes for the MCP3001, MCP3002, MCP3201 and MCP3202 ADC chips too.

[pcopa]

Yes this should be simple, just a matter of adjusting the channel count. This would address all the spi ADC chips (that I know of).

[lurch]

And looking at https://raw.githubusercontent.com/RPi-Distro/python-gpiozero/master/docs/images/composite_device_hierarchy.png it would seem to make sense for MCP3001, MCP3002, MCP3004 & MCP3008 to inherit from MCP30xx; and MCP3201, MCP3202, MCP3204 & MCP3208 to inherit from MCP32xx?
Or would that just be taking things too far for the sake of 'neatness / symmetry' ?
(Is this something that can be squeezed into 1.1 ?)

[bennuttall]

• MCP3001
• MCP3002
• MCP3004
• MCP3008
• MCP3201
• MCP3202
• MCP3204
• MCP3208

[bennuttall]

I'm adding these now. Got MCP3002 and MCP3202 in.

MCP3001 and MCP3201 seem to have two input channels but +/- rather than CH0/CH1 - I'm not sure what that means. Is it that one to read an input as 0->max, the other max->0? (if that makes sense?)

[bennuttall]

Reading up on datasheets it's definitely one channel, I just don't understand what the IN+ and IN- are. I'll add them as I have them, and they can be fixed before release if there's something else needed.

[waveform80]

That means that like the 3301 they operate in differential mode only (and I'd guess they don't require any input in that case); i.e. you're expected to connect both. However, if the chip's like the other 32xx and 30xx's it won't support fully signed operation; i.e. IN- will only be for background noise compensation, not for measuring a fully negative value.

[waveform80]

Have a look at the existing MCP3301 class and that should shed some light on things.

[bennuttall]

Yeah, just spotted it! That makes sense.

I was going to leave the channel parameter in, even though it has to be 0, just so the arguments were the same as the others, but I see you removed it from MCP3301 so I will too.

[waveform80]

Yeah, not much point having a parameter you can't do anything with :)

[bennuttall]

Yeah

[bennuttall]

Pushed - please take a look!

[waveform80]

I think you need to remove differential=False from the xx01 constructors; those chips will very likely operate in differential-only mode; i.e. firstly you shouldn't be able to specify differential and secondly its value should be True.

[waveform80]

Yeah, just skimmed the data-sheets and they're both differential only. Also, because both of them require no input at all you'll need to push MCP33xx._send up into MCP3xxx and override it in the xx01 classes to just return [0, 0] (to allow two bytes worth to be received without sending anything).

[bennuttall]

Like that?

[waveform80]

Yup, that's the first bit - now you need to deal with the fact they don't expect any inbound data (and will only send back two-bytes worth when the SPI CLK activates)