I just found this post. Hopefully I'm not too late to chime in.
Let me provide my input by means of a story.
First, my background. I have been an embedded developer for the past 15+ years using about a half-dozen different architectures at the C, assembly, and HDL level. My passion is to make things work and to control and interface to the real world without sucking. I discovered MicroPython during the ESP8266 kickstarter. I was instantly intrigued as I have been using Python to solve problems on the PC side of things for many years. When the opportunity presented itself I purchased a half-dozen pyboards for a rapid development project at work. I estimate the pyboard cut my development time to a tenth of what I would have expended with a traditional microcontroller development; nevermind the fact that I had never used it before (Hats off to the pyboard and MicroPython!).
Shortly after this experience I was talking with one of my employer’s most senior engineers. He is an idea man. He would like to build controllers for his home solar hot water system. He would like to automate his PV hybrid battery charging system. These are just the start of his ideas. He has the knowledge and the desire but he doesn’t have the time for the boring and painful parts of embedded development. He certainly doesn’t have time to read the datasheet to see how to setup the ADC, SPI or the PWM modules--nor should he have to.
I told him what I had discovered with the pyboard--and how it had made microcontrollers more fun and not tedious as they so often are. The highlights are that the common microcontroller peripherals (and everything on the board) are fully supported, built-in, and easy. Plug-Import-Play, so to speak.
After I had pitched the merits of the pyboard and MicroPython he asked the simple question:
How do I hook it up to the world?
That, I realized is what the pyboard lacks. There are no relays. There are no thermistor/thermocouple inputs. There is no high-endurance FRAM or EEPROM. There are no analog buffers on the ADC inputs that can handle outdoor/industrial voltages. There is no 16x2 LCD interface. The only button is TINY. There are no stepper motor controllers. If you want real-world interface for the pyboard you have to design your own board or deal with a rat's nest of proto boards stitched together--which moves the project back to tedious, un-fun, and impractical for those of us with lots of ideas and very little time.
At first I thought the answer was a micropython board with the ability to use Arduino sheilds. The problem with this approach is that there are so many options for shields that the support complexity increases quickly. It would be near impossible to support more than a small combination of shields from a few vendors (and, as others have noted, certainly not economical).
My current wishful thinking is for a board or an expansion board with real-world (industrial/scientific) interfaces that are fully supported by MicroPython so that the plug-import-play ease of the pyboard remains. I asked my senior engineer friend about what he would like to see and this is the summary of our brainstorming for an outdoor, data logging and controller application board:
- First, keep the basic pyboard features!
I/O with screw-type terminals and transient protection
USB UART and mass storage
Solar/interruptible power aids
- Buffered analog inputs to support wider input voltage ranges (+/- 10V or 0-28V)
“Open collector” digital outputs
“Ground-open” or “volt-open” digital inputs
2 or 3 relays suitable for AC/DC circuit switching
Rugged communications interfaces, again with transient protection in mind
- Battery or supercap backed, temperature stable RTC
Lithium battery charger/fuel gauge
High write endurance NVM
More “appliance” friendly user interfaces
- 10/100 (wired) Ethernet
RS422 or RS485 serial port(s)
- 16x2 character LCD module
Small key matrix (either on-board or external)
To support these more “real time” operations I would be interested in C implementation of python modules that could support tunable digital filtering of the analog inputs and debouncing of digital inputs. I’d love to see C implementations of tunable PID modules that can drive DAC or PWM modules. The idea being that once initialized, and connected these modules would operate “in the background” with higher and more deterministic performance while leveraging the ease of MicroPython to stitch the system together.
Yes, wishful thinking I know. My take is that MicroPython could be today's Basic or Forth once we can easily hook it up to the world around us.