allenneuraldynamics / harp.device.treadmill Goto Github PK

As-is, the resistor wattage rating is a little low for the amount of current they could theoretically see. Let's bump this up to 127Ohm, 1/10W to be on the safe side.

• raw position
• raw torque (reaction torque from mouse)
• raw current (applied torque from brake)
• 5 register spaces
• aggregate register of the first 3 registers in an array: [raw position, raw torque, raw current]
• raw sensors dispatch rate hz
• current setpoint
• "tare" (position, current, torque)
• Error register
  • brake disconnected
  • current setpoint and sensed current mismatch. (out of range)

This driver is copy-pasted into two different repos: this one and the lickety split repo (bad!). Let's tease this device into its own repo that can be submoduled and compiled into other projects like the Harp Core.

The treadmill uses Hammond Mfg pn: 1455L801, but we will need to make custom side panels for it that expose all connectors.

The voltage references for U9 and U17 should be 3.3V, not 1.65V.

Like the Lickety-Split, we need to make this connector extend a bit further out so that the plug makes a good connection to the socker

Since cables can be unplugged, it would be ideal to be able to check the presence of all sensors and actuators.

Practically, we might not be able to detect all devices, but checking as many as possible would be a good start.

Can the treadmill be hot-plugged? The use-case is to be able to install-and-remove the treadmill platform where the mechanical hardware, sensors, and brake are a single unit. This is currently going to be done with an electrical spring-based contact pins. We need to make sure every io pin (brake, encoder, transducer) can be unplugged or plugged-in with the device turned on without damaging anything.

Currently, we use a Hammond Manufacturing 1455L802 enclosure (drawing) for this project, but its worst-case tolerances are 99.5mm, which means our PCBs will be too big in some cases. We need to adjust our PCB to accomodate this worst-case dimension.

The pico-sdk has a fairly heavy footprint (on the order of gigabytes).

We should enable some sort of option to not need to clone the pico-sdk if the user can specify its location elsewhere. We could do this with some sort of environment variable definition, like $PICO_SDK_DIR, that gets checked beforehand.

We should be able to:

• write a new 16-bit value to it (i.e: put something in the TXFIFO).
• dispatch the value over SPI.

The harp core repo should be a submodule in this repo.

Before that, the harp-core repo needs to be refactored with an updated CMakeLists.txt.

The Treadmill has 3 sensors:

• brake current
• encoder
• torque sensor torque

We should be able to dispatch these readings as periodic Harp Events at a configurable interval (up to 1 KHz).