Programmable Fluid Transfer System

How am I going about building this thing?

I have a gut-level reaction to expensive, locked-down tools that sit just out of reach of the people and labs who could move the needle but lack the force-multiplier of automation. The best thing I can do with my time is build alternatives.

I designed the first version of this pump on my bedroom desk. I was fresh out of college, no EE lab, no institution backing me, and a 0% APR credit line. I picked a bare RP2354A microcontroller instead of a dev-board stamp because I wanted to learn how to build the real thing. I designed the schematic to be as boring and reliable as possible so I could spend the risk budget on the build itself. The board was paste-stenciled on a 3D-printed jig because I didn't want to buy one, hot-reflowed on a hot plate, and hand-assembled under my stereo microscope.

When I needed thermal data on my motor driver and didn't own a fancy thermal camera, I built a fixture from a smartphone thermal dongle and had it working by the afternoon. When the linear-stage vendor didn't publish CAD, I picked up my calipers and modeled it. This is an exercise in "if it doesn't exist at a price you can afford, build it".

I'm not building this for the labs that can already afford a Harvard Apparatus. I'm building for the PhD student who lost their lab budget to NIH cuts, the biohacker running serial dilutions who doesn't have enough hours in the day, and the artist making small-batch fragrances in their kitchen who needs batch consistency. The bottleneck is the same for all of them: cheap equipment isn't designed with the human in mind, and the best equipment is priced out of reach. 

Syringe Pump Controller — RP2354A + TMC2209

A 4-layer board for a syringe pump: RP2354A microcontroller, TMC2209 stepper driver, dual PWM-heated reservoirs with PID, force-sensing resistor for plunger feedback, and limit switches for end-of-travel. USB-C for logic, 24V barrel jack for motor and heaters.

This is the first complex board I've designed without a dev-board stamp for the MCU. I designed it to be boring on the schematic and unforgiving on the build — paste-stenciled, hot-plate and hot-air reflowed, hand-placed under a microscope on my bedroom desk. With one shot at JLC turnaround and no EE lab, I put the risk budget into the build, not the schematic.

Designed for debug

The board has 17 test points, including probe pads on every rail (+24V, +12V, +3.3V, +1V1, GND) and on every signal that matters during bring-up including the stepper motor rails. A 16-pin header breaks out 13 GPIO plus power and ground for any expansion or instrumentation I haven't thought of yet.

The dual-rail power separation is a debug feature, not just an electrical one: I can program and exercise firmware on USB alone, with no 24V supply anywhere near the bench, until I'm ready to drive the motor and heaters.