The printed StackCalc enclosure is not a generic case waiting for an electronics design. Its CAD reads fixed coordinates from the KiCad board so the rails, display opening, key interfaces, and clearances are designed around the same board that is now in fabrication.
The difficult part of RPN is often not arithmetic. It is knowing where a value is, when the stack will lift, and what a binary operator will consume. A screen can show those facts after a keypress; the StackCalc Learning Lab asks a learner to predict them first.
A calculator can look consistent across devices while disagreeing at exactly the moments that teach people how RPN works: ENTER, stack lift, an error state, a program step, or an unfinished number. We did not want the Watch, iPhone, and eventual handheld to agree only on the happy path. StackCalc has one calculator core so a named key sequence has one meaning before any surface decides how to draw it.
Power management begins as a behavior problem before it becomes a battery-life number. The StackCalc firmware has to decide when to scan the matrix, wake the display path, skip an unchanged frame, save state, and return to a lower-work loop. We can exercise those decisions in the RP2350 emulator while the PCB is in fabrication.
An expression tree can make postfix notation less mysterious: finish each branch before visiting the operation above it. The Learning Lab expression-tree board turns that traversal into an arrangement of removable tokens.
Fractions create two useful RPN questions: what quantity does a numerator and denominator represent, and what sequence turns those quantities into a result? The Learning Lab fraction studio keeps those questions side by side instead of asking a learner to jump straight to a symbolic entry.
“Tool-free” is easy to write on a feature list and hard to make useful in a printed enclosure. The test is not whether an assembly has no screws; it is whether a person can insert, retain, inspect, and later release the parts without turning every opening into a breakage risk.
The current StackCalc assembly is an unpowered, four-part mechanical prototype: a tapered chassis, unified faceplate, TPU membrane, and snap-retained top cap. Its RP2350 KiCad board is in fabrication, but no assembled board has been installed, so this remains a mechanical design lesson rather than an electronics assembly claim.
RPN calculators already exist. StackCalc is not a nostalgia project trying to recreate a familiar beige machine. The gap we see is a learning path: a way to move from a familiar screen to a physical instrument while understanding what the stack is doing.
That is why StackCalc joins the current iPhone and Apple Watch apps, a shared calculator core, Learning Lab prototypes, and an RP2350 handheld whose board is in fabrication around the same calculation behavior. The connection is semantic, not cosmetic.
For many children, the first personal computer they wear is not a phone or laptop. It is a watch, often after an iPad. That changed where we started StackCalc. The Apple Watch was not a companion screen added after the calculator had been designed; it was the first place we had to prove that an RPN calculator could explain its next action in a very small space.
The current StackCalc prototype is a 3D-printed, tool-free mechanical assembly. Its RP2350 calculator PCB design is in fabrication, but no board has yet been assembled and powered, so we cannot yet make claims about powered bring-up, battery life, display behavior, or key feel on the finished electronics.
That is why the firmware simulator is useful. It gives us a fast way to exercise the calculator software and its proposed row-and-column interface while the physical electronics are still being developed. It is evidence about the simulated build, not evidence that a physical PCB works.