Nick Steele Reverse-Engineered Digital Calipers Into an Open-Source $32 Linear Encoder, Gapped With Two Layers of Kapton Tape

Chris Pirillo, drawn as a blueprint cyanotype, presses a strip of tape along a long sensor board beside an opened digital caliper with a blank display. — Illustration: LockerGnome

“I started by reverse-engineering digital capacitive calipers,” Nick Steele writes in the project’s README. “I wanted to publish this working capacitive encoder because I had trouble finding open information about them.”

He went as far as the silicon. In the KiCad forum thread where he posted the design on October 17, he describes opening up a cheap dial indicator: “after decapsulating the die, I found it was the same flip-chip part that the capacitive digital calipers always seem to have,” he wrote.

His version swaps that mystery chip for parts you can order. A sensor board carries eight repeating transmit antennas he calls emitters. A passive reflector board with a row of T-shaped copper pads runs the length of the axis. “They do not touch. The copper faces sit about 0.1mm apart.” An STM32L4 steps the emitters through a 16-step pattern at 16kHz, reads the charge that couples back through the reflector, and sends a position over UART every millisecond.

The gap took the least engineering. “I tried a 3D printed shim to set the gap, but two layers of Kapton tape turned out to be much easier and more accurate,” the README says.

The accuracy took the most. “When I started, I hoped for 10µm electrical accuracy. My first revision was about 0.5mm.” The 40µm rating came from plots against the Mitutoyo, which he “read the accuracy figure visually from the plots and rounded it up conservatively.”

What it still can’t do is remember where it is. “At power-up it assumes it is at position zero, so it needs zeroing or homing every time it starts.” Steele says he’s “pretty dead-set” that the next revision gets “two measurement scales for absolute position sensing.”

Why it matters

That’s the cheap caliper’s oldest flaw. “The main annoyance of those cheap supermarket calipers is that when you need them, the battery is empty. The reason for that is that it is an incremental encoder,” forum member MartinS replied. To skip re-zeroing, the measurement has to keep running, which drains the cell. An absolute scale knows its position the moment it wakes up.

Custom silicon still wins on price. Steele priced out adding a display to make a full caliper. “After adding a display, the materials cost per build would be on the order of $100 right now,” he wrote. “While I don’t expect to be competitive with custom silicon, at that price it didn’t even seem reasonable to pursue.” The open board’s job is a DRO sensor for your own machine, not a cheaper caliper.

The board stack is a measured choice. He ordered the sensor in two-layer and four-layer versions. The two-layer one showed extra error “on the order of 0.1mm” from asymmetric coupling, and the four-layer board’s internal ground planes fix it, per the hardware README.

The chip that makes it small is getting scarce. The design leans on the STM32L4, and the README lists “less reliance on the STM32L4, which is getting harder to source” as a goal, with a plan for “a more standard STM32 with a discrete analog frontend.”

Steal this

Before you put any incremental scale on a DRO or a printer axis, give the axis a homing switch or a zero routine, because this kind of sensor “needs zeroing or homing every time it starts.” This is like a trip odometer that resets every time you turn the key.
When you need a small, repeatable gap between two flat parts, try layers of tape of known thickness before you print a shim. Steele’s printed shim lost to two layers of Kapton. Kind of like leveling a wobbly table with a folded receipt instead of carving a wedge.
If you mount a capacitive encoder on a machine that throws coolant, cover it first: the README warns it “is sensitive to liquids between the boards,” and on long axes thermal expansion matters too. It’s like a phone touchscreen in the rain, still working, just not telling the truth.

Every repeat in this design is imperial: the emitters sit every 0.635mm, a period every 5.08mm. “I do prefer metric, and the next version probably will be,” Steele said in the thread. That next version is supposed to be absolute, metric and off the hard-to-find chip at once. Another forum member has already offered to write a script that generates the reflector board; Steele told him to send a pull request. Whether a parametric reflector generator lands before the absolute scale does decides how fast anyone else can cut one to the length of their own machine.

More Tools & Gear