The RFID readers communicate using the SPI protocol. SPI was originally designed for short-distance, high-speed communication on a single PCB. However, Musical Steps required SPI to function reliably over distances of up to seven feet, well beyond its intended range.

This limitation was a known concern from the outset and became a central focus of early unit testing. In those tests, I found that the RFID readers operated reliably at short distances or in small groups, but began to fail and become unresponsive when the wiring extended beyond four feet.

I achieved early success by combining several signal integrity strategies:

• Reducing the SPI clock speed from 1 MHz to 500 kHz.
• Adding 100Ω series resistors on the SCK, MOSI, and CS lines to dampen reflections.
• Alternating GND and signal lines within each ribbon cable to minimize crosstalk and ensure clean return paths.

With these improvements, I was able to reliably operate two RFID readers simultaneously over distances of up to 7 feet.

But scaling beyond that, even to three readers, remained unreliable.

I tried just about every trick in the book: adjusting clock speeds further, experimenting with multiple RFID and Software SPI libraries, and even borrowing Professor Allan Weber’s portable oscilloscope to diagnose the issue.

One idea remained. Since I had verified that the readers worked reliably in pairs, I theorized that I could run SPI over five independent sets of Arduino pins: one for each pair of readers. The catch? I could no longer use the Arduino’s built-in hardware SPI library.

So, I painstakingly rewrote the entire SPI communication for the RFID readers in software.

And, it worked! With a bit-banged implementation of SPI, long-range communication was no longer an issue. The RFID readers responded reliably, even at full distance.