XIAC Museum — 2007 — PDI-1206

The story

The accuracy of linear-fed counting machines lives and dies by the timing of the memory gates. When a gate is only a few milliseconds late (or rebounds at impact), you can get launches, splits, missed catches, and the classic overcount/undercount pair.

In the field, fault-finding could be painfully slow — especially when the problem was intermittent. The common “fast fix” was to replace cylinders and valves in bulk, with no guarantee the root cause was actually removed.

Outcome: PDI‑1206 became a practical instrument for quickly separating electrical, pneumatic and mechanical causes — and it laid the foundation for later in-production real-time validation electronics.

Design goals (the non-negotiables)

Before building anything, the must-have list was clear: fast, repeatable, operator-friendly, and machine-safe.

• Self-calibrating — no specialist technician required
• Quick — around a 10-minute check, practical as a routine
• Minimal disassembly — connect like the unit itself
• No external power — powered from the machine
• Portable — one instrument for many machines of the same type
• Meaningful outputs — a simple channel score plus deeper component breakdowns

The “dud mouse” beginning

The original hack. A sacrificial computer mouse used as a quick carrier for early electronics and wiring. This experiment directly kicked off the PDI‑1206 project.

The first breakthrough wasn’t a polished device — it was proof. A hacked enclosure, a scope, and stubborn troubleshooting turned a vague customer report into an observable problem.

Prototype electronics (PIC16C73 era)

Early boards. Control + I/O prototypes in the UV-erasable PIC era.

Bench reality. Prototypes, probes, and lots of iteration — the fast path to understanding.

UI bring-up. Simple display and indicators — focused on truth, not decoration.

Optical flap-speed sensing

A key development step was an optical reflective sensor: two infrared emitters with a photodiode between them. The emitters shone into the channel, and the moving flap reflected a portion of light back to the photodiode.

Sensor construction. Two IR emitters + photodiode, designed to look down into the channel.

Homogeneous field. The reflection profile was intentionally even across the channel width.

Position → speed (by differentiation)

Signal insight. The optical signal (position proxy) was differentiated to reveal flap speed — making slow, sticky and bouncing behaviour visible.

On-channel sensing. The sensor mounted above a CF‑1220 channel, watching the flap pass beneath.

From electronics to enclosure

Making it real. The shift from prototype to portable instrument starts at the front panel.

Inside view. Compact, serviceable wiring and a dedicated PCB for reliable field use.

Completed unit. Portable, self-powered, and designed to connect in seconds.

On-machine validation

Cremer CF‑1220. Testing in real machine conditions — no lab assumptions.

Later evaluation. Archived photos show the instrument was also tested on another Cremer platform at a later stage.

This step mattered: intermittent faults don’t live on the bench — they live inside real machines with real air systems, real wiring, and real wear. PDI‑1206 was designed to measure exactly that.

Legacy: foundation for real-time validation (2015)

2015 in-production validation. Real-time validation circuits mounted inside the memory flap unit, underneath the cylinders (holes for cylinder mounting visible just above the hall sensor board).

The PDI-1206 established the core validation philosophy: measure dynamic behaviour and convert it into actionable pass/fail information. By the mid-2010s, these ideas had evolved into the first in-production real-time validation electronics, embedded directly into memory flap units and operating continuously during normal machine use.

One of the key constraints during this 2015 development phase was budget. The project had to be executed with tightly controlled costs, which meant that design decisions were often made with a strong preference for getting things working on the first iteration. If something was functional but not ideal, it was sometimes accepted rather than risking the cost of second or third redesign cycles.

This was particularly true for PCB development. At the time, prototype PCB fabrication was significantly more expensive and slower than it is today, making multiple design spins difficult to justify. As a result, the 2015 architecture represents a careful balance between innovation, practicality, and the manufacturing realities of that period.

Modern validation systems benefit from a very different landscape. Affordable rapid-turn PCB manufacturing now allows aggressive iteration, tighter integration, and architectural refinements that were simply not economically feasible a decade earlier. The underlying validation principles remain the same — but today they can be implemented with far greater precision and completeness.

Museum takeaway: the tool came first — and the production systems followed once the physics was understood.