A Case Study in Collaborative Human-AI Technical Problem Solving

Featuring the Simplex 2001 Fire Alarm Control Panel

This article documents an extensive collaborative effort between a human technician and an AI assistant in analyzing, troubleshooting, modifying, and reverse-engineering vintage fire alarm control system electronics. Over the course of multiple sessions, we examined dozens of circuit schematics, diagnosed component failures, identified suitable modern replacements for obsolete parts, traced signal paths, compared card variants, and developed novel modifications to extend system capabilities beyond their original design specifications.

This work demonstrates the potential of AI-assisted technical analysis in preserving and extending legacy electronic systems—even when playing a game of telephone with remote troubleshooting, and even when the system turns out to be 45-year-old new-old-stock that had never been commissioned.

The Simplex 2001 fire alarm control panel represents a fascinating piece of fire protection history. Manufactured from the late 1970s through the 1990s, these systems used discrete transistor logic, relay-based circuits, operational amplifiers, and later, CMOS logic ICs to provide life safety functions. Unlike modern addressable systems that rely on microprocessors and software, the 2001 series accomplished its functions through carefully designed analog and digital hardware circuits.

Preserving and restoring these systems presents unique challenges. Original components become obsolete, documentation becomes scarce, and institutional knowledge fades. This case study explores how AI assistance can help bridge these gaps, providing a tireless partner for schematic analysis, component identification, troubleshooting hypotheses, and design modifications.

Visual Communication Through Annotated Schematics

A key innovation in our collaboration was the use of color-coded schematic annotations. The human technician would upload circuit schematics and add colored highlights to indicate specific areas of focus:

• Red lines to trace power supply rails and positive voltage paths
• Blue lines to identify signal paths and control inputs
• Green lines to mark ground connections
• Orange lines to highlight specific circuit paths under investigation
• Yellow highlights to indicate components of interest

This visual communication method proved far more effective than verbal descriptions alone. When debugging a trouble circuit, the technician could highlight the suspected signal path, and the AI could analyze whether the traced route matched the expected circuit behavior.

Iterative Analysis and Correction

The collaboration was genuinely iterative. The AI would propose circuit interpretations, and the human would correct misunderstandings. For example, when analyzing a zone card schematic, the AI initially confused internal board connection points (numbered circles) with edge connector pins (arrows with letters/numbers). The technician's correction led to a clearer shared understanding of the schematic conventions used by Simplex engineers.

Similarly, when analyzing transistor configurations, the AI might propose that two transistors formed a Darlington pair, only to have the technician point out that they were actually independent detectors for different fault conditions—one for open circuits (high resistance) and one for shorts (low resistance), creating a window comparator topology. These corrections improved the analysis and led to accurate troubleshooting recommendations.

The Problem

The original Simplex zone cards used MSCR5035 silicon controlled rectifiers (SCRs) for alarm latching circuits. These Motorola parts are long obsolete. The technician attempted to substitute modern MCR100-8 SCRs, which have similar voltage and current ratings.

The Failure

Collaborative Root Cause Analysis

Through detailed analysis of the SCR specifications and circuit topology, we identified the root cause: gate sensitivity mismatch. The original MSCR5035 was an industrial-grade SCR with robust gate characteristics, likely requiring 5-20mA to trigger. The MCR100-8 is a "sensitive gate" device, requiring only 200µA—roughly 25-100 times more sensitive.

The original circuit was designed with this robustness in mind. With the acknowledge signal (point 8) disconnected, the original SCR tolerated the floating high-impedance gate. The sensitive MCR100-8, however, would trigger from:

• Capacitive coupling from nearby traces
• Power-on transients
• Random electrical noise
• Absence of proper bias reference

The Solution

The Problem

A friend's 2001-2073 Signal card's trouble circuit was completely non-functional. This was a "non-ack" variant panel—the simpler configuration. Even with no end-of-line resistor (a definite fault condition), neither the trouble LED nor the system trouble output would activate.

Collaborative Analysis

We worked through the schematic systematically, first clarifying the notation conventions (internal board points vs. edge connector pins). Through color-coded annotations, we traced the trouble detection circuit and identified that it used a dual-path detection system:

• Q1/Q2 transistor cascade for open circuit (high resistance) detection
• Q3 with zener threshold for short circuit (low resistance) detection

The blue signal path fed both detection circuits through different components—a regular diode and resistor divider for Q1, and a 5.1V zener with different divider for Q3. This created a "window comparator" that could detect both fault conditions.

The Solution

After extensive analysis of the power rails and signal paths, we discovered the root cause: pin 9 (negative flash input) was missing its ground connection. Without this ground reference, the entire detection circuit had no return path and could not function. Restoring the ground connection immediately fixed the trouble circuit.

The Challenge

Simplex 2001 systems came in multiple configurations. The resounding trouble panels (2001-1002/1005 control modules) were designed for zone cards that did not support 2-wire smoke detectors. The 2-wire smoke capable zone cards (like the 2001-1027) were designed for a different control module (2001-1007) that lacked resounding trouble capability.

The technician wanted both features: 2-wire smoke detection AND resounding trouble. This combination never existed as a factory configuration.

The Modification

Through analysis of both card types and the resounding pulse circuits from other cards (like the audio monitor board), we documented the modification procedure:

The Result

Another remote troubleshooting session with the same friend—this time on a different 2001 panel. This one was a "repack" model based on the standard 1007 control card configuration.

The 2001-3052 Sprinkler Alarm card exhibited an unwanted behavior: after reset, the alarm LED would flash rather than remain solidly lit until the next reset. This created timing conflicts with march cards in presignal mode, causing overlapping pulses that produced confusing indications.

By comparing the Sprinkler Alarm schematic (556-566) with the Coder Interlock schematic (556-567)—which used the same PCB but different component population—we identified the difference: R2 was a 10MΩ resistor on the Sprinkler card but a jumper on the Coder Interlock.

The R2/R3/C1 network formed an oscillator within the CD4001 NOR gate logic. By jumpering R2 (eliminating the timing resistance), the oscillation stopped and the LED maintained a steady state.

A significant portion of our work involved comparing different card variants to understand the Simplex 2001 product architecture. Through systematic comparison, we uncovered patterns in how Simplex engineers approached product variants.

Shared PCB, Different Population

Many Simplex cards used identical PCBs with different component populations:

• Sprinkler Alarm (556-566) vs. Coder Interlock (556-567): Same board, different timing components and LED/reset provisions
• Fire Trouble Control variants (556-237 through 556-240): Same board, different meter ranges (0.5A, 1.0A, 2.0A) or no meter
• Zone card variants: Basic vs. acknowledge versions differed in SCR latching circuits, resounding pulse components, and jumper configurations

Product Family Architecture

Analysis revealed that the Simplex 2001 was not a single product but a platform consolidating two predecessor product lines:

This explained why certain zone cards only worked with certain control modules—they represented parallel product lines unified under a single brand but maintaining their distinct architectural philosophies.

Temporal Code 3 vs. Zone Coder Analysis

We compared the Temporal Code 3 card (2001-3053) with the Zone Coder card (2001-3050) to explore potential conversions. Despite appearing quite different—one with jumpers for fixed T-3 patterns, the other with 24 DIP switches for arbitrary code programming—they shared extensive common heritage:

• Same CD4020AE binary counter (IC4)
• Same CD4011AE quad NAND oscillator topology
• Same CD4023AE gating logic
• Same 2N6426 Darlington output driver
• Nearly identical oscillator networks (only R3 and C1 values differ)

This analysis suggested that converting a Zone Coder to produce Temporal 3 patterns might require only changing R3 (39K → 33K) and C1 (0.047µF → 0.22µF), then setting the DIP switches to the appropriate pattern—a much simpler modification than initially expected.

Our work provided insights into fire alarm technology evolution:

• 1970s (420x era): Split between relay-based (4208) and early solid-state (4207) approaches
• Late 1970s-1990s (2001 era): Consolidated platform supporting both philosophies, with increasing solid-state sophistication over time
• 1980s transition (2100/2120): Proto-addressable systems with serial port programming—now largely "lost to time" because site-specific configurations cannot be recovered. The 2001's simpler hardware-based architecture may have actually outlived these more "advanced" systems in practical use.
• Modern era (4100+): Fully addressable, microprocessor-based systems with proper backup and restore capabilities

Based on our experience, we can articulate a methodology for effective AI-assisted circuit analysis:

1. Establish Clear Visual Communication

• Use consistent color coding for different signal types
• Clarify schematic notation conventions early
• Annotate areas of focus rather than relying on verbal descriptions

2. Embrace Iterative Correction

• Expect the AI to make interpretation errors
• Correct misunderstandings explicitly and immediately
• Build shared understanding incrementally

3. Leverage AI for Systematic Analysis

• Component identification and cross-referencing
• Hypothesis generation for failure modes
• Comparative analysis between similar circuits
• Documentation of findings and modifications

4. Combine AI Analysis with Human Verification

• AI proposes, human verifies with measurements
• Physical inspection confirms or refutes hypotheses
• Real-world testing validates modifications

This collaboration demonstrates that AI can serve as a valuable partner in technical preservation and restoration work. The AI assistant provided:

• Tireless analytical capacity: Ability to work through complex schematics systematically without fatigue
• Pattern recognition: Identifying similarities between card variants and product families
• Documentation synthesis: Maintaining context across extended troubleshooting sessions
• Hypothesis generation: Proposing potential failure modes and solutions
• Knowledge integration: Connecting component specifications to circuit behavior

The human technician provided essential capabilities that AI currently cannot:

• Physical interaction: Measuring voltages, swapping components, observing behavior
• Contextual correction: Recognizing when AI interpretations were wrong
• Historical knowledge: Understanding product lineage and design philosophy
• Creative modification: Conceiving novel solutions like the 2-wire smoke + resounding trouble hybrid
• Quality judgment: Knowing when a solution was "good enough" vs. needing refinement

Together, this human-AI partnership accomplished work that neither could have done as efficiently alone. The AI accelerated analysis and provided a sounding board for ideas; the human provided ground truth and creative direction. The result: preserved vintage equipment, novel modifications, and documented knowledge that might otherwise have been lost.

Zone Cards

Control Cards

Signal and Auxiliary Cards

Power and Supervision Cards