Tech developer Jeff Geerling restomodded a 1995 TrueTime XL-AK industrial GPS time server using a Raspberry Pi 5 and a Jimmy Paputto L1 GNSS HAT in June 2026. Inspired by a major Telstra network outage caused by a legacy timing appliance bug, the project delivers precise Stratum 1 network time while retaining vintage hardware aesthetics.
The intersection of 1990s industrial hardware and modern single-board computers rarely looks this clean. Back in June 2026, developer and maker Jeff Geerling acquired a TrueTime XL-AK. It’s an industrial-grade GPS time server manufactured in 1995, originally built to maintain hyper-accurate time synchronization via satellite constellations. Geerling wanted to explore the engineering history behind satellite timekeeping. He just didn’t expect a real-world disaster to immediately validate the project’s entire premise.
Barely sixteen days after the hardware arrived on his desk, a nearly identical legacy time server triggered a catastrophic 12-hour blackout across Telstra’s mobile network in Australia. Millions of customers lost voice calls, electronic payment capabilities, and railway communications. That glaring vulnerability in critical infrastructure turned a retro tech experiment into a timely masterclass on the dangers of aging enterprise timing systems.
Why the Telstra Outage Forced a Rethink of Legacy Time Servers
According to reporting from Hackaday, the July 8, 2026, Telstra blackout crippled services for 8.8 million people nationwide. The root cause was a Microchip SSU 2000—an NTP appliance dating back to the early 2000s. During a routine maintenance task, the server rebooted. Upon restarting, it rolled back its internal clock to the year 2006 instead of 2026. This triggered a known GPS rollover bug that the manufacturer had documented as far back as November 2000.
Telstra’s technical teams were fully aware of the risk. Hackaday noted that the vendor had issued patch reminders for the GPS module in both 2020 and 2022. However, internal tracking failures led to a decision in January 2026 not to act because the change lacked proper documentation in enterprise systems. The resulting cascade caused digital certificates signed with incorrect timestamps to be rejected across the entire network. Time synchronization took until 16:00 local time to contain.
TechRepublic highlighted the sweeping operational impact. Alongside consumer mobile failures, railway communications collapsed across Victoria and New South Wales. EFTPOS electronic payment terminals failed in stores, and at least 604 emergency calls to the 000 dispatch failed to connect during the critical window. Telstra CFO Michael Ackland confirmed publicly that cybersecurity exploits were entirely absent. He pinned the blame squarely on software defects embedded inside the Sydney and Melbourne time servers.
Inside the TrueTime Pi Hardware Architecture
Rather than letting vintage gear gather dust, Geerling gutted the obsolete internal logic of the TrueTime unit while preserving its original enclosure and front-panel interface. The modern rebuild relies on a lean hardware stack designed for high reliability and low thermal drift.
At the heart of the build sits a Raspberry Pi 5 with 4 GB of RAM, though Geerling noted that a base 1 GB model priced at $44 easily suffices for pure NTP workloads. Attached to the Pi is a Jimmy Paputto L1 GNSS HAT. This expansion board features a u-blox NEO-M9N receiver that extracts position and time data from GPS satellites, outputting an exceptionally precise Pulse-Per-Second (PPS) hardware signal.
Software-wise, the system runs Chrony as the NTP daemon. Chrony ingests two primary timing references: NMEA data streamed via shared memory and the direct PPS signal wired straight into /dev/pps0. To maintain the retro exterior, Geerling reconnected the original Densitron 16×2 LCD screen directly to the Pi’s GPIO header using wire wrapping—a technique forming gas-tight connections around square pins with lower electrical resistance than standard solder joints. A custom 3D-printed mounting bracket and a bicolor LED round out the physical build.
Thermal Optimization and Precision Tuning for Stratum 1 Accuracy
Getting Chrony running is straightforward, but achieving true Stratum 1 stability—meaning a direct reference to an atomic or satellite clock—demands rigorous hardware tuning. Temperature fluctuations directly impact the internal crystal oscillator of the Raspberry Pi SoC, introducing timing jitter.
Geerling experimented with various configurations to isolate the oscillator from thermal drift. Adding force_turbo=1 to /boot/firmware/config.txt locks the SoC frequency permanently at 2.4 GHz. While this draws roughly 1 watt of extra power, it eliminates micro-variations caused by dynamic CPU load scaling. He also pinned the cooling fan to a constant 50% duty cycle, striking a balance between optimal thermal dissipation and desk-friendly acoustic levels.
To eliminate interrupt latency, Geerling isolated CPU core 3 by adding isolcpus=3 to the kernel boot parameters, routing all PPS hardware interrupts directly to that dedicated core via systemd. He also shielded the underside of the Pi—where the physical oscillator resides—to protect it against drafts from external HVAC systems that measurements showed could destabilize the clock.
Geerling also added support for legacy network standards, including RFC 867 (Daytime Protocol) and RFC 868 (Time Protocol). By enabling xinetd under Debian, vintage machines can query the server over TCP and UDP. For attendees at the VCF Midwest retro computing convention, he integrated a Docker container running netatalk with TimeLord. This serves AppleTalk time and AppleShare file services, letting visitors sync vintage Macintosh computers using the Tardis plugin just as they did thirty years ago.
Infrastructure Lessons for Modern Startups
While building a retro GPS appliance makes for an engaging weekend project, the underlying engineering takeaway is deadly serious for modern SaaS, B2B, and fintech operations. Distributed microservices, TLS certificate validation, JSON Web Token (JWT) verification, and financial ledgers all rely fundamentally on accurate time synchronization.
Startups must audit their infrastructure dependencies. Relying on an unverified, single-source cloud NTP provider without internal redundancy introduces a single point of failure. Implementing a drift budget helps define acceptable tolerances—such as maintaining clocks within ±10 milliseconds of UTC—and triggers automated alerts before token validations start failing across distributed nodes.
Teams can verify local NTP quality immediately using diagnostic utilities like chronyc tracking or ntpq -p on any production Linux instance. High stratum levels exceeding 3 or jitter measurements climbing past 10 milliseconds indicate a weak timing chain. Deploying independent NTP sources alongside a fallback internal GPS hardware appliance eliminates an entire class of obscure synchronization bugs for under $100 in components.
Documentation remains the ultimate safeguard against outages. The Telstra incident proved that unrecorded legacy configuration choices can block emergency patches decades later. Keeping precise operational logs ensures that when system clocks drift, engineering teams spend time fixing the code rather than guessing what changed.
