Why I Built This

It started with a $5,600 repair bill. Then it kept going.

I didn’t set out to build a product. I set out to understand my own car.

In early February 2025, on a snowy day with ice on the road, my 2017 Tesla Model X 100D slid and hit a bridge pier head-on. The exterior didn’t deform. The car drove normally afterwards — I drove it to the body shop myself. I figured it was going to be a bumper-scuff job, maybe a sensor.

Two and a half months later, the body shop handed me a $25,905 CAD invoice. Once they’d pulled the front end apart they’d found the subframe, the frame rail, the A/C condenser, the headlights, and the air suspension compressor all needed replacing. The car spent 8 hours on a Celette frame bench. ICBC covered the claim; I paid the $500 deductible.

That was the first lesson: on a modern Tesla, what you can see from the outside — and what the car drives like — has almost nothing to do with what’s actually wrong underneath.

When I finally got the car back in mid-April, I didn’t fully trust that the body shop had caught everything. The invoice listed a new frame rail, a new subframe, 8 hours on a frame bench — all from a slide that had left the car still drivable. So I booked a Tesla post-collision inspection for two weeks later. On 2025-04-30, Tesla’s service center ran their standard air suspension leak test — which in 2026 is still: put the car in “very high” mode, tape all four corners, leave it overnight, come back in the morning to measure how far it sagged. Both front corners had dropped. Tesla confirmed both front struts were leaking. That inspection cost $619.

That’s when the dispute started.

I took the car back to the body shop and asked them to repair the leaking suspension under the collision claim. Three parties, three positions:

• The body shop insisted the leak was not caused by their repair process. On whether the collision itself could have caused it, they deferred to ICBC.
• ICBC insisted it was normal wear-and-tear on an 8-year-old car, and asked me to provide evidence that it wasn’t.
• Tesla — the only party with actual diagnostic data on the air suspension — refused to take a position on whether the failure was collision-related or age-related. No comment on probability.

Two weeks of back-and-forth between the three, car sitting at the body shop the whole time, no movement. I finally gave up on the insurance route and decided to just pay Tesla out of pocket to fix the suspension. I called the body shop to tell them I was coming to pick the car up.

The car wouldn’t start. Not “slow to start” — nothing. No screen, no lights, no click. The body shop had noticed that same day when they tried to move it around the lot and got nothing. They’d swapped in a fresh 12V battery, assuming that was the classic Model X first-response failure mode. Still dead. Zero response.

I had the car flat-bedded from the body shop to the Powell St Tesla Service Center in Vancouver.

That tow turned into the most expensive service visit of the year. Tesla found three things on the same work order:

1. Both front air struts leaking — the thing I already knew. Replaced.
2. The HV battery pack had actually failed. That’s why no amount of 12V swapping was going to bring the car back — the DC-DC converter had nothing to pull from. Tesla replaced the HV pack for free under the HV Battery Limited Warranty.
3. Since the car was already in the shop, I took the opportunity to retrofit MCU1 → MCU2 + HW3 autopilot hardware.

Repair portion of that visit: $5,612. Voluntary MCU upgrade: $3,024.

The HV battery failure was the part I couldn’t see coming. If I’d had continuous pack telemetry for the six months leading up to that morning, there would have been signals — cell spread widening, pack internal resistance creeping up, charging behavior shifting. Tesla has that data. They don’t share it with owners. They told me about it after the car was already on the flatbed.

That was the moment I stopped being frustrated with Tesla and started writing code to read the CAN bus myself.

By the time I tallied it up, my 12-month Tesla service total — on top of the ICBC collision claim — was $17,009 CAD. Combined with the body shop, the year cost $42,914.

12 months on a 2017 Model X — the real numbers

* Also on the 2025-05-30 visit: a $3,024 voluntary MCU1→MCU2 + HW3 autopilot retrofit. Not listed above — it’s a feature upgrade, not a repair, so it doesn’t belong in the failure-cost narrative. Mentioning it for transparency.

Year totals:

• ICBC collision claim: $25,905 (insurance paid; I covered $500 deductible)
• Tesla service & follow-ups: $17,009
• Combined: $42,914 CAD

These are all real invoices, real dates, on a single car. Some of it was unavoidable — a frame-bench job after a collision is a frame-bench job. But the slow leaks, the missed sway bar end links, the HV battery decline that happened to coincide with the collision — those are exactly the kind of things a continuous monitoring system could have flagged earlier or caught while the warranty was still covering them.

That’s the gap T800 is built to close.

The question that started everything

Was the suspension really healthy now? Was the 12V battery stable? Were there early warning signs I could have caught months earlier?

No one could give me clear answers — not Tesla, not ICBC, not the repair shop. So I started collecting CAN telemetry data from my Model X and analyzing it myself.

Over several months, I reverse-engineered 230+ CAN signals across 6 vehicle subsystems: battery health, air suspension, 12V system, thermal management, drivetrain, and charging efficiency.

That work became T800.

What T800 is

T800 is a long-term health monitoring system for Tesla Model S and Model X. The T800 Box runs quietly in your garage, listening to the CAN bus through an OBDLink MX+ adapter 24/7, and tells you when something matters.

It doesn’t send commands to your car. It doesn’t need your Tesla account. It just watches.

• Air suspension pump running more often than it should? You’ll know 30 days before a $5,000 repair.
• 12V battery voltage drifting below threshold during sleep? You’ll get an alert before you’re stranded.
• Battery cell spread widening over months? You’ll see the trend, with context.

The goal is simple: avoid surprises on expensive components.

Your car deserves a health record

Here’s the thing I kept bumping into while building this for myself: the value of vehicle telemetry is not measured in days. It’s measured in years.

A week of cell voltage data won’t tell you your battery is degrading. Three years will. A month of suspension pump intervals won’t catch a gas bag that lost 2% of its seal per month. Two years of pump frequency, compared against itself, will catch it cold — 60 days before the corner starts sagging.

But that only works if the data survives. And a garage computer full of SSDs and thermal paste is not the right place for a multi-year record. It’s fine as a buffer. It’s terrible as an archive.

That’s why T800 treats your telemetry the way a clinic treats medical history: the record follows the car, not the filing cabinet. The T800 Box in your garage keeps about 30 days of local buffer — enough to keep processing if the internet dies, enough to handle short outages. Everything else lives in your account in the T800 cloud, and follows you forever. Swap the T800 Box, lose an SSD, move cities, or sell the car — the history is yours.

The two practical reasons this matters, beyond peace of mind:

Resale. A multi-year T800 Sell Report is a real asset at sale time. Used-market studies put the premium for documented, auditable vehicle history at 5–15%. On a 2017 Model X 100D where the biggest buyer fear is “what did the previous owner do to it,” three years of daily cell voltage, suspension pressure, and charging behavior is the most credible answer you can possibly hand them. I built this system for my own car, and the sell report is the receipt I want to have in five years.

Retroactive insight (the part I’m most excited about). Our CAN parsing and monitor algorithms improve every month. Signals we can only weakly interpret today will become sharp, reliable indicators a year from now. When we ship a new monitor, we re-run it over your whole history. Your data today is not just today’s snapshot — it’s a receivable against every future algorithm we build. That’s a kind of compound interest you only get if the raw data is actually around.

If I’d had this running for the three years before my collision, I’d have the complete dataset for every failure that showed up afterward — every cell voltage, every pump cycle, every thermal sensor reading for the weeks and months leading up to each problem. Instead I have a few patchy local logs. That’s the gap this archive closes.

And the privacy piece, briefly, so I don’t bury it: the Tesla CAN bus I read does not carry GPS, destination, route, cabin audio, or camera feeds. Those signals are not on the bus. That’s why I was comfortable making cloud storage the default for T800 — there isn’t a location story to worry about, because there isn’t any location data in the pipeline to begin with. If you still prefer the 30-day local buffer, Strict Local Mode is one toggle away in your account. I documented the full trade-off in the Privacy Policy — short version: you keep strict local privacy, you lose multi-year analysis, the Sell Report, and retroactive insight. That’s a real trade-off, and it’s yours to make.

Why I’m sharing this

I built this primarily for my own car. I plan to keep my Model X for at least another 10 years.

But Tesla Model S and Model X are reaching the age where expensive failures become common — suspension, 12V, battery degradation. And with S/X production ending, the aftermarket diagnostic ecosystem is only going to get thinner.

If other long-term owners find this useful, I’d like to build it together. Real data, real problems, no hype.

About me

I’m Harrison — an engineer based in Vancouver, Canada. I’ve spent months reverse-engineering Tesla CAN bus data, building analytics pipelines, and testing everything on my own car.

This project is documented on the Tesla Longevity Lab YouTube channel, where I share findings, data, and real ownership stories.

If you have questions or want to talk about your S/X, reach out at [email protected].