When the Robot Says No: A Field Guide to the Metso IC70C and What Happens When Automation Bites Back

I got the call on a Tuesday. 2:47 PM, to be exact. The voice on the other end was tense, the kind of tense you get when a machine that costs more than a house has decided to throw a tantrum. The issue: a Metso HP cone crusher, the workhorse of the quarry, had gone into a full stop. The IC70C, the brain of the operation, was flashing a fault code I’d never seen before. The operator, a guy named Dave with twenty years in the pit, was ready to take a sledgehammer to the touchscreen.

“It just shut down,” he said. “No warning. One minute it’s chewing granite, the next it’s dead. Says something about the ‘tramp release system’.”

In my role coordinating emergency service for heavy equipment, I’ve handled maybe 40+ rush orders in the last three years. This felt different. This wasn’t a missing part or a scheduling snag. This was a system-level brain freeze. And the quarry had a 6:00 AM deadline the next day for a 5,000-ton order.

The Setup: The Promise of Automation

The Metso IC70C is a good system. I’ll say that upfront. It’s designed to be the ultimate operator—no coffee breaks, no fatigue, no bad judgment calls. It monitors the crusher’s power draw, the chamber level, the setting, and—critically—the hydraulic pressure. It’s supposed to automate the “tramp release” cycle: when an uncrushable object (a piece of steel, a shovel tooth) enters the chamber, the system should instantly drop the main shaft, let the object pass, and then return to the original setting.

Here’s the thing: automation is a promise. And promises, especially in the mining world, have fine print. The fine print is that the system is only as good as its sensors. And sensors, in a cloud of rock dust and vibration, can be… temperamental.

The Turning Point: A False Positive

I arrived on site about three hours later. The crusher was still down. Dave was standing ten feet away from the control panel, arms crossed, like it was a wild animal. The IC70C’s display was a memorial screen, listing the last fault: “Tramp Release Induced Shutdown – Critical Pressure Anomaly.”

I ran through the standard diagnostics. Did an object actually pass? No visible damage. Chamber clear? Yes. Hydraulic oil level? Good. Then I checked the pressure sensor log—and this is where it gets interesting. The system had recorded a pressure spike that lasted for 0.8 seconds. Enough to trigger the release sequence. But the waveform didn’t look right. It was too perfect, too clean.

“I think it was a ghost,” I told Dave.

“Come again?”

“A ghost signal. A spike from a loose wire, or a sensor that’s about to fail.” I pulled up the data. “Look. A real tramp event shows a chaotic spike and then a rapid drop as the system reacts. This is sharp and clean, like a computer simulation. The system saw a problem that wasn’t there, and did exactly what it was programmed to do. It protected the crusher from a threat that existed only inside its own wiring.”

This is the dirty secret of advanced automation. The IC70C is very, very good at following rules. But it can’t tell the difference between a real event and a sensor glitch. It’s like a guard dog that barks at every shadow, but you can’t tell it the shadow is just a cat across the street.

The Fix: The Human Override

We bypassed the offending sensor, swapped in a spare we had in the truck, and recalibrated the tramp release settings. The whole process took about 45 minutes. Dave re-started the crusher, and it purred back to life. The 6:00 AM deadline was saved. But the real work came after.

I sat with Dave and the shift supervisor and walked through what had happened. (Honestly, I still don’t know why that sensor failed. It was a standard Bosch-Rexroth unit, less than 6 months old. My best guess is a micro-fracture in the housing from vibration. If anyone at Metso has insight, I’d love to hear it.)

We created a new protocol. Once per shift, the operator was to manually check the pressure sensor’s live reading against a mechanical pressure gauge. A 5-second check to validate the brain’s perception of reality. As I told them: “Five minutes of verification beats five days of correction.”

The Reckoning: What I Learned About the Metso IC70C

This incident happened back in March 2024. Since then, I’ve seen this pattern repeat at three other sites. Here’s what I know now:

• The IC70C is excellent for optimization, but its fault responses are rigid. It assumes any anomaly is a real event. This conservatism saves parts but costs hours when a sensor lies.
• The tramp release system is a mechanical miracle, but the sensing is its weak spot. The hydraulic circuit itself is robust. The pressure sensors, located on the accumulator line, are exposed to high-frequency vibration that can cause intermittent failures.
• Don’t trust “100% uptime.” The industry standard for automated systems is often described as “improved reliability.” That doesn’t mean perfect. A false positive shutdown is still a shutdown.

Look, I'm not saying the IC70C is a bad system. It's basically a necessity for modern crushing operations. It manages power draw in a way no human can, adjusting the feed rate in real-time to maximize throughput. The problem is, we treat it like a magic box. We trust the display without questioning the data. Real talk: The lowest total cost of ownership for an HP cone crusher includes the cost of training your operators to not just trust the screen, but to understand the physics the screen is trying to represent.

Think of it this way: The IC70C is a world-class chef. It can make a perfect meal every time. But if the ingredients (sensor data) are bad, even the best chef cooks garbage. You still need someone to taste the soup.