The $28,000 Separator That Cost Me $94,000 (And What I Check Now)
The cheapest magnetic separator is rarely the one with the lowest sticker price. My $28,000 "deal" in 2019 ended up costing $94,000 over three years—in rebuilds, unplanned downtime, and lost throughput.
Here's the pattern I missed, and the checklist I use now.
I've been buying mineral processing equipment for nine years. Not as a manufacturer rep—I'm on the operator side, specifying for recycling and recovery facilities. I've approved POs for 14 magnetic separation units. Seven of those decisions had problems. Some minor. Some not.
Let me be specific about the worst one.
The $28,000 separator that cost me $94,000
In March 2019, I signed off on a high-intensity magnetic separator for our e-waste line. On paper, it looked superior to everything else we'd quoted. Higher gauss rating. Bigger throughput number. 30% cheaper than the next best option.
It arrived on time. Installation went smooth. Then we ran actual material through it.
The unit was rated for 12 tons per hour. We were feeding it 8. It clogged twice a week. Reason: our material stream had moisture content that fluctuated between 3% and 9%. The separator worked fine at 3%. At 9%, it was a $28,000 paperweight.
That moisture issue showed up in our feed data. I just didn't look at it. I was comparing gauss ratings and price sheets—the numbers that looked impressive in a quote, not the ones that mattered on the floor.
The lesson: separator specs mean nothing until you match them to your actual feed variability.
What "best" actually means
Every vendor claims their equipment is the best. The Steinert rep said it. Every other rep said it too.
"Best" only means something when you define the variable. Best recovery rate at what feed size? Best throughput at what moisture level? Best uptime at what maintenance budget?
I now ask every vendor: "Show me a test result that matches my material." If they can't, I walk. If they can, I want to see the actual report—not the summary slide.
That's why I started requiring material tests before buying. Steinert has a practice center for exactly this—you send them a sample, they run it through different configurations, and you get actual numbers instead of a spec sheet. The first time I did it, in 2021, I found that the configuration I would have ordered (based on the spec sheet) recovered 8% less aluminum than the alternative. That's $1,900 per month in recovered material I would have lost.
Spent $600 on a material test. Saved $1,900 per month. Not a complicated ROI calculation.
The checklist I use now (before I sign anything)
After that $94,000 mistake—yes, I added up the rebuilds, the downtime, the cost of sending material to landfill because we couldn't process it—I built a pre-purchase checklist. It's not fancy. It's just the questions I should have asked in 2019.
- Feed variability, not feed average. What's the standard deviation on moisture, particle size, and iron content? Not the average—the range. Equipment that works at 3% moisture may fail at 9%.
- Throughput at what condition? "12 tons per hour" means nothing without the test conditions. Ask for the derating curve: how does throughput drop as moisture rises?
- Maintenance window compatibility. A separator that needs a 4-hour rebuild every month doesn't fit a plant that only shuts down quarterly. I learned this one the hard way in 2022.
- Test with your material. Not a sample from a trade show. Not a "representative" sample your vendor used. Yours. With all its imperfections.
The one thing I got right
So glad I started requiring material tests before purchase. Almost skipped it in 2023 on a small order—just 3 units, $18,000 total. The vendor said testing would add two weeks to the lead time. I pushed back and said we'd wait.
Dodged a bullet. The test showed the units would have worked, but needed a different screen configuration. Without the test, we would have received equipment that ran at 70% of rated throughput. The vendor would have blamed our material. We would have blamed the vendor. Nobody would have been happy.
The test cost $800. The first six months of reduced throughput would have cost us roughly $11,000 in lost recovery. Simple math.
What I still don't know
I'm not a metallurgist, so I can't speak to the physics of magnetic field gradients or why certain rare-earth configurations outperform others on fine particles. That's beyond my lane.
What I can tell you is how to evaluate a quote without a PhD in material science: ask for test data, define your variables, and don't trust a spec sheet that wasn't generated with your material.
Honestly, I'm still not sure why some vendors consistently provide honest derating curves while others consistently hand-wave. My best guess is it comes down to whether their sales team is paid on units shipped or lifetime customer value. But that's speculation.
One caveat
This approach works for operations processing variable feed streams—e-waste, mixed recyclables, construction debris. If you're running a homogeneous ore body with consistent mineralogy and stable moisture, you can probably trust the spec sheet more than I can.
But for anyone else: the best separator is the one that performs on your worst feed day, not your best.