The pump impeller I filed by hand to silence a midnight rattle in a 180-gallon sump

The pump impeller I filed by hand to silence a midnight rattle in a 180-gallon sump

I’d been running the new system for about eleven days when the rattle started. Not a hum, not a vibration you could feel through the floor—a rattle, sharp and metallic, like a loose screw being shaken inside a tin can. It started around 2:47 AM on a Wednesday. I know the time because I remember staring at the microwave clock while standing in my kitchen in boxer shorts, holding a flashlight, trying to pinpoint where the sound was coming from.

The tank is a 180-gallon custom build I’d pieced together over six months. It lives in what used to be my home office—I moved my desk into the living room to make space. The sump sits in a plywood cabinet I built myself in the garage over a long weekend, and it’s not pretty. The cuts are slightly off on the left side panel, and I never bothered to paint the inside. It works, but it looks like something a high school woodshop student turned in for a B-minus.

The rattle was coming from the sump. Specifically, from the return pump—a Reef Octopus Varios-8 that I’d bought used off a guy named Dave from a forum. I paid $180 for it, which was about $80 less than retail, and Dave had assured me it was “barely used.” He’d run it for three months on a 120-gallon before upgrading to a closed-loop system. I believed him because the pump looked clean when I unboxed it, and because I wanted to believe him.

Standing there at 2:47 AM, I was starting to doubt Dave.

What midnight sounds do to a person

There’s something about a repetitive noise at night that bypasses rational thought. During the day, a rattle like that would have been an annoyance—something to fix after work, maybe over the weekend. At 2:47 AM, it was a crisis. I couldn’t sleep knowing the pump was making that sound. I’d lie in bed, listening, waiting for the next rattle to confirm the pump was still rattling. And it always did. Every few seconds.

I tried the obvious things first. I turned off the pump and pulled it from the sump. I checked the mounting bracket—tight. I checked the volute for debris—clean. I reassembled it and put it back in the water. The rattle came back within two minutes.

I did this four times. At some point around 3:30, I realized I was being an idiot. I was doing the same thing over and over and expecting a different result, which is the definition of something I should have learned not to do by age thirty-seven.

I sat on the floor of my office, back against drywall that still smelled faintly of new construction, and watched the pump run. The sump water was moving fine. The display tank looked fine. But the rattle was there, persistent, mocking.

I texted Dave at 3:45 AM. I don’t know why. I didn’t expect a reply. He wrote back at 7:14 AM with a single sentence: “Check the impeller shaft, mine had a burr.”

That was all. No apology, no offer to take the pump back. Just a sentence that would end up costing me two hours and saving me $180.

What a burr actually looks like under a desk lamp

I didn’t get to the pump until Saturday morning. Work happened. Life happened. But the rattle was still there every night, and by Friday my wife had started sleeping on the couch. She was kind about it—said she didn’t mind, that she’d been meaning to rewatch something on Netflix anyway—but I knew the sound was getting to her too.

Saturday morning, I pulled the pump again. I emptied the sump down to the baffle level, disconnected the plumbing, and carried the pump to my workbench in the garage. The temperature was about 45 degrees that morning, and I remember thinking I should have worn a jacket, but I was too focused on the pump to go back inside.

I took the volute off and pulled the impeller assembly out. It was a magnetic rotor with a ceramic shaft, held in place by a plastic bushing on one end and a rubber grommet on the other. I held it up to the light—one of those cheap clip-on desk lamps I’d bought at a hardware store for twelve dollars—and turned it slowly.

There it was. A tiny ridge of metal on the ceramic shaft, maybe a millimeter long, near the tip. Not a crack, not a chip. A burr. Probably from when the pump was assembled at the factory, or maybe from when Dave had dropped it while cleaning it. Ceramic is hard, but it’s brittle. A small impact can cause a raised edge without breaking the shaft.

That burr was catching on the bushing every time the impeller rotated. The rattle was the sound of something catching and releasing, catching and releasing, hundreds of times per minute.

I stared at it for a while. I considered ordering a replacement shaft. I checked online—$32 plus shipping. Three to five business days. Another weekend of sleeping on the couch.

That’s when I got the idea to file it.

The wrong tool for the right job

I didn’t have a proper ceramic file. I didn’t even know if such a thing existed. What I had was a set of four jeweler’s files I’d bought years ago for a project I never finished—the kind you get at a hobby store in a plastic tube for about eight dollars. The smallest one was a flat file, maybe three millimeters wide, with a grit that felt rough against my thumbnail.

I also had a magnifying headset—the kind with the flip-down lenses that make you look like a mad scientist. I’d bought it for soldering circuit boards and never really used it. Now it was sitting on the workbench, gathering dust, and I was grateful for its existence.

I put the headset on and clipped the impeller shaft into a small vise I’d bolted to the workbench years ago. The vise was overkill for the job, but it held the shaft steady without damaging it. I took a deep breath and touched the file to the burr.

The first stroke didn’t do anything noticeable. The file just skated over the ceramic. I adjusted my angle, put a little more pressure, and tried again. This time I felt it—the file was actually removing material. A tiny amount, barely visible, but it was working.

I filed for about ten minutes. Not continuously—I’d take a few strokes, then examine the shaft under the lamp, then take a few more strokes. The burr was slowly disappearing. The surface of the shaft was getting smoother. I wasn’t trying to reshape anything, just remove the raised edge that was causing the rattle.

After ten minutes, I couldn’t see the burr anymore. I ran my fingernail over the spot where it had been and felt nothing. I wiped the shaft clean with a microfiber cloth—the same kind I use for the display glass—and reassembled the pump.

I carried it back to the sump, connected the plumbing, filled the sump back to operating level, and plugged it in.

Silence. No rattle. Just the gentle hum of water moving through the system and the trickle of the return pipe into the display tank.

I stood there for a full minute, listening. Then I went inside and told my wife the couch was available again.

What I wish someone had told me before I bought a used pump

I’ve been keeping fish for twelve years now, and I still make stupid mistakes. Buying the pump used wasn’t a stupid mistake in itself—saving $80 on a piece of equipment that retails for $260 is a legitimate move. The mistake was not testing the pump before I installed it.

I was in a hurry. The tank had been cycling for three weeks, and I was eager to get the fish in. I’d ordered a batch of six ocellaris clowns from a local breeder who was retiring his operation, and they were scheduled to arrive in two days. I didn’t want to tell him to hold them while I waited for a new pump. So I took Dave at his word, installed the pump, and crossed my fingers.

I should have run the pump in a bucket of water in the garage for a day. I should have listened to it, checked its temperature, measured its flow rate. I should have treated it like a used car—test drive first, buy later.

Instead, I learned about the burr at 2:47 AM, standing in my underwear in a room that still smelled like silicone.

If you’re reading this and you’re about to buy a used pump, here’s what I’d tell you: run it in a bucket for at least 24 hours before you put it in your sump. Fill a five-gallon bucket with tap water—dechlorinated if you want to be extra careful, but for a test run, plain water is fine. Drop the pump in, plug it in, and let it run overnight. Put your ear next to it. Feel the housing for heat. Check if it’s pulling an air bubble or making any noise that doesn’t sound like moving water.

It would have taken me an hour. It would have saved me a week of bad sleep and a conversation with my wife about whether the pump was worth the money we’d spent on it.

The cost of being cheap

I’ve done the math on the pump. The $180 I paid Dave was a good price for a working Varios-8. But the pump wasn’t working when I got it—or at least, it wasn’t working quietly, which for a sump pump in a living space is a critical part of working. A pump that makes a noise you can’t ignore is a pump that needs to be replaced, even if it’s moving water.

I got lucky. Filing the impeller shaft worked, and the pump has been running silently for six months now. But I could have easily ruined the shaft. Ceramic is hard, but it’s also brittle. Too much pressure, the wrong angle, a moment of impatience—and I’d have a cracked shaft and a $180 paperweight.

Then I would have had to buy a new pump anyway. I would have been out the $180 I paid Dave, plus the $260 for a new pump. That’s $440 for the privilege of learning a lesson I could have learned for free by running the pump in a bucket.

The cost of being cheap isn’t always measured in dollars. Sometimes it’s measured in hours of sleep, or in the irritation of a spouse, or in the feeling of standing in a cold garage at 8:30 AM with a jeweler’s file and a prayer, wondering if you’re about to turn a fixable problem into an unfixable one.

What I’d do differently

If I could go back to the day I bought that pump, I’d do three things differently.

First, I’d test the pump before I installed it. I’ve said that already, but it bears repeating because it’s the single easiest way to avoid the whole situation. A bucket test costs nothing and takes a day. It reveals burrs, unbalanced impellers, loose magnets, and a dozen other problems that are easy to fix on a workbench but hard to fix in a sump full of water and live rock.

Second, I’d ask Dave if he’d ever had the impeller out of the pump. Most people who maintain their equipment have pulled the impeller at some point to clean it. That’s normal. But every time you pull a ceramic shaft, there’s a chance of introducing a burr—from a dropped shaft, a dirty work surface, a careless reinsertion. If Dave had said “yeah, I pulled it once,” I would have inspected the shaft before I ever put the pump in the sump.

Third, I’d buy a set of diamond files instead of the cheap hobby files I used. The jeweler’s files worked for this job, but they’re not ideal for ceramic. Diamond files are designed for hard materials—they cut cleaner, leave a smoother surface, and are less likely to cause micro-fractures. A good set of diamond files costs about $20 online. That’s $20 I should have spent before I touched a $180 pump with a $8 file.

I bought a set after the fact. They’re in my tool drawer now, next to the headset, waiting for the next time something goes wrong.

The pump, six months later

The rattle is gone now. The pump runs so quietly that I sometimes have to put my hand on the sump to confirm it’s still operating. The clowns I got from the retiring breeder are doing well—they’ve been in the tank for five months, and they’ve started pairing up. The smaller one has developed a habit of swimming into the return flow and letting the current push her across the tank, like a kid on a waterslide.

I think about Dave sometimes. Not in an angry way—he didn’t know about the burr, or at least I don’t think he did. I think about the randomness of it all. A million-to-one manufacturing defect, a drop on a concrete floor, an impeller that was just slightly out of spec—any of those could have caused the burr. Dave sold me a pump that had a problem he didn’t know about, and I figured out a way to fix it without spending more money.

The pump is still running. The files are in the drawer. The headset is on the workbench. And somewhere out there, probably, someone else is standing in their underwear at 2:47 AM, holding a flashlight, wondering what that noise is.

I hope they find this article before they file anything.

📷 Photos: Dheeraj Udatha (Unsplash), Eugene Chystiakov (Unsplash)