My Aging Fridge
Before the cave there was a used refrigerator, and it is still the build I’d tell most people to start with.
I bought it secondhand and rebuilt it, following the New England Cheesemaking cave guide nearly exactly for the basics — if you want the canonical version of those, read theirs. What’s below is what I actually did, in the order I actually did it, including the two designs that didn’t work — because the failures are the useful part, and because the rotation mechanism at the end is the thing I’ve never seen published anywhere.
It has been running about twelve years.
Why a refrigerator needs rebuilding at all
Two problems, both covered in the numbers and both worth restating because they’re why every step below exists.
It’s 10 to 15 degrees too cold. A fridge holds 35–38 °F; a cave wants 45–55 °F. At fridge temperature the ripening cultures don’t stop — they slow down so far that the wheel dries out faster than it develops flavour.
And it is, by design, a dehumidifier. Cooling air condenses water out of it. That’s the mechanism, not a side effect, and it runs directly against the 85–95% humidity you need.
So the conversion is two corrections: make it warmer, and put the water back. The first was easy. The second took three attempts.
Temperature: a PID controller on the compressor
I used a PID controller to control the compressor, and left the fridge’s own thermostat simply running inside. The fridge can’t be set anywhere near 53 °F, so you don’t try — you leave its thermostat calling for cold and put your own controller upstream of the compressor.
This is the same approach I use on my smoker modifications, which is not a coincidence: once you’ve wired a PID to hold a temperature in one box, every other box in the house starts looking like a candidate. A PID is worth the small extra cost over a simple on/off controller because it anticipates rather than just reacting, so you get less overshoot and less compressor short-cycling.
Humidity, in three phases
This is the part where I want to be honest about the iteration, because a build article that presents only the final design teaches you much less than one that shows the dead ends.
Phase 1 — a humidifier inside the fridge
The obvious approach: put a small humidifier in there with the cheese.
It only sort of worked, and it took up a lot of space. A fridge cavity is mostly shelf, and a humidifier standing in it is volume you wanted for cheese. It also has to fight the fridge’s own condensation directly, in the same small airspace.
Phase 2 — a hose through the door
So I moved the humidifier outside. I drilled a hole about an inch in diameter in the front of the fridge, roughly midway up the door, ran a vacuum hose through it, and sealed and secured the hose where it enters with RTV. The hose runs to a small humidifier outside, controlled by a second PID controller — one for temperature, one for humidity.
Better. Also wrong, in two specific ways I hadn’t anticipated:
The hose crossed the door opening. Every single time I wanted to look at a cheese, I had to disconnect the hose. That’s the kind of small friction that makes you check on things less often, which is the opposite of what you want.
And any sag in the hose became a water trap. Warm humid air moving down a cool hose condenses, and condensate runs downhill to the lowest point and sits there. Once a low spot existed it collected water and eventually made a huge mess.
Phase 3 — rigid pipe through the side wall
The fix was to move the entry from the door to the side, and to use rigid pipe instead of hose so it cannot sag.
🔴 And here is the part to be careful about. I very carefully found a spot on the side where I could drill a one-inch hole without running into any of the inner tubes. The walls and skin of a refrigerator carry refrigerant lines, and drilling into one ends the project immediately — you release the charge, and depending on the unit’s age you may be venting something you shouldn’t. Find the lines first. Do not guess.
Into that hole went one-inch rigid PVC, sealed and secured with super glue and RTV, connected to the top of the humidifier.
That single change fixed both phase-2 problems at once: the door is now completely clear, and rigid pipe has no low spot for condensate to collect in. This is the design that has been running since.
⭐ The rotation
Aging cheese has to be turned. It distributes moisture evenly through the paste, and it encourages uniform microbial and mould growth across the rind rather than a wet face on the bottom and a dry one on top.
The obvious way to turn cheese is to open the door and turn the cheese — and every door opening on a small humid box dumps the conditions you spent all day holding. Worse, it ties the turning schedule to the door schedule, when the guide points out that opening the door once or twice a day is actually desirable for air exchange. Those are two different rhythms and manual turning forces you to serve the wrong one.
So I made the cheeses rotate themselves, with the door shut. Twelve years on, it works.
The frame
Two vertical rails of 80/20 half-inch extruded aluminium, bolted to the outside left and right of the fridge. The rails are spaced so the motors bolt straight on without fabrication.
The drive
BBQ grill motors — the kind that turn a rotisserie — bolted to the right-hand rails. They’re 110 VAC, cheap, widely available, already geared down to a slow rotation, and designed to turn an unbalanced load all day. For this job they’re close to ideal, and I doubt I’d design a better actuator from scratch.
On the left side, where the humidifier is, I made simple brackets with a hole for the far end of each rod to rest in.
Getting the shafts through the wall
I drilled holes on either side for bearings, and the BBQ rods pass through those.
The rods are square, not round — which is exactly what you want for driving a load without slipping, and exactly what you don’t want for sealing a rotating shaft. There’s inevitably a gap. My solution was to glue styrofoam around them, and I’ll be straight that this is the least elegant part of the build: it is not a proper seal. In practice I have no leaks to speak of, helped considerably by the fact that the whole thing lives in a cool basement, so the pressure and temperature differentials across that gap are small.
The shelves
Three rods, 18 inches apart, the first 18 inches down from the top (inside dimensions).
The cheese sits in wire baskets that attach to the rods with thumb screws — two baskets side by side per rod, each about 12 inches square. Thumb screws matter more than they sound: baskets come out for cleaning and go back without tools, and anything requiring a tool eventually stops being done.
The control
Each rod carries a glued-on magnet, read by a hall effect sensor on a bracket mounted to one of the 80/20 rails near the motor. So the controller knows the rod’s position rather than just running the motor for a guessed length of time.
The controller is an Adafruit Metro Mini 328 with an OLED FeatherWing, driving each motor through a relay (the motors being mains voltage). The rods are numbered 1, 2 and 3 top to bottom, and the FeatherWing’s buttons let me pick a rod and set how often it turns — once a day, once a week, whatever that shelf’s contents want. Which is the real win: a bloomy on rod 1 and a hard wheel on rod 3 don’t need the same schedule, and now they don’t have to share one.
The whole controller lives in a 3D-printed box mounted above the topmost motor on the front-facing 80/20 rail, so I can read the display and reach the buttons without opening anything.
The two things I only found out by running it
Both of these are the marks of a real build rather than a design, and they’re the details I’d most want somebody copying this to know.
1. The motors don’t stop instantly. A geared AC motor coasts. So the hall effect positions needed adjusting, and I had to add a dwell time into the Metro’s logic to account for the overrun — cut power slightly before the target and let it coast in.
2. An unbalanced cheese keeps rotating after the motor stops. This is the one I didn’t see coming. If the load on a rod isn’t balanced, gravity finishes the job the motor started, and the basket creeps round after power is cut.
The fix is a brake: a heavy-duty spring that clamps onto the square rod, plus a solenoid, also controlled by the Metro, that releases the brake just before the motor starts and lets it re-engage when the motor stops. It took some tweaking to get the timing right. It works well.
I like that the square rod, which was the problem for sealing, is what makes a simple spring clamp work as a brake. You get to use the flat faces for something.
What’s next: an 8×8 walk-in
The fridge is not the end of the line. Beyond it there’s already the basement cave — the 4×10 room with cement on three sides and a salvaged wooden cooler door — and beyond that there’s a proper 8×8 walk-in cooler I bought off Facebook Marketplace.
It is currently in pieces. I need to assemble it and confirm the compressor actually works, which is the risk you accept buying refrigeration secondhand. I’m hoping to have it together and running by next summer.
At which point I will have gone fridge → found room → real cooler, which is more or less the ladder on the building-a-cave page climbed one rung at a time over thirty years.
Why I still think the fridge was the right first move
It’s cheap, it’s reversible, it fits in a corner, and — the part that matters — a small box is a harsh teacher. Forty square feet of basement with concrete walls forgives your mistakes because the thermal mass rides them out. A fridge-sized cavity responds to everything immediately, so you find out within a day when your humidity control is wrong, and you learn what each variable actually does because you can watch it move.
And constraint produced the only original thing here. The motors exist because the box was small enough that door-opening genuinely mattered. In the cave, the problem that prompted them is much milder. If I’d built the room first, I’d never have thought of it.
Back to the numbers, or on to the cave that came after this.
Sources
- The base conversion — controller, humidifier, target ranges, shelving and air exchange — follows New England Cheesemaking Supply Company’s “How to Make a Cheese Cave”: https://cheesemaking.com/blogs/learn/how-to-make-a-cheese-cave
- Everything from “Humidity, in three phases” onward is my own build, including the rotation mechanism, which is not from that guide and which I have not found published elsewhere.
Comments (0)