Tue, Nov 3, 2015

Cider Cave Cooling

Building an Internet-enabled walk-in cooler by hacking air conditioners.

In my last post I described how we built a cider cave — a large insulated box to house our fermentation tanks. To make excellent cider it is crucial to have the right temperatures while the cider ferments and rests. This post is about how we achieved those ideal temperatures despite the extreme Okanagan summertime heat.

We started with a couple of window-mounting air conditioners salvaged from Mike’s parents’ basement. Perfect! They’re each about 6000 BTU and draw maybe 4 amps @ 120VAC. Units like this are meant for cooling your bedroom though so they only go down to about 17°C. We want this cave to be 10°C or even less depending on the stage of production, so they needed some modification hacking.

Refrigeration primer

Fridges, air conditioners, heat pumps, and even large industrial refrigeration systems all work on the same principal. A compressor pump forces a refrigerant gas around a closed loop of (usually copper) pipe. At one stage the gas is compressed and gives off heat, and at another stage it is allowed to expand and so it absorbs heat. This is known as the refrigeration cycle and you can usually spot it by looking for the big thermal fins and fans that help it efficiently absorb and dissipate heat.

Ultimately it’s the electric compressor doing the business. A thermostat switches it on when it wants things colder, and off when things are cold enough. But AC units like these have thermostats that refuse to go below 17°C. They figure that ought to be cold enough for anyone’s home, and it provides a conservative margin to avoid things getting so cold that they seize up with ice.

We wanted to make them go much colder, so we replaced the thermostat. I built one based on an electronic microcontroller, but you could also use a Ranco or even a mechanical thermostat. I chose the microcontroller because it can do a bunch of extra tricks.

A new temperature controller

Our particular AC units had two circuit boards inside. The relay board does the business of switching mains voltage to the compressor and fans, and the other one has the little LCD screen, buttons, and a processor of some kind. The two were connected by a ribbon cable where each pin was a 5v signal corresponding to one of the relays. After ten minutes poking around with a multimeter, I had mapped out what each pin was for.

The relay board didn’t need to be modified, which was nice because that way I didn’t have to worry about switching high-current mains voltage directly. The other circuit board, however, was of no use. I replaced it with an ESP8266 microcontroller with some additional circuitry to generate the 5v signal levels required by the relay board. The ESP8266 has wifi and plenty of programmable pins to use for other purposes. This controller is equipped with a 1-Wire bus that can accommodate lots of DS18B20 digital temperature sensors just by chaining them up in parallel.

Avoiding frosting

With the modified thermostat controller in place, it’s possible to keep the compressor running until our sensors get to a much lower setpoint. In fact you can just target it to lower and lower temperatures until the compressor is basically running all the time. The problem with doing this is that the metal fins will eventually frost up from the ambient humidity and impede air flow. The whole refrigerant cycle will become ineffective and the compressor would probably just burn out and stop. This is why the original thermostat was limited to 17°C.

To prevent this from happening, I simply added another temperature sensor directly on the fins, and modified the controller software to ensure the fins never stay below freezing for too long. This limits how cold it’ll go, but ensures it won’t get colder than it can actually sustain.

There’s a commercial product called CoolBot that does this too, and also a free open source version called ColdSnap.

Testing it out

When we first got this rigged up, we had a test batch of a couple hundred liters that we wanted to carbonate in kegs. Normally we’d use a fridge or deep freeze for this, but it was the perfect opportunity to see how cold our cave could go. We got it to around 3°C which was just sufficient for carbonation. We won’t be using this for carbonating production batches, but it was neat to see how low this thing goes.


I couldn’t have done this without carefully monitoring the temperatures throughout the system and making tons of adjustments. By using a programmable microcontroller with wifi, I was able to send high-resolution measurements over the Internet to a server where I could store and analyze them. Overkill? Yeah probably. I’ll describe that in another post.

Posted on Tue, Nov 3, 2015 by Luke Cyca