DIY Aquarium Chiller

From modified dehumidifier or air conditioner

The most common thing to hear about DIY chillers is that it's not worth the hassle, will not be cheaper than buying a commercial chiller, and the simple home made solutions don't work. As someone who has been studying the issue for many years, and building my own chillers, I absolutely agree that for most people that is true. If you want a cheap chiller, buy a small, China-made, used one. There are some people though, who can get parts for free and already have the necessary tools and expertise. For those it can certainly pay off, particularly if they need large chillers. Personally, I make chillers because I can't buy the type of chillers I need for my tanks. Also, I totally enjoy building! So my personal motivation is not saving money on this. Building a refrigeration chiller is challenging, and refrigeration technology is difficult to understand. This article can't take you from being totally green on the subject to being ready for building your own chiller. There are so many more things that would need a lot more explaining. But it will describe the whole process step by step using a real build as an example, and show you everything you need in terms of tools and parts. So hopefully it will get you started, or able to judge if this is something you want to embark on or not.

Safety warning

I don't write a lot about safety in this article. There are two reasons for that: First, it would make the article much longer. There is an amazingly large number of ways you can potentially hurt yourself when modifying a refrigeration system. This is not a complete guide, you have to study each separate subject more before being ready to build. Second, building a chiller is not for beginners. It is for people with both practical experience and knownledge of relevant physics. The safety issues should be obvious to such people.

How a chiller works

Again, it is not in the scope of this article to explain all the physics of refrigeration. So I will only do a short summary. I really recommend that you read more about the details of chillers if you plan on building one. But enough about that, let's jump right into it: If you've ever looked at the inside of a chiller you must have noticed a bunch og ugly looking copper tubes, a fan and radiator, and a black painted, round, heavy, steel thing that makes a clunky noise every time you move the chiller around. That last thing is the compressor. An important thing to know is that all the tubes in there are connected together in a cycle. A fluid called the refrigerant is pumped around in there by the compressor. The first half of the tube cycle is called the high side. Here the pressure and temperature are high. The refrigerant condenses from gas to liquid here. It then passes, in liquid form, through the long, thin, curled up, copper tube called "the capillary tube". The capillary tube's function is to hold back the liquid so that the pressure on the high side stays high. Some larger chillers use a thermostatic expansion valve instead of a capillary tube. The official name in both cases is "metering device". On the other side it enters the low side where pressure and temperature are very low. Here it evaporates into gas. Now the point of all this is of course to get the cold section in contact with aquarium water and the hot section in contact with air so that heat is transferred from the water to the air. The part that is in contact with flowing air is called the condenser, and the part in contact with water is called the evaporator.

Dehumidifiers

A certain type of dehumidifiers, and some air conditioners, are amazingly similar to aquarium chillers. The casings they come in are virtually identical. The compressors are identical. The condensers, with fans, are identical. Their metering devices are even tuned to work with the same temperatures as aquarium chillers for tropical aquariums. The way a dehumidifier works is that a flow of air is first blown over the evaporator (or through it, you could say), which is cold, and then over the condenser, which is hot. That's it really. So how does that dehumidify the air? Well, since the evaporator is cold, water vapor in the air condenses on it. The condensed water then drips down into a tray and flows on into a container. The reason the condenser is in the air flow is that the heat from the air, water and compressor must be removed from the dehumidifier so that it doesn't overheat. What needs to be changed for it to become an aquarium chiller is the evaporator and the control logic. The last part is actually much simpler than it sounds. You don't really change it, you just rip it out and instead run the chiller on an external aquarium temperature controller. The evaporator is more complicated. A completely new one has to installed. Then the system has to be evacuated and charged with new refrigerant.

Stripping down the dehumidifier

The easiest to modify dehumidifiers I have worked with were made by Woods in Canada. They seemed to be of a very old design. They were almost completely made from metal, which I like because of the fire safety. Even though the performance was low the size was enormous, and there was a lot of room inside. All the parts gave the impression of being very overbuilt compared to the modern, small China dehumidifiers. The electronic parts were down to a minimum without any circuit boards. So all wiring was simple. The main problem with these is the price, they cost up to 4 times more than you can get a similarly specced China dehumidifier for. The clunkyness is also an issue. The one I use as the main example here is a modern, very cheap, China device.

Electrical work

The internal parts of my dehumidifier were a bit intimidating I must say, particularly the electrical system. But I must also say that it gave every impression of beeing a cleverly designed system. Also, there were good markings and wiring diagrams for the electronic parts. My previous cooling devices had only 3 cables coming out of the compressor and 4 from the fan. One is ground, and for the compressor, the two others are live, so the wires map directly to the 3 on the mains line. It is almost the same with the fan, but there are 4 wires. One of them you don't connect. You choose that one from a group of two, and depending on which one you choose you get high or low fan speed. This dehumidifier had 4 wires from the compressor and 6 from the fan. If it wasn't for the wiring diagrams I don't think I would have been able to wire the fan and compressor correctly. Both were connected to capacitors in addition to the mains power line. I don't even know what a capacitor is, so that came as a surprise. A few days of googling and reading later I felt safe that the capacitors were so called "run capacitors", and that I had understood the diagrams and wired them correctly. There was also a fuse on one of the circuit boards, so I bought fuse holders and installed a fuse on the mains line. The wiring can be even more complicated for some compressors. There can be a start capacitor and a relay in addition to the run capacitor.

The evaporator

The evaporator is where heat from the aquarium water is taken up into the refrigerant and the refrigerant evaporates in the process. It is crucial that there is only a thin heat conductive material separating the refrigerant from the water. So the industry standard for aquarium chillers is to use titanium tube. It is possible to buy thin titanium tubes, both straight and curled, online. I decided to go with a cheaper alternative called 316L stainless steel. It is designed to be saltwater resistant. I have done some tests, and so far it seems to be acceptable as long as I don't get any cut, scratches or heated surfaces in contact with seawater. So care must be taken when working with it. I hope to get away with using this material this time, but in the future I will probably use titanium. I invested in some high quality professional tools to work with the stainless steel tubing. Copper tube can be cut and bent with very cheap tools that are sold in all hardware stores, but steel needs bigger dimensions. I assume that is the case with titanium too. I had no idea how big the heat exchangers needed to be before I started making and testing them. For the 200 Watts (engine power) compressor I decided to use 130cm of 3/8 inch (9.5mm) diameter tube. For the 420 Watts compressor I went with 170 cm. They seem to work just fine. Evaporators don't need to be optimal in size, just large enough. I made prototypes from soft cheap copper tube before bending the more expensive steel tube.

Brazing

The high side pressures of a refrigeration system can easily reach 15 bar. The tubing should be pressure tested to 25 bar, I've heard. So we are talking real pressures here. Professionals build systems that are to run leak free for decades. Mechanical joints with or without gaskets are very prone to leaking, and avoided. You will not find a refrigerator or dehumidifier with a flare joint or service valve. The method of choice for joining pipes is brazing. In practice, you need to learn how to braze to build your own chiller, even though it may be theoretically possible to build one without. So, what is brazing? It is something in the middle between soldering and flame welding. Like soldering, the material used to join two pieces is different from, and has a lower melting point than the pieces themselves. This makes it easier than welding and suited for pipe work. But while ordinary soldering tin has a melting point of about 200 C, brazing rods typically have a melting point of 650 C. The copper, steel, or titanium must be heated to between 800 C and 1000 C when brazing. For this, flame welding equipment is normally used. I won't dig much further into the art of brazing in this article. There are so many articles about it, discussion boards, youtube videos etc. But I will tell exactly what equipment you need. The good news is that you don't need flame welding equipment like an acetylene-oxygen torch with a 3400 C flame, even though that is what professionals use, and the best. Then again, you can't really use the cheapest type of crème brûlée kitchen propane torch with a flame temperature of 1300C either. "Why not?", one may ask. Shouldn't it be possible to heat something to 900 C with a 1300 C flame? Well, you see, copper is a very heat conductive material. Heat will run away from the point you are heating and into the rest of the tubing system. Air flow also steals heat. The heating of air around the pipes create a chilling wind. So the longer it takes to heat, the more heat you use. This is an evil cycle. Heating is not good for metal, as it gets oxidized. So you want to heat an area that is as small as possible, as fast as possible. I first tried a high quality propane-butane torch that was to hold 1950 C. But it couldn't spit out the effect needed. Then I tried something called a MAP/Pro torch. It burns mostly propylene. The temperature isn't that much higher than the propane torch, 2400 C, but the bottle pressure is much higher. So the flame is much more powerful. It gets the job done and costs much less than even the cheapest acetylene-oxygen hobby kit. When it comes to the brazing rods, I use the cheap phosphorous containing ones for copper. They are excellent. For joining to stainless and titanium I use flux coated silver rods.

Purging

When brazing pipes it is necessary to purge the oxygen from the inside. Metal will oxidize in the extreme when heated. Copper gets a thick layer of black ash after beeing heated. On the inside of the tubing the ash pollutes the refrigerant and can clog the capillary tube. So it must be avoided. To get rid of the oxygen, a gentle flow of inert gas should flow through the tubing while it is brazed. The point is that air must not get in. Nitrogen, argon or CO2 are cheap gasses that can be used for this. I use CO2 because I happen to have a bottle of that for aquariums. The reduction valve and hoses for aquariums are exactly the same type that are used for welding. If there is more than one opening in the piping when brazing, care must be taken to avoid backdrafts of air into it. Gasses expand and flow fast when heated.

Modifying the refrigeration circuit

The refrigeration circuit must be cut in at least 2 places: Before and after the evaporator. But in practice I found it necessary to cut in 3 places because I didn't want to replace the capillary tube, and it is necessary to install a service valve on the high side. A good place to cut and install the service valve is at the port at the refrigerant filter just before the capillary tube (though I couldn't do that in the example build). The second cut can be done at the cold end of the capillary tube. Here I attach my evaporator setup. Finally, I cut and install a T junction with a service valve just before the return to the compressor. Here I attach the exit from my evaporator. In my first test chiller project I simply just brazed the evaporator assembly, with rigid copper piping, in place. This was extremely impractical and fragile. It made the system virtually impossible to move and store. It had to be built in place and destroyed if it was to be moved. So for the next build I decided to split the interior part of the evaporator assembly and the exterior part from each other. The interior part would be firmly attached to the casing at the exit point. That, at least made it possible to take the chiller apart for storage or transportation. I should have used flexible pipes. Flexible piping is really a must for a drop-in chiller. So I regret not having taken the time to find a source for buying that. I tried, but just couldn't find any good places to buy it at the time. But next time I will.

Cutting the refrigeration pipes of the dehumidifier is a point of no return in the build. The dehumidifier is permanently broken after that, and air can get into the system and damage the compressor oil and add moisture. Also, it is illegal to just cut the pipe and vent out the refrigerant to the atmosphere. That is because of the global warming potential of the rerigerant. It should be recycled by piercing the pipe with a special valve and pumping it out into a gas bottle so that the system is empty when you start cutting the pipes. Since refrigerant is expensive it should be possible to get professionals to empty the system for you for free, as long as they get to keep the refrigerant. That said, if you live in a place where it is legal to let out refrigerant, just cutting the pipe works fine to get rid of it. I try cutting at a high point to avoid oil leaks. Ideally the system should never be completely empty of refrigerant (or inert gas) and open at the same time. If air gets into the system the compressor oil will be damaged. It is a special type of oil called polyolester oil and it cannot be in contact with air. Actually, a tiny bit of air is OK, but not for a long time, and the less the better. All components that are to be attached to the system must be completed and ready for brazing when the cutting happens. They must be brazed in advance so that only the actual attachment joints need to be brazed during the final assembly stage on the chiller. An opening for gas must always be present while brazing, or red hot gas will force itself through the flowing brazing material and create a leak, or worse. The whole operation must be planned in detail and be done with minimum heating. It must be planned so that it can be done without risk of the flame touching anything but the pipe it is supposed to touch. This may require bending pipes out slightly or setting up heat shields. Preparation is everything here. When the brazing is done I cool the pipes quickly with a moist cloth.

Compressor oil spill

A problem I have had occasionally is that compressor oil comes leaking out when I cut the tubes, or even from open schrader valves. For example, when I vacuumed my first build, oil came squirting through the hose and out of the vacuum pump. I am still cleaning oil off things in my work area, one year later. I don't know exactly how to avoid this, but try to cut and vacuum from high places in the system as oil will naturally tend to gather in low places. There was an oil leak in the system I show here too. This seemed fine at first, but when I was going to mount the evaporator, oil dripped from an open pipe section that I had brazed on. So this oil had entered that section when the system sat idle with low pressure in it.

Refrigeration piping and tools

Copper pipes come in various dimensions. It is specified by its outer radius. As usual there are two different, incompatible types: Metric and imperial. I went with imperial, since that was what my supplier, www.bes.co.uk, had. The most relevant dimension is one quarter inch (1/4") which is 6.35mm. I also bought 3/8" copper pipe (9.53mm), but didn't use it. The basic tools can be bought cheap in many hardware stores. You need pipe cutters, deburrers and benders. You also need some tubing parts. The most important of those are couplings and schrader valves.

Schrader valves

The schrader valves used in refrigeration are absolutely brilliant little devices. They can be taken apart into 3 different parts; the main part, the core and the cap. The main part has no gaskets, so it can be brazed. Remember to take off the core and cap before brazing! The core is screwed in with a special tool that is usually part of the cap. When both the core and the cap are on they form a double seal. So the cap is not just there to protect the valve, it is part of the seal. Both must be on when the chiller is finished and operating.

Flaring

While brazing is the most common way to join copper tubes, sometimes you may want a way to be able to take them apart without destroying anything. Flaring is a popular technique that gives good seals that allow high pressure and temperature. But there is a higher risk of leaks than with brazing. So it should be used with care, and as little as possible. The technique uses the copper in the pipe itself to form a gasket. This works well because copper is a soft metal. You need a special tool kit for flaring. They can be bought at aliexpress or shops that sell car tools, because flaring is used with automotive copper pipe, for example in the brake system. I found that the cheapest china kits were completely useless. So I invested in a slightly more expensive one. It forms what is called a "double flare", which means that the actual "gasket" consists of two layers of copper. The way it works is that the end of the copper pipe is held firmly in one part of the tool, the flaring bar, while another part, the yoke, allows you to put a great force on it and shape it correctly. It is easiest to just show it with pictures. Using a little oil on the tool make the process go better.

Insulating cold pipes

The cold copper pipes need to be insulated from air, otherwise they collect condensated water. Remember the dehumidifier principle above. After some trial and error I found 3 things that did the job nicely, in combination: First I bought special copper tubing insulation from www.bes.co.uk for 1/4" pipe. I also bought a type of insulating tape made for use with that material. The tube insulation had to be cut lengthwise to be put on after the system was finished. The tape worked nicely as a means to put it together again. It was also nice as an insulator of small details like valves, flaring nuts etc. But the tape could let go in exposed places like curves. So I strengthened the whole thing with cable strips.

The manifold

The manifold is a standardized and very characteristic tool of the refrigeration technician. You will need one of these. They can be bought cheap on ebay or aliexpress. Typically there are 3 hoses with 1/4 inch screw-on connectors. The hoses have schrader valve depressors in one end so that they open schrader valves when you attach them. The hoses have colors blue, red and yellow. The red is to be connected to a valve on the high side and the blue to a valve on the low side. The yellow one is connected to whatever tool you are using with the system, like pressure testing gas bottle, vacuum pump or refrigerant bottle. There are two valve wheels. These control the two manifold valves. They open or close the connection between the blue and yellow, or red and yellow hoses. Then there are two manometers. One measures the pressure in the blue hose, and the other the pressure in the red hose. When the manifold is connected to the chiller, these pressures will be the same as in the low and high side. Knowing those pressures are vital when tuning the system and filling refrigerant.

Pressure testing

After the chiller is assembled it should be pressure tested. That means to fill the system with pressurized gas. As long as no, or virtually no air is in there it can be done before evacuating. The test pressure should be higher than maximum operating pressure. I have read that 25 bar is good. The exact pressure should be noted and then the system should be left standing for a few days and the pressure measured again. If it is still the same, no leaks are present. Personally I have gas bottles with high enough pressure, but I couldn't find any valves to let me get the pressure out. All valves were reduction valves at max 3 bar. So I use that for pressure testing. It will give me reasonable leak testing at least. I always detected a little drop in pressure if I left it on pressure testing for over a week. I hope that was because of tiny leaks in the manifold or schrader valve connectors.

Evacuating the system

After pressure testing, the system has a mix of pressure testing gas and maybe a tiny bit of air inside. Now it is time evacuate it. The system has to be pumped clean before refrigerant is filled. It is particularly important that all moisture is gone. This is very important for systems that have been standing open so that air can come in. Some sources recommend multiple evacuations in that case. I felt that one time was enough for my systems. You need a vacuum pump for this. They can be bought for reasonable prices online. There are two types; single stage and dual stage. The dual stage ones have two pumping units in series, creating a better vacuum. They are the only ones used by professionals or expert hobbyists. I have never tried a single stage pump myself, but if they can create the kind of vacuum they advertise then they should be fine for this application. The pump is connected to the yellow hose of the manifold and started so that it empties the hose. Then one or both of the valves on the manifold are opened so that the pump sucks gas from the system. The pump is then left to run for a few hours. After evacuation is done, close the valves on the manifold and swith off the pump. Don't disconnect the pump right away as it will suck air into the manifold.

Refrigerant

The refrigerant used in all dehumidifiers that I have seen is R134a. The problem with this refrigerant is that there are environment taxes on it. The taxes themselves are not a problem, but the administrative restrictions around it makes it impossible to get where I live. So I decided to buy something called EnviroSafe R134a replacement. It is a secret blend of chemicals, but as far as I have understood it is mostly propane (R290). Propane has the disadvantage of beeing highly flammable, but I think it is a better refrigerant. EnviroSafe claim that their blend can replace R134a in any system. I believe international safety rules allow up to 150 grams of flammable refrigerant in household systems, which is plenty for my chillers. The refrigerators in my apartement run on R600a, isobutane, which is equally flammable. R290 is compatible with all compressor oils, so it is not necessary to worry about that either.

Refrigerant charging

Charging refrigerant was one of the great mysteries of refrigeration for me for a long time. In the books it just says that you should charge the correct amount, by weight, according to the system manufacturer's specification. That didn't help me, since I was the manufacturer myself. I am not able to explain everything around this in this article. This is an area where you need to study way more. But I'll talk about some parts of the subject at least. One thing that makes it so mind boggling is that when you add refrigerant to an already running system there will be several effects that can work against each other: The first and most obvious thing that happens is that more refrigerant evaporates and condensates, and thus the whole cooling effect of the system increases. It performs better. But there are of course preconditions that must be met for this to happen. All the components must be able to handle the added load. The second effect is that both the low and high side pressures will rise, which also means that the evaporation and condensation temperatures get higher. For the high side this means that the condenser can get rid of more heat, which is good news. The compressor will work harder, which increases performance, but there is an upper limit to how high this pressure can be. The limit must not be exeeded, or the compressor will be overloaded. Higher temperature is bad news for the evaporator because it will have its capacity reduced, just as the added refrigerant demands more capacity. So care must be taken to make sure that the evaporator doesn't get overloaded. Eventually one of these two will hit the limit and the system will then be charged to the max. After much testing and thinking I concluded that as long as the evaporator and condenser are oversized, that is, they have plenty of capacity compared to the compressor, and the capillary tube is reasonably sized, it should be possible to just slowly charge refrigerant into the system until evaporation/condensation temperatures are acceptable. In practice this means starting the system with too little refrigerant and adding more, which causes both temperatures to rise, until one of them has reached a predefined maximum. If the other one is, by chance, at its maxium at the same time, that means the capillary tube has perfect length, and all is fine. If the evaporator reaches max first, the capillary tube is too short. If the condenser reaches its max first, then the capillary tube is too long. They won't reach max at the same time in practice, but the closer the better. You need to change the capillary tube if things are too bad. Note, as mentioned above, you don't measure evaporation/condensation temperatures with a thermometer, but with the manometers on the manifold. So what should these temperatures be? For the condensation temperature this may be 45C which is OK for the compressor and for removing heat. For the evaporator it depends on how cold aquarium you want. For my example chiller I decided to go with evaporation at about 0C as the water temperature would be around 10C. According to my pressure/temperature chart for the refrigerant, that meant that the high side pressure should be 12 bar absolute pressure and the low side should be about 2.9 bar absolute pressure. On my manometers I would read 11 bar and 1.9 bar, called "bar gauge" or "barg" because the manometer shows pressure relative to earth's atmosphere which is one bar (atmosphere) less than absolute pressure.

To charge EnviroSafe from small boxes I need a special valve that pierces the top of the boxes with a needle. I have this attached to a special hose with another valve to control the flow. I call this hose the refrigerant hose. When connecting the refrigerant charging equipment it is important to make sure no air enters the system. If I just unplug the yellow hose to the vacuum pump, air will get sucked into into the manifold. Also, the refrigerant hose has air in it. So I first open the control valve on the refrigerant hose so a tiny flow of refrigerant flows out to flush it. I then connect it to a schrader valve on the yellow hose connector on the manifold (se picture of manifold). The refrigerant leaking into the refrigerant hose will be sucked into the manifold and yellow hose. When the pressure is higher than atmospheric I unplug the yellow hose and put the refrigerant hose where it was. The vacuum pump is now unplugged, and I am ready for charging. It would be better to have a valve on the end of the yellow hose so I didn't have to remove it at all. It is important that the refrigerant always leaves the bottle as a liquid. Otherwise the different components can be distilled away from each other. So it must be held upside down. The first part of charging is easy: The high side manifold valve is opened and then the refrigerant valve and liquid refrigerant is allowed to flow freely until the pressure is at about 4-5 bar. Then all valves are shut. Most of the refrigerant should be in the system at this time. Now the tuning starts: First I start the chiller and carefully watch the manometers on the manifold. Before I start I have decided a maximum pressure for both of them. If there is too little refrigerant in the system, then the pressures will be too low. I then add refrigerant, very slowly and carefully, to the low side. Only small quantities must be added at a time. This is to allow the pressure to distribute across the system, and to be sure that only refrigerant in gas form enters the low side. Liquid refrigerant in the compressor will damage it. But it must still leave the bottle as a liquid! Both pressures will now rise, and when one of them reaches maximum, the charging is done.

Thermostat

My chillers are wired so that they will always run as long as there is electricity. So to get the right temperature I use external temperature controllers. I found the "Lucky reptile thermo control PRO II" to be user friendly and affordable.

Conclusion

This must be the longest article I've written! Still, I have only been able to write a short summary. There is a lot more to be said, particularly in the area of refrigerant charging, refrigeration theory and design of heat exchangers. But this will have to do for now. Hope you find it useful!