DIY Air Conditioner Built From Weird Donor Appliance

There are some parts of the world where living without air conditioning borders on unthinkable. But in more moderate climates, it isn’t all that unusual. [Josh’s] apartment doesn’t have central air conditioning — the kind that connects to a forced-air heating/cooling system. It does, though, have a water circuit for air conditioning, so he decided to hack a few experimental air conditioners.

He’s not starting completely from scratch. The two attempts he made at building his AC came from donor parts. The successful one started out as a hot water heater. The very first attempt didn’t quite work as well, using a refrigerator compressor and an evaporator from a baseboard heater. The flow control through the heat exchanger turns out to be very tricky, so [Josh] claims he mostly got ice right at the inlet and minimal cooling through the evaporator.

The more successful one works better but still has a problem with the evaporator freezing that he’s trying to solve. He’s looking for suggestions on how to make it work better. As much as we like a good hack, our advice is to move to a different apartment building.

We’ve seen other homemade coolers, but they are more like swamp coolers. If you just need to cool your desk, you might just get some ice in a metal can.

58 thoughts on “DIY Air Conditioner Built From Weird Donor Appliance

  1. Use a bigger evaporator, or run the compressor less.

    I saw a construction in one showcase cooler where the evaporator ended in a long-ish copper coil that was placed inside a bucket. The condensing water would slowly fill the bucket and make a water bath with enough thermal mass to boil any remaining coolant during the compressor’s running cycle. Point being that if there’s any liquid coolant going into the compressor, it will seize – it all has to evaporate.

    Likewise, if the problem is freezing the evaporator, then a solution would be to put all the coolant through a copper loop in a water bath with enough thermal mass that it won’t freeze solid. Then use the cold water through the radiator.

    1. You need to figure out how many BTUs the compressors output is then figure how many CFMs you need to move across that coil and it won’t freeze it’s 400CFM per 12000 BTUs you could probably get a bathroom exhaust fan or stove exhaust fan or some type of fan should do the trick maybe put the evaporator in its own case and force the air through it

    2. The expansion valve is supposed to regulate the superheat of the gas leaving the evaporator. If your evap is freezing up that suggests to my mind that you definitely do not want more refrigerant, you are shifting more heat than your existing fan and evap coil combination can handle and, I think, you need to try much harder to get better airflow over the evap coil. It is not enough to just put a stronger fan. What you need is an air flow that is more effective at getting warmer air into good thermal contact with the evap fins and then move the air on asap to get fresh warmer air over the fins, not just a stronger fan. I have found with some component arrangements it can be better if the fan blows air on the coil and other arrangements it was better to suck the air over the coil. It is also possible I think that your tx valve may not be adjusted properly. I’ve only used electronic valves with my own controllers so I don’t know how to adjust the type you have (assuming it is adjustable?).
      A guess as to what the downward loop in the suction line before the accumulator is for: if it is not just to isolate vibration then it might be the first line of defense against flooding. Being a thin tube it will not hold a lot of liquid and so will also not hold a lot of lubricant. If it did the compressor may be damaged. The accumulator is designed to hold liquid refrigerant and uses the natural stratification of oil and refrigerant to return the oil to the compressor but not the liquid refrigerant. This is what the aperture in the bottom of the loop tube inside the accumulator is all about.
      With your second system, you have done a nice neat job of that but I didn’t see a vacuum pump in your kit. Did you fully evacuate the system before charging it? If not then all the partial pressures of the contaminant gasses you would have would certainly explain poor performance and the moisture could exceed the filter drier capabilities and then destroy the lubricant in the compressor.
      I think you mentioned or someone mentioned the evap is only cold at the entry point. That could have hundreds of causes but they all come down to this: all your refrigerant is evaporated way too quickly. If the pressures look ok then maybe contaminated gas (other gas mixed in), over zealous tx valve causing a compressor overload and low flow rates and the one that gets me all too often, flash gas at the expansion valve which can kill system efficiency stone dead.
      The best solution to an evap that gets icy, as long as it isn’t every few minutes is to do a de-ice cycle (turns fans off and reverse the refrigeration for what is usually less than 60 seconds to get the evap coil a couple of degrees above freezing and the ice gone).
      I hope some of that might help and not be an insult to your skills and knowledge.

    1. If the point isn’t to collect water/dehumidify, could you get extra mileage out of the cooler by running a swamp cooler right next to it? Dribble the condensate onto a canvas with a fan running next to it.

      1. Swamp coolers only work well in low humidity environments. I used to spend my summers in New Mexico, everyone had a swamp cooler and they worked well…but move several hundred miles east to the DFW area and they don’t work at all. My boss bought one for the shop and all it did was make the machines rust, we just use it as a big fan now.

        If your collecting enough condensate to wet a swamp cooler, the RH is too high for it to work correctly…at least in my experience.

        Think about when it’s hot outside and you’re sweating, if the air is dry you more comfortable than if the air is humid sweating just makes you more miserable.

    2. If the refrigerant level is low in the system, increased airflow will not help. Aside from low refrigerant, too small of a metering orifice will do the same thing.

      Either way your suction pressure is too low.

      1. No, the suction is too high. Never ever use “vacuum pressure” or “suction pressure”. Vacuum and suction are just that, pressure lower than ambient, so there’s no need to include the word “pressure”. Pressure is for higher than ambient pressure.

        That’s why you suck on a straw but a leak in a spacecraft causes things to be blown out. In space the ambient is zero in any scale of measuring pressure. A ruptured tire is a blowout not a suckout.

        Same reason why in electricity the word “second” isn’t used since the base units are defined by the second. A Watt is a specific amount of electric power in a second so Watt/second = Watt, but if you need to know how much in an hour you use Watt/hour or kilowatt/hour, or megawatt/hour etc.

        1. So Rob says the suction pressure is too low, he means the suction line pressure is too low, not that the suction amount is too low. The suction line pressure in a system using residential refrigerant will be well above ambient atmospheric pressure, but it is still referred to as the suction pressure because that is the line where the refrigerant is being sucked into the compressor.

          1. Partially wrong. With most common refrigerants also the suction pressure is well above 1bar_abs, especially for air conditioning.. Only refrigerators with R600a (isobutane, boiling point at 1atm = -12°C) have suction pressures below ambient.

  2. I do hvac for a living, and I don’t even know where to start with this. The first video is frustrating to watch, as he states “I know why it’s icing up, but I can’t do the two things that would make it stop (adjust charge, increase airflow), how do I make it stop?”

    Hopefully he’s just doing this as a fun hobby project, and I wish him hours of enjoyment tinkering with this stuff.

    If he actually wants his building hvac problems solved, he should complain to his landlord to get a professional to fix the issues, because “working hvac” should be a condition for his lease.

      1. Your right, it is a luxury, a luxury hes clearly paying for when he pays his rent. and you must deal with some great slumlords in “south MS”. up here if shit doesn’t get acted on, cool or heat, people start withholding rent payments and getting lawyers.

        1. Maybe in the state you live in, there’s a legal requirement for a landlord to provide cooling – but in most jurisdictions they are only required to provide functional heat.

          Providing cooling is NOT a legal requirement, so unless his lease explicitly states that functional cooling will be provided (which it doesn’t in most cases), then he has no legal ground to stand on.

          You assume he’s paying for this luxury, but I’ve only once lived in a rental where cooling was part of the lease. In every other case I’ve had window ACs.

          He’s actually lucky he seems to have plumbing for water circulation that he’s alllowed to touch. My lease explicitly states that I am not permitted to make plumbing modifications.

        2. Obviously you dont read. Withholding rent will get you evicted
          Commercial the tenant is responsible. Residential its on the owner to fix a rentals ac, if the house does not have it the owner does not have to supply it in a residential application. Most slum lords have window units in each room of the trailor.
          Luxury = can live without. And whats up with quotations on south MS. or do you not understand

    1. Playing with propane is fine for a free standing house out in the country. I would not like it so much for someone without a clue to be playing with a flammable refrigerant in an apartment where their mistake can burn other people’s apartments up.

      Personally I would just save my quarters and buy a mini split system. The new ones are super efficient, very reasonable price wise, and easy to instal. The R410a units can heat to down near zero F or less, and cool in the summertime. The variable speed is wonderful. Whisper quiet the vast majority of the time. Once you have tried one you will not go back to a window rattler again.

  3. Get a thermostat that has a defrost cycle. After x amout of hours the compressor is shut off for x amount of minutes while the fan continues to run. This is called air defrost. I could recommend a Dwyer tsw 150.

  4. The ISS uses Ammonia in their cooling system. Hard to get with out a license. Seriously though Man has survived for a million years without A/C. I believe it is one of the causes of climate change. Fans [DC] [I use a radiator fan ]can easily be powered by solar and work great when Sol is shining

  5. Guy in video may just need to run the Frankenconditioner at a lower duty cycle.

    The compressor may be too efficient dumping all that heat into the water loop (remember normal AC units are designed to dump heat with large air exchangers which are not nearly as effective as water in dumping heat) and bring the cold side down to the point that it’s freezing the condensation. Run the fan to blow air for a while before and after each compression cycle and lower the duty cycle to avoid the freezeups.

  6. Purely imagineering at its best and Will never work properly. All the suggestions made here to resolve the problem are based on pre- engineered systems that weren’t functioning properly. Good luck!

  7. Need a cap tube or txv to meter refrigerant and constant air flow over evaporator. The water cooled condenser is a lot larger than you need. Need to control water temperature rise thru condenser.

  8. I want to see someone mod a large swamp cooler with an air-to-air or air-to-water-to air heat exchanger so the damp air stays outside while absorbing heat through the exchanger.

    A swamp cooled building feels good when you go in out of the hot sun, for about 5 seconds. Then it’s worse than being outside because our stupid human autonomic systems attempt to dump heat as fast as possible by cranking up the sweat, instead of allowing for conduction and radiation to the air to handle it over time.

    Rather than do that, your heat management system assumes that you might just suddenly walk right back into the hot outdoors instead of staying indoors for an hour to cool off.

    The same happens in an air conditioned building but since the air is drier that initial flush of sweat evaporates quickly, cooling the temp sensors in your skin. Your heat management then twigs to the reality “Oh, hey, I’m in *cool air* so I can shut down the evaporative cooling system.”.

    With cool and humid air that initial flush of sweat doesn’t evaporate, you don’t get that cooling. Thus your heat management assumes you must still be hanging out in the heat and *sweats more*. As you get coated in more and more sweat that’s not going anywhere, your skin is insulated from being able to conduct and radiate heat to the air, so you can get *warmer* than when you first walked inside.

    Depending on your individual biology, you *might* be able to acclimate to cool but damp air after a while, but some people just can’t and will sweat a huge amount and overheat in a swamp cooled space.

    That’s why I hate swamp coolers, they’re useless for me, making me sweat worse than just standing outside in some shade on a hot day.

  9. Basic water source heat pump. System sized in btu’s. In this case heat in equals heat out. Air side or evap for sensible cooling is cfm X 1.08 X temperature difference / 12000 =’s tons or btu’s. Water side or condenser is gpm X temperature difference / 24 =’s tons or btu’s.Thus heat in =’s heat out. Equipment lable should provide size. Assuming air flow and water flows are correct and a icing condition exists probable cause is low refrigerant charge or faulty tev. Reading evap pressure and suction superheat are your indicators.

  10. Need high and low side pressures and temperatures to figure out your subcool and superheat. Cheap eBay or harbor freight gauges should be good enough if you don’t want to do it for a living.

    Honestly best upgrade would be a txv to control your refrigerant flow. Hard to recommend one without knowing the specifics of your compressor, a Sporlan fj12c might work going just by the size of your evaporator coil.

    Your current metering device probably isn’t doing much to spray the refrigerant in, should be a mix of 70% liquid 30% vapor. Cheapest devices are a fixed orifice or piston.

    AC system design is a whole field of engineering. Start with your cooling need in BTUs per hour, select a compressor that can pump the required amount of refrigerant, select a condenser that can adequately cool that refrigerant back into a liquid, select a metering device to control the flow, select an evaporator big enough to absorb the heat without requisite airflow sounding like a jet exhaust and finally size all of the piping to maintain adequate flow and velocity to avoid oil separation. Then you can get into tricky things like efficiency.

    Also doesn’t look like you properly evacuated your system of air and water, noncondensables and water vapor will wreck havoc on a system, it will never run right even if properly designed.

  11. First thing you have to understand, you are not “making cold”, you are REMOVING HEAT. That little tidbit should go a long way to help understanding why it does or does not work. That being said, you have a long way to go to make this work. Trying to get a good balance between your evaporator coils area and airflow to match your condensor coils area and water flow is going to take a good bit of math into consideration. The water is going to conduct the heat out much faster than the air is going to shed heat into the system. Compounded by the fact that your using a different refrigerant without a proper orifice, and a much lower area evaporator coil.
    Would have been better to leave the original device intact, find out if it had proper refrigerant pressure, consider wether or not it could push enough BTUs to do the job, and finally figure out how it’s going to shed the heat outside of your dwelling. You are never going to create a quiet system without very large coils and very large, low speed fans.
    I don’t mean to discourage, it’s a learning experience. Nothing ventured, nothing gained.

  12. With only 11 more years before the planet is ruined by man-made CO2,.. I was really hoping to see a cheap, easy, ‘green’ air conditioner solution. This doesn’t seem too cheap or easy, unless it’s just a matter of finding the right scrap parts to patch together. I grew up out west, never had air conditioning, didn’t really need it. But, down here in Florida, and the extra few degrees from climate change, I can’t see living long without it. But, then again, I’ve never owned a car that had AC that worked very well. My 2003 Explorer doesn’t even have heat, least I won’t need to worry about that…

    I have to agree that AC is a luxury, and don’t recall it ever being mandatory. If you pay for it, you’re entitled to it, but doesn’t mean it has to be good. Have to know what you are getting, before you sign the paper. Personally, I think AC is over used, lot of energy wasted. Likely to change with the climate, green energy ain’t cheap, and more people will want to beat the critically hotter climate. Our electric cars will need to be charged as well. Guess we’ll all be good, long as the wind keeps blowing, sun keeps shining…

      1. There’s a brand – Gree, I think, that provides a model that works directly off solar PV while the sun is shining, and switches over to mains when solar drops below a threshold. It requires ~1KW of panels dedicated to the unit, and the best thing is that it’ll work during a power outage – so when everyone else’s AC is causing brownouts or rolling blackouts because there’s insufficient supply, you’ll stay cool. During the day, at least.

  13. Given that it’s from a fridge essentially you would be lucky to cool a bathroom or closet and have a reasonable duty cycle on the compressor.

    The things that will grow in a swamp cooler are much scarier than any hole-e-wood B film. A week after Legionnaires was discovered they found it in the student union swamp cooler at Indiana U. Not exactly an arid part of the west.

  14. They do make portable air conditioners that do not go into windows BUT the heat needs to be vented outside (through a provided hose)!

    You can get one at Walmart for about $200.00 and up (walmart . com) OR about $170.00 at (brandsmartusa . com). I’m sure you could get a used one somewhere for cheaper. Either way, they would work as needed. ;)

    1. Portable AC are terribly inefficient: if you do not vent all exhausted hot air (and in most case you don’t), you are reinjecting heat generated by the device inside the room.

      And if you really vent all hot air outside, it means that you are removing a large volume of air from your room, that needs to be replaced by new air coming from the outside, where it is probably hotter than in your room.

      So these devices are constantly wasting energy to fight against heat generated by themselves or from constant hot airflow coming into the room.

      So you should never consider using such a portable device. Split systems are the only efficient solution.

  15. Sorry, a few have come close, with the water taking up more heat, but no one really gave a solid answer.
    You need to restrict your water flow. Typical water cooled condensers need a 15 -20 delta t between the inlet and the outlet, This does 2 things:
    One, slower water flow won’t erode the condenser coil (I have seen this happen on numerous occasions)
    Two. It will build up a higher liquid line pressure and push the refrigerant through at a higher pressure, as long as the pressure equates to a temperature above 32 degrees, your golden.

    1. This is an interesting suggestion I had no thought of! So if I raise the temp of the water, the high side pressure goes up, so the mass flow though the compressor goes down, so the low side pressure goes up, so the low side temp goes up. Right? Would not be hard to modulate the water flow rate to control water loop output temp. I will try this! But I guess I was hoping for a solution that increased the low side pressure rather then the high side because this should make the compressor run quieter(?), but anything steady state is better than having the compressor turn on and off like it does not (wakes me up every it turns on!). Thanks!

      1. Most water cooled condensers have a regulator on them to prevent over cooling, but too much heat loss can mimic a low on charge condition, same reason you warm your car up, even in the summer, you need a certain amount of heat to keep everything flowing, this includes keeping the pressure up in the system, you will get better refrigerant flow, which will bring the suction pressure above waters freezing point, Once you have achieved that, no more frost on the coil.
        It’s also important in the cooling of the compressor, you want a certain amount of sub-cooling of your compressor will overheat, usually around 10 degrees.
        As to the compressor noise, that will go down because your compressor is expecting a certain amount of back pressure, it wants to push, but with lower pressures it’s not working hard enough and tends to over heat. Have you ever taken a furnace blower out and just let it sit and run on the floor? with no air resistance, it will over amp the motor and not move much air, If you put an amp probe on it and slowly cover the output, you can watch the amp draw go down and feel the air flow increase, this is called tuning the blower. Your furnace is expecting a certain amount of air resistance from the ductwork, same effect.
        Lastly, this isn’t increasing the water temperature TO the coil, this is using the coil to raise the water temperature, I have seen these water cooled coils erode from the inside out and leak water, sometimes into the refrigeration system, no bueno for sure. if you use a regulator, adjust it to about to around 120 degree equivalent, this pressure will vary depending on what refrigerant is used.
        Sorry for the excessive detail, kind of my field for the last 30 years.

  16. he used r134a though you can also use canned air as it uses r134a or r152a and canned air maybe easier to get because in wisconsin you have to get license to handle r134a

  17. Thanks for all the helpful feedback! To all those who suggested adjusting the duty cycle of the compressor to match the load – take a look at the right side of the unit. Bolted on you’ll see a solid state relay and an Arduino. The Arduino monitors the temp at the bottom of the evap coil and turns off the compressor when ever it gets too cold, and then leaves it off until it thaws. This is totally practical and works, but my whole reason for building this thing is to try to have some quiet cool in my bedroom, and cycling the compressor is LOUD so I really want to find a steady state solution – hopefully one that lets the compressor run at reduced load (and therefore hopefully quieter) rather than only off and on. I think this should be possible thanks to the AC synchronous motor.

    1. Looking at your setup, a lot of your noise is coming from sitting that compressor directly onto wood. If you look at how compressors are mounted in refrigerators, AC units and even air compressors. Always separated from support structure with thick rubber bushings. When those bushings wear out things get really loud regardless of how well the compressor is working for it’s given setup. It can be the difference between a low background hum and hearing the compressor kick in clear across the house. Of course this is moot if the pressure is too far out of balance.

  18. For those of have suggested increasing the airflow, while this should work in theory it does not work in practice. Firstly, there is only so much airflow you can get though the geometry of that little evap coil and when you try to push too much it gets very loud- which defeats the whole purpose of the project (to make quiet cool in my bedroom). Secondly, the bottom of the coil where the refrigerant first flashes is cold enough that even a tiny bit of condensate will quickly freeze and this is a runaway process- once there is a bit of ice then the airflow over that spot is blocked and so the metal cools more so more condensate freezes and then airflow is blocked even more…. etc until it starts snowing! :) I really want to find a way to increase the low side temp above 32F by increasing the low side pressure so no frost ever. I really think I could do this with direct control of the metering device by targeting a low side min pressure (rather than targeting superheat like TXV). I just need to find a good way to do this without having to open up the refrigerant loop. Any ideas from more experienced HVACers welcome! Thanks!

    1. Yes, enjoy to destroy! You can not do useful regulation of an AC induction motor with a TRIAC (voltage regulation). This somehow works for a fan because of the highly nonlinear speed/torque ratio, but on a compressor it ruins efficiency and shortly after the whole motor through overheating.
      The only possible way of regulation is with frequency regulation. Use a VFD.

    1. I believe it is properly charged for its intended application, which is pumping heat into 120F-140F water (for domestic hot water). I am pumping into constant 90F water, so system pressure is lower. I think adding charge would be one solution, but I’d rather find a way to do it without cracking the refrigeration loop both because I’d only need a couple ounces more and don’t want to get stuck with mostly full leftover cylinder of R404A and also because I am I almost sure there is a non-invasive to do it and I want to know what that is! :)

  19. Presuming your apartment was built in the ’50’s or ’60’s, hot or cold water circulating in your thru the wall unit was the engineering marvel at the time. Since you opened the wall unit, did you examine the blower motor to determine why its noisy? If the blower motor is too old for repairs, tapping into the water system may be the easiest way to have cold air without the unnecessary hacks. With enough radiator area and a fan to blow warm room air thru, chilled water should provide cool air. All you’re doing is circulating chilled water thru an equivalent radiator of similar size to the wall unit (or larger) with a quiet fan to move warm air thru the coils.

    Unless you’re an hvac engineer, leave the ac hacking to experienced hvac techs trained in heating and cooling.

    1. >chilled water should provide cool air.
      The water loop is supplied from the building at about 90F, so I can’t just run air over that without a heat pump!

      >Unless you’re an hvac engineer

      I am EPA 608 part II, but would rather find a solution that does not require cracking the refrigeration loop here because it is R404A… and I am lazy. :)

      1. Hot water should not be circulating in summer weather. If there’s a shut off valve, use it. I presume 90F water is for heating season and not used or supplied during summer weather.

        Is your wall system using a wholly separate ac cooling system? You’re not very clear on how your wall unit operates.

  20. Sounds to me his sysem is improperly balanced. Evap and condenser are to be sized properly. If the water cooling is dropping the pressure down to low it will never work. Try slowing down the water cooling. Also everything depends on what refrigerant is being used and is the txv matched to the refrigerant. If you dont have balance you wont get it to work properly.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.