Reverse air conditioner freeze up problem
 

I'm experimenting with a 5000 BTU window air conditioner to heat a small work room in my shop. The "outside" part is in the bigger area of my shop where it was 57 degrees. I put the temperature sensor in a can of warm water to fool it into coming on. I am getting lots of heat this way and it seems to work as well as a 5000 BTU heater I normally use.

Problem is, after about 45 minutes the cold side freezes up completely. Information I read from other people said this should work well down to at least 45 degrees. Why is this thing freezing up at 57? Humidity was only at 64%. I can only assume the problem will get worse as the outside temperature drops.

This unit uses R410a and the evaporator looks smaller than what I've seen on older R22 units. Is that the problem? Would an older unit work better for this?

http://img.photobucket.com/albums/v4...0-25015323.jpg

The dew point is high.
57 degrees F and 64% humidity means that you will have dew at 45 degrees F. Once that wet coil reaches the freezing temperature and you've got a very quickly frosting coil.

I'd say increase the blower speed, but you are already on high. The only other suggestion that I could give you was from a case study done by the DOE that suggested that much of the air exhausted from the evaporator coil of window air conditioners gets cycled back through the coil again because of the supply and return air being in such close proximity. Since you took off the cover, the louvers that would usually push this air away from the unit aren't there. I'd suggest making a makeshift duct of sorts or something else to reduce the amount of air leaving the blower from getting back to the coil, otherwise you are inviting the frost. Something as simple as a cardboard sheet sticking out a few feet between the top and bottom of the coil could make a big difference

It looks like what you need to do is install a temperature sensor that shuts off the compressor and leaves the fan running when the coil gets near freezing to give it a chance to blow the warmer air over and allow it to defrost. With temperatures well above freezing and with the coil not completely frosted over, the defrost cycle would probably only take a few minutes. Cheaper heat pumps usually use a timer, I'd imagine some trial and error over the amount of time the compressor runs and the amount of compressor off time needed could be figured out to develop a crude but useful defrost system.

Good luck.

A quick warning: don't run the unit frosted up like that for any extended amount of time. You'll be finding a nother unit to experiment with when the compressor burns up.

It would be interesting for you to buy a cheap non-contact IR thermometer (harbor freight has them for about $10), and measure the temperature of the evaporator HX (the cold one).

Once you see how cold it gets, you will no longer be surprised by the formation of frost on the coils.

What you need to remedy this situation is a heat pump controller that will sense frost in the coils, and either pause the compressor, and wait for the frost to melt, or else run a de-frost reverse cycle.

A project thread has been started that would provide you with a working controller for just such a conversion as you are attempting.

Unfortunately, after three and a half months, and 270 posts, that project thread has yet to produce even an elementary controller that could serve your purposes.

Best,

-AC_Hacker

That's what I was thinking. Problem with that is when I shut it off, and the ice melts, that rush of water hits the fan and I get some water mist out the hot side. I suppose I need to modify it to give the water a better path out.

I will put the covers back on if I get it working how I like, just wanted to fool the temperature sensor before I go hacking up the wiring.

Another thought I had, what if I hooked it to a VFD, and slow it down some? I would get less heat, but maybe I could stop it from freezing and then it wouldn't need to cycle on and off as the room gets warm enough.


Just curious, why is that? I thought they could just run that way to make a block of ice? (Not that I need to)

the cheapie thing you want is this:

DT45 Refrigerator Defrost Thermostat L 45 Degree w Clip for for 1 4" 5 16" | eBay

Wire it in series with the compressor, clip it onto the tube running from the evaporator back to the compressor (not the one with the capillary tube). When the coil freezes up, the contacts open and stop the compressor. The contacts close again when the tube has warmed from thawing. You can tweak the operation by moving the thermostat to different points on the copper tubing.

To answer your question about running a frozen evaporator: the compressors in window a/c units are not built the same as those in freezers and refrigerators. They basically are all guts on the inside, with not much breathing space. There is not much room for extra heat, refrigerant, oil, or anything in there. They are also optimized to run at a certain compression ratio and volume flow.

When the evaporator freezes, a few things happen. Heat flow into the evap slows down, causing the boiling of refrigerant to slow. This forces more liquid further through the evap and sometimes into the compressor. Compressors don't like to see liquid at all. During moderate events, things like bent or cracked valves happen. During wet events, burnt windings, bearings, stalled rotors happen. Not good...

Is this what you are using on your converted air conditioner to solve the frost problem?

-AC

One of them. It used to be a dehumidifier, one of the old school brown boxes. It runs off r22 and has not been modified as far as the refrigerant circuit goes. It was torture tested in a few different configurations. I got a couple of extra degrees of operating out of it by catching the condensate and warming it with the compressor discharge line. When the klixon switch opened, a little fountain pump sent the warm water up to the frosted evaporator coil and sped up the defrost operation.

The window unit was equipped with a reversing valve and a defrost control that could be wired with the same type of switch or a thermistor. I ended up using a thermistor because it was adjustable.

I don't recall that you did a thread on these...

-AC

AC,

I didn't figure they were really worthy of their own threads. They were more or less proof-of-concept devices to help me get an idea of this whole phase-change realm of things. I have made mention of some of them in other threads belonging to whoever. I believe I have posted a few pics along the way.

Once I gained enough insight into the potential of these units, as well as some idea of how they act, I built the window a/c heat pump in my thread. If you'd like, I can start a couple of threads about them on their own. Most of them still live, but wifey banished them to the barn or shed for one reason or another.

F357,

There is a quick, easy way to make the unit you have work off its own thermostat for heating duty. Grab a 120VAC SPDT relay from radio shack or some other parts store, along with an SPDT switch. Disconnect the thermostat from its load, leaving the live side connected. Rig the relay coil to operate off of the thermostat. Connect the common terminal of the relay contacts to the live side of the thermostat, and the load to the common terminal of the SPDT switch. Connect the NC contact of the relay to one side of the SPDT switch, and the NO contact to the other side of the switch. In one switch position, the thermostat will be active in cooling. In the other, it will be active in heating.

For heating operation, you just need to fish the sensing bulb through to the back side of the window unit. Since the unit draws air in from the sides and out the back, the sensing bulb will be exposed to fresh air before it is heated.