A better defrost for heatpumps
 

I was looking around to find out if there was a better way to do defrosting than my current method (which is very basic, if the return temperature is less than X for however long then defrost).

I came across this interesting article:
Why Are Heat Pumps So Dumb About Frost?

So it got me thinking. I have noticed that frosting is related to external dew point so a combination of humidity sensor, barometric pressure and temperature seems a good combination.

There are really cheap pressure/temperature/humidity sensors available, the BME280 being one so seems like a good idea to order some and experiment...

Acuario

Please please help figure this one out! AC hacker and I started a thread awhile ago to try to get together a simple general purpose heat pump controller. A lot of versions became the outcomeb rather than one, mainly due to custom systems.

The one element missing from my stamp is a reliable defrost control. I played around with a few different algorithms , but was never really happy with them. Either the unit would defrost o.k. way too often or hardly ever. I ended up using a defrost board in that unit.

I put up a sketch on github and posted one inline in the thread but nobody had anything helpful to offer. The code is done on arduino IDE for the uno. I don't believe AC got a finished controller built either. His was for a water to water unit, and he had an air source outdoor unit he was operating on as well. He is no good at writing code, and I'm not very much better.

I like the modular and robust nature of your control system. Much more flexible than what we were doing. Please elaborate on your control scheme. If i know which direction to go, I can probably get there.

Well I finally found time to do a bit of very provisional testing using BME and BMP pressure sensors.

It looks like it is going to work as there is a measurable pressure difference when the airflow is restricted. I hacked together a very basic test rig with a sensor in a sealed box with a tube going to the back of the fan and the other sensor hanging outside and was getting pressure differences of nearly 0.9atm between an unobstructed and 20% obstructed airflow - I think this is roughly when efficiency starts to be lost.

So now to build and incorporate the sensors into the machine and some further testing. Plenty of time till next winter to fine tune it..

Acuario

I've now built the pressure difference measurement sensors into the machine.
One sensor is enclosed in a sealed box (ex screws!) with a tube going from the box to the back of the fan. The box is sealed airtight with tape and hot glue, there is a BMP280 inside, the other sensor which is located on the outside of the machine case is a BME280 so I can detect temperature and humidity as well.

http://ecorenovator.org/forum/attach...1&d=1492336475

http://ecorenovator.org/forum/attach...1&d=1492336475

http://ecorenovator.org/forum/attach...1&d=1492336475

I'm getting readings of:
Machine not running difference 34 -> 43
Running no restriction difference 5 -> 17
Running restriction 20% -2 -> 6
Running restriction 50% 0 -> 6
Running restriction 100% -20 -> -9

The difference when the machine is not running is due to the sensors reading different values so will be different for each installation and needs to be taken into account.

From the above (and yes I need to do some more sampling) it's certainly possible to detect (ice) restriction to airflow and so it looks easily possible to incorporate this into the defrost detect code.

Acuario

Slick! You could also write some false triggering software to recognize a evaporator blockage from other than frost ie dust, mold, bent fins, errant plastic bags - if its defrosting in non frosting humidity/temps it would alert you.

Be sure to put some kind of foam or aquarium filter on the end of your sensor tubes - mud dauber wasps, spiders etc love to plug up little open tube ends.

Early heat pumps used simple spring loaded flaps in the air stream which, when the frost blocked enough airflow, tripped a limit switch to trigger the defrost cycle - not sure why the industry moved away from that.

Acuario,
the pressure difference may also be related to the length of the tube so keeping it as short as possible might help. Also, there should be a way to match the sensors before installing them to get the smallest differential. Tighter tolerances means more money I suspect. I can see it being a PITA if many of these went into installations with differing results.

Outstanding. I may have to try this out. My pump defrosts way too often in my opinion and there never seems to be any ice buildup on it. Wasted evergy.

..the tests continue now it's winter..

So far sadly not too good. The tests I did initially were using plastic bags to restrict the airflow and then I was getting good results. Now I've had a few (not many) real scenarios and disappointingly the results didn't match the tests.

I'm continuing to monitor but as we have few days cold enough to cause frosting progress is slow.

I'm going to investigate another approach in parallel; monitoring humidity (from one of the sensors) as well as temperature. As the humidity hugely influences frost then this may be a more reliable method and it may be possible to calculate probability based on humidity/exterior and evaporator temperatures.

If this still doesn't produce the required results then there is always the IR emitter/detector method but I'm going to see how I get on with the humidity one first.

Acuario

Have you given any thought to using a color sensor like the one below? Detect white, defrost until non-white. I dont know if sensors in the price range below will work or not. Worth a try maybe.

Color Sensor TCS230 - RobotShop

Thanks for the pointer. On ebay the basic version is £1.59 with free shipping - so cheap I've bought one just to see what it's like.