Automatic defrost circuit
 

Well winter is just around the corner again in Northern Spain as I write this on 29th September so it's time to revise the heating system.

Last year the home made heat pump was a fantastic success so I'm obviously going to continue using it. The only problem I had was a lack of an automatic defrost for those extra cold nights. The result was a block of ice hung on the wall - not too efficient.

So here is my attempt at solving this minor problem - a PIC processor controlled defrosting circuit.

The circuit itself is pretty basic and uses a Dallas DS18B20 digital thermometer to measure the temperature of the evaporator. I've started with a temperature of -2 degrees celcius as my trigger for the defrost cycle - this is a bit of a guess, anyone have any better figure?

I could have used a thermistor and the pic ADC circuit but I wanted to have a go at coding for the 1-wire protocol and I also use the same code in other projects so why complicate life?

When the temperature reaches (currently) -2 then the defrost cycle kicks in. Again, the cycle itself is a best guess but it can be easily modified if necessary. The cycle is:

• Turn off compressor
• wait 1 minute
• Switch to cooling mode
• wait 1 minute
• Turn on compressor for 5 minutes (defrost)
• Turn off compressor
• wait 1 minute
• Switch to heating mode
• wait 1 minute
• Turn on compressor

The pic code was compiled using the Microlab X IDE (free) and the schematic using KiKad (free).

I decided to use a relay module bought from ebay for the valve control and a 40A solid state relay (also from ebay) for the compressor control. You could design/build your own bits for this but I had them available and they hardly cost any money.

I've attached the circuit, some photos and the code I wrote in case anyone else needs a defrost circuit. I'll update with how it performs when the temperature drops!

Acuario

Your best bet would be to use 2 thermisters, one for the ambient temp (air on) ad one on the fins (at the air on location). Remember that your liquid line is going to be X deg colder than ambient so you will need to start monitoring when the ambient is X above 0C (maybe 3-4C). Turning on defrost is not the hard part but knowing when to turn it off is, and that is where comparing the two temps helps. You can set the differential to determine when you think there is no more frost.

Optical detector 'optical frost sensor'?
 

For my systems, I would need a detector to work down to -15 C, so measuring temperatures to detect ice would be difficult.

But seeing ice isn't hard at all. :)
Here are two examples.

Frost Controls SP-820 for Debron 1052 Optical Sensor | eBay

This sensor is wing mounted. Once ice fills the gap/notch, the Ice alarm triggers.
http://www.newavionics.com/Images/9732-oem_text.jpg
New Avionics: Introducing Ice*Meister Model 9732-OEM, Aerospace Sensor for embedding into host flight systems
PRINCIPLE OF OPERATION

Ice*Meister Model 9732-OEM ice detecting transducer probe functions as an optical spectrometer. It monitors both the opacity and the optical index-of-refraction of whatever substance is on the probe.

The transducer probe is excited by a microwatt IR LED coupled to the driver plastic optical fiber. A likewise-IR detector is coupled to the receiver fiber which inputs to an op amp and comparators.

The threshold comparators register the IR signal voltage-analog from the op amp.

NO-ICE state is reported back to the op amp as a stable mid-level signal. When the ice detecting transducer probe enters an icing domain, it can attract either rime ice or clear ice.

~~~~

I'm pretty sure that I could make a frost sensor using an LED (or laser LED) pointed at a light sensor.
With the light 'beam' going pass the area where frost builds up on my outdoor coil.
I would design the detector circuit so it would send a Frost signal when 70% of signal was blocked by ice or frost.

I would have to insure the beam passed within ~5 mm of the HX coil's outside surface. (near the bottom, where it gets thick).

The ice on the sensor might melt (during defrost) before the HX was completely clear, so a timer might be necessary.

That looks cool. I will have to look into it further. All ASHPs have the same problem of working down to -15C or so although there is not much moisture in the air at that temp.

What is the spacing between fins on your unit? Can the Debron 1052 fit?

Measuring temps is not that hard with thermisters and most of the HPs I see have them installed plus all the outdoor reset systems use them so it is a known technology. I would like to figure out the programming to have it as a couple of inputs on a PLC and do away with a separate circuit.

I wasn't able to find much info on the Debron 1052 Optical Sensor,
But, it's a good example of an emitter-detector open-gap sensor.

These sensors are commonly used to detect solid material across an air gap.
Kinda like the object sensors used on a garage door opener.

It would be easy to try out a 99 cent reflector sensor, and glue a small mirror on the HX coil.
http://www.vishay.com/docs/83760/tcrt5000.pdf
(Has a Daylight blocking filter • Emitter wavelength: 950 nm)
New 10 Pcs TCRT5000L TCRT5000 Reflective Optical Sensor Infrared 950mm 5V 3A | eBay
Have the sensor 10-12 mm away from the mirror on the coil.
When frost weakened the signal, the transistor would signal your defrost alarm line.
But, not if the ice acted as a mirror!! Have to try it to be sure..

I'll order some to try. It's takes a few weeks to get them,
and it will take a couple of days to set up a test rig and post some results.

I'm thinking of trying to make the mirror-to-sensor spacing adjustable,
so I can widen the space until signal loss causes signal switch.
Then, move it in a bit and do the frost testing.. Maybe CO2 spaying?

http://4.bp.blogspot.com/-CTkCMFcNEB...tcrt5000_2.PNG

TCRT5000 part test
 

I used the humidity input on the CAI board to read the optical detector with the circuit above.

The range using a mirror was about 2.25". When I inserted foggy plastic
between the mirror and sensor, I only got change of about 30%.. (90% to 60%).

But when I didn't use the mirror, it displayed 0% until I placed an ice cube in front of the sensor.
The ice is very reflective. At 1/2" it displayed 30% and at 1/4" it was 90%.
I checked the ice in both white and clear parts of the cube, the reflectivity was good in both cases.

It seems like setting up the sensor so it points right at the HX coil
at a distance of 1/2" would detect ice/frost building up.
I would use a shot of flat-black spray paint on the coil, for near zero reflection. (When the coil was ice-free)..

Using a standard analog input would have worked better, but the humidity sensor input has screw Euro-terminals.

With a control board, it should be pretty easy to use this type of IR detector to control defrost cycles.

It will have to be tested in real time under many types of snow, rain, sleet etc. I would be interested in putting it in the next HP I make but that is a few months away.

Mike, I noticed the IR LED light level was hardly effected when I placed semi-clear plastic between the LED and mirror.
That means it might be simple to make a water-tight enclosure using a plastic bottle with a flat surface for the IR window. (the bottom maybe).

Or, just encapsulate the leads and wires on the bottom of the device, using RTV or non-conductive epoxy.?.
Maybe use the adhesive to bond the device to some aluminum angle iron, or some clamp-on mounting device?
Whatever is used, it should be a solid mount that won't allow movement.


This device should be mounted in the shade, to avoid interference from the sun's IR,
but also to allow the plastic lens on the LED and photo-transistor to age more slowly..
(If you did leave them sticking in the weather).
It should be facing the area on the HX where the ice normally forms first.



http://4.bp.blogspot.com/-CTkCMFcNEB...tcrt5000_2.PNG

When configured pointing at a black spot on the HX, (~1/3" away) I think the output is going to be very close to a full 5V. (TTL logic 1)

Once some reflective ice or frost forms in the IR spot light (LED),
the bounce back is going to turn on the photo-transistor,
and pull the voltage down towards near zero volts (logic 0).

I'm not sure, but I think the ice might bring the voltage down
low enough to be a true TTL compatible logic 0..

TTL Low or Zero= 0V to 0.8V High or '1'= 2V to VCC VCC = 5 V ±10%

So, maybe a fancy A/D input isn't really needed..?.
That could make these cheap sensors a lot more useful..

I've got to find my clip-on voltmeter probes.. :)

I was wondering whether a system like this coupled with something that measures the exact distance between two fins, which would have to be precise, and compare the distance with frost/no frost on it. There are many types of frost or snow, some more reflective than others so maybe it would work better with two sensors........or maybe I am making things a bit too complicated. I had a big problem with defrost last year on an ASHP and it is still not resolved.

The evap design i have now will have the coil at about a 20deg angle from vertical so your sensor may work without being in a special compartment as it is under the coil and away from the sun.

http://i46.photobucket.com/albums/f1...r/NCL/A017.jpg
(Looks like lower middle would be a good place for a sensor).


Since there are air passageways in the HX fins, you could use two these detectors units, one on each side of the HX unit.
(using the LED on #1 and Photo-Transistor on #2).
I think the range in a dual device setup would be 3 to 4 inches.
Shoot the IR beam between the fins, to turn on the transistor, putting out near zero volts...
When ice filled the voids, less light passing would cause the voltage to increase towards +5.

The only problem with that idea, is really clear ice. (Optically clear).
Clear clean ice would likely pass a pretty good amount of IR light. :(
To work well, you would need a good ADAC to monitor the analog level.
And most likely a programmable controller to decide what voltage level = ice..

My Sanyo HX coils are stacked layers (dual coils, I think), so I'm not sure if light would make it from one side to the other..

~~~

Using the device normally (IR bounce mode), seems better IMHO, since the surface of clear ice has a substantial mirror effect.
Ever taken a flash picture, shooting through a car window? :(
I think IR bounce mode will work on white frost or clear ice, equally well.
It has the added advantage of near zero signal being reflected back to the photo-transistor.
Meaning it will be Off and the collector will be at +5V. However, if anything
reflective gets within 1/2" of the beam LED lens, it's going to bounce back and turn On the ~transistor.
If the transistor was half way on, (2.5V) or all the way on (0.2V),
either state means, start melting the ice.. :)

Edit:
Wow, the price of 10 units just shot up by $8.. I paid $1.93 for 10 units.
http://www.ebay.com/itm/160742387966

Here's a better price..
http://www.ebay.com/itm/280760529898