Refrigerator Thermostat Replacement

[conachair]

Originally Posted by BambooSailor

But I am talking about electronics properly designed for marine environment, otherwise they are almost guaranteed to fail in a very short time.

It's worth bearing in mind here that an electronic thermostat isn't linking some electronics to a boat refrigerator, it's linking some *more* electronics. The 101n0220 controller isn't mechanical, all the worries about salt water corrosion and lightning apply to that as well. I don't know much about the design, but get the impression it isn't designed solely with damp boats in mind.

I had a thermostat go after about 6 years, whether a microprocessor inside a decent ip66 or above enclosure with a little care on regulating power would do in better is unknown, but I'm willing to give it a try. The mechanical thermostat is just a switch with an adjustable spring, no way of controlling the cycling and speed dynamically based on actual data. (maybe 6 years was just bad luck with that thermostat, dunno.)
A processor device will certainly provide continuous data about cycle times and temperatures which would be very time consuming and difficult to get accurately otherwise. With units fairly widely available and very inexpensive, easy to have a few spares onboard. The sensors hopefully will be fairly robust, they might not like lightning either though. But they cost so little you can buy them by the handful.
The belt and braces approach would be to fit both and leave the mechanical one disconnected.

Sent from my SGP312 using Tapatalk

[conachair]

Some more lovely data

So yesterday the pump on / off temps were 5deg C & 3deg C measured near the bottom of the box.

Average temp was 4.34deg C, pump ran for 25.4% of the time.

Today the on/off temps were 4.5deg C & 3.5 deg C.

Average temp was 4.14deg C with the pump running 30.1%of the time.

Yesterdays data included nighttime which might have skewed the data a bit.

[Richard Kollmann]

If you are satisfied with box temperatures then 30% compressor cycle times would indicate you are in refrigerator’s sweet performance zone. The question now is there room for more improvement in reducing daily amp-hrs consumed? And is this control method an improvement over a mechanical thermostat?

I use data logger to run a system as the manufacturer designed it first to establish a base line performance, and then evaluate modification performance changes. I installed a logger in a boat’s Adler Barbour system’s freezer in transit passing through south Florida Monday with instructions over the next two weeks to run tests increasing compressor speed, Turning CU200 water pump off then on, Eliminating thermostat using a jumper to find this system’s low temperature equilibrium based on insulation. Boat owner will mail logger back next month and I will copy you in on my analysis. This boats original problem was compressor ran all the time and box temperature stayed at 26 degrees F. In order to preserve frozen food quality +26 degrees is not recommended for a freezer.

[belizesailor]

Interesting to see in Conachair's plot just how much evaporator plate temp varies vs box temp. I knew this in general, but helps enforce that knwoledge by seeing it so clearly in a graph.

Given that most cruisers I think just set and forget their t'stats, I don't think an electronic t'stat would be that beneficial to most cruisers, but the data logging has obvious potential in design and trouble shooting. Would be nice to have a data logger, built of inexpensive components, with flexible software for those purposes.

With that in mind, Conachair, could you share the details of you project?

(I also have the ulterior motive of use in beer brewing...where an electronic, software controlled t'stat has definate application!)

[conachair]

Originally Posted by Richard Kollmann

The question now is there room for more improvement in reducing daily amp-hrs consumed? And is this control method an improvement over a mechanical thermostat?

Very probably room for improvement in power draw, the arduino board is pulling nearly 0.25a so something wrong there, it should be much lower. Lots to learn, which is half the fun
I'll try with a higher pump speed next as well and see what that does. To measure current and voltage is easy, just add a current sensor, again very cheap off ebay. All of which will probably happen at some point but the real jobs list is suffering badly as this is much more fun
Before heading off the well beaten track again it would probably be wise to have a mechanical spare just in case.
As to improving performance over the mechanical unit-- if the on/off temps were known it would be simple to replicate, then play around with the values to see if they can be bettered. But I've no idea what they are for the different setting or high accurate they are from unit to unit.

[conachair]

Originally Posted by belizesailor

With that in mind, Conachair, could you share the details of you project?

The kit is actually quite basic, all based around an arduino ..
Arduino Mega 2560: Components & Supplies | eBay

In a realy stupid moment the other day I blew an Uno R3 board, they draw less and cost less that UK£10.00 so I'll go back to that probably and save the mega for other things.

Sensors are these...

Digital Temperature Temp Sensor Thermal Probe DS18B20 For Thermometer Waterproof | eBay

Then the pump is controlled just with a transistor being triggered by the arduino. Programming isn't too difficult if you have done anything like that before and there's loads of resources and forums on the net to help.

I'm getting little carried away with thinking about adding a decent screen and making something multifunction - this guy has a nice setup and a nice screen..
Arduino Thermostat
Combining fridge monitor/thermostat with a gps repeater and monitoring the temps of stuffing box/alternator/exhaust watertrap would be nice.

But one thing at a time....

[conachair]

And some sort of card read/write as well..

http://www.ebay.com/bhp/arduino-sd-card-shield

Sent from my SGP312 using Tapatalk

[belizesailor]

Thanks much. I used to be heavy technical (decades ago, but the engineering mind set is still useful) so I can probably sort out the details...now I just need to allocate the time.

[conachair]

Well that's interesting.. 1st image is with the pump running at 2000rpm. 2nd at 2500rpm. The evaporator plate temp varies quite different than at a 2000rpm pump.
Running at 2500rpm the pump ran for 23% of the time, or with the extra current draw a fraction less efficient, which could well be nothing more than variability elsewhere.

[Richard Kollmann]

You know the temperature control is constant between test 1 and 2 How about the amp-hrs consumed?

[conachair]

Originally Posted by Richard Kollmann

You know the temperature control is constant between test 1 and 2 How about the amp-hrs consumed?

From a clamp meter at 2,000rpm the pump was pulling 2.3A and at 2,500rpm it was 3.1A.
Lots of assumptions going on, same voltage, same ambient temp etc so there may well be inaccuracies creeping in. And there's still no fan on the compressor, though England in April is hardly tropical.
But pretty graphs all the same
Next stop a little fan in the box.

Sent from my SGP312 using Tapatalk

[belizesailor]

Very insightful I think to see this in graph form. Great potential for education/design/troubleshooting use. Be interesting to see this same data with and without a fan (and decent ventilation) in the tropics.


(...and I've been researching Adruino, way cool, I see a board or three in my future. So many applications on the boat and off: fridge monitoring, engine monitoring, beer brewing, Jeep engine diagnostics... Haven't written a line of code in about 20 years so be fun to get back into that too.)

[Richard Kollmann]

What you should be looking for with any change to refrigeration you make is will it maintain your desired box temperature at the lowest daily amp-hrs consumed? For your results to have any meaning start with a base line performance before any change is made from manufacturers design.

I maintain a test room temperature of 80+- one degree F. With mobile units in a boat I record average ambient temp during base line test. Later tests when ambient temps are different from base line I add 4% to amp-hrs for each degree F increase over base line temp for each Degree F warmer and subtract 2% for each degree colder. This may not be perfect but it works for me.

Trying to improve reliable and performance with electronic thermostats is possible, maybe, someday but mechanical thermostats will still be more friendly in marine applications.

[senormechanico]

Originally Posted by conachair

From a clamp meter at 2,000rpm the pump was pulling 2.3A and at 2,500rpm it was 3.1A.
Lots of assumptions going on, same voltage, same ambient temp etc so there may well be inaccuracies creeping in. And there's still no fan on the compressor, though England in April is hardly tropical.
But pretty graphs all the same
Next stop a little fan in the box.

Sent from my SGP312 using Tapatalk

At first with a full load of food, our box had a huge temp difference between top and bottom (frozen on the bottom and 45 deg f. on top).
Compressor was running at highest speed, and getting hot.

After installing a 2" computer fan running 24/7 in our box, it made a huge difference in run time.
The temp was only about 2 deg f. difference top to bottom. Compressor runs at lowest speed now.

Happy Admiral, too!

[Zanshin]

Originally Posted by conachair

...
I had an arduino uno r3 and some temp sensors onboard already... overall power consumption should be minimal, the arduinos don't pull much anyway and I think underclocking and some other tweets can be brought down to very low power...

I do a lot of Arduino programming (plus now I have the Atmel studio and program the ATMega chips directly once I have the software debugged) and you can do a lot to reduce power consumption, the display is probably going to use more power than the Uno. While somewhat more complicated than necessary you have given me some good ideas - I've got a number of small One-Wire bus thermometers and some LCD displays as well as extra Arduino Micros aboard which aren't doing anything and I might build a display with controller for the fridge and freezer (a simple poti to control the temperature will do).
Have you written the C code for the controlling circuit yet? What sort of time delays do you consider realistic?

p.s. I just checked my specs, an the DS18S20 thermometers only go to -10C, so they aren't good enough for the freezer, unfortunately. But I have been wanting to log my fridge temperatures for a long time and I am going to put together a Micro today with 2 or 3 temperature sensors and logging temperature data every couple of seconds to an attached SD card - thanks for the impetus!