Re: Refrigerator monitor ideas?

[Marc_F_Hult <MFHult@xxxxxxxxxxxxxxxxxxxxx>]

Bobby, Bobby, Bobby ...

One of the reasons that it's so hard to help you is that you indulge in
hippospeak -- hypothetical hyperbole.

Do you understand that you now claim that your refrigerator uses only (115 x
1/10th time on) = 11.5 watts = 101 kWh/yr ?

There have been several  exhaustive studies on energy use of refrigerators.
For example, the calculator here:

http://www.nwcouncil.org/comments/calculators/ResFrigDecomSavingsCalcv2_0.xls

indicates that a 24-year-old refrigerator (yours is about 30, but the
calculator doesn't go that high) used in a heated space as in your case
averages about 1,540 kWh/yr. This is 15 times your claimed usage.

And that even a very efficient US Department of Energy 2004 rated  top-mount,
no through-door ice uses 532kWh/yr (5 times your claim)?

http://www.energystar.gov/ia/business/bulk_purchasing/bpsavings_calc/CalculatorConsumerResidentialRefrigerator.xls


And that your claim that the duty cycle on your refrigerator is 1:9 (ON:OFF)
is at odds with (eg) the study by the utility in my metro region that
measured in-use performance of about 5,000 actual refrigerators and cit as
typical a 20-min ON, 20 minute OFF --> (1:1) duty cycle ?

Because you already have an Ocelot running 24x7, the power needed to do this
is less than 1/1000th of what the refrigerator uses even without optimization
despite your claim/implication to the contrary. By shutting down power to the
sensors when not measuring, you could likely get it down to 1/100,000th. Use
reality-based data and the figures will fall into place.

So you are off again into your hypothetical world where few care to follow.

But I'll stick with you long enough to follow through with what is intended
to be a useful answer to your question. Conceivably others may benefit even
if you don't.

First, since *you* stated that:
	 "... the first clue is that the freezer temperatures begin
	    to rise, and

         *you* summarized and ended your long original post with:
	  " What I really want to know is when 5 on the freezer control
	    no longer means an average of X temperature in the
	    freezer compartment,

it seems to me that a responsive answer to the question you pose would
include temperature measurement despite your rude and incorrect comments. I
also suggest that you measure compressor "ON" time and calculate as fraction
total time, i.e,  duty cycle.

So consider this:

1) Measure the temperature of the freezer compartment (as per your summary)
2) Measure the temperature of the coil
3) Determine fraction "ON" time and duty cycle by detecting power to the
    compressor in one or more of several possible ways.
4) Use your Ocelot (and therefore analog Lm34/5-style sensors, not 1-wire)
   for the temperature and duty cycle measurements.
5) Ask your wife to stop fiddling with the thermostat. If she _has_ to turn
	it up, its probably time to clean/unclog. End of experiment.
6) Make your measurements between 3AM and the earliest anyone uses the
   fridge each day (say 7AM) to avoid perturbations. The problem
    presumably takes at least several days to develop.

You should be able to figger out how to program your Ocelot in CMAX to do
this. If not, ask in the ADI user group.

In brief:

1) manually monitor the temperature for a few of days to get the general
range of values for your calculations. This will depend in part on how clean
the coils/filter/clog/ is at that time.

2) Program the Ocelot to compute a 'running average' of n measurements by
measuring a new T_FreezerNow and computing:

T_NewFreezerAve = (T_FreezerNow + (n-1)*T_FreezerRunAve)/n
T_FreezerRunAve = T_NewFreezerAve

3)Determine the duty cycle and(or) percent "ON" time. A study by my utility
of 5000 in-use household fridges states that this is typically 20 minutes on,
20 min off, --> 50% duty cycle = 0.50 "ON" fraction.

Trigger whatever you want to trigger (Gong, LED, email, mousetrap) when duty
cycle and average temp indicate a problem,

eg:  T_NewFreezerAve > 5 degreesF AND Fraction_ON > 0.65

Additionally, if you measure the temperature of the coil that is supposed to
be defrosted by melting, if it stays below 32F  for an extended period of
time, (say 48 hours) it is frosted. That's another condition that you can use
to improve your notification algorithm.

Past experience suggests that you will say that you can't/don't/won't know
how to do this because of umpteen convoluted reasons including, as you wrote
earlier, that the dogs open the refrigerator!

So don't.

Just like you have umpteen convoluted and specious reasons why HA PC's
running 24x7 don't work dependably despite not actually having one, and
despite extended testimony by those that have been successful that they are
dependable.


... Marc
Marc_F_Hult
www.ECOntrol.org