wrote:
On Apr 11, 3:07 pm, James Sweet wrote:

There are very few mechanical thermostats which can work that close in
differential. The only one I can recall used a mercury thermometer with
leads embedded in the glass tube at the desired temperature.

Even a design with electronics needs careful design to get that close of a
differential. However I wonder if the spec. needs to be that accurate since
this appears to be an anti freeze system.

The Dallas sensors I refer to have a straight digital output so there's
no fiddly analog design to deal with. Still, that is quite a tight range.

If he provides some more detail about the intended application it will
be easier to give suggestions.


This is the application:
A nearby apple orchid apparently is in danger of having the trees
damaged due to an early Spring frost. Every year around May they have
gotten this box out. Inside is a temperature monitoring control unit
and what appears to be a low power radio transmitter. There is also a
probe, a thermistor I suspect mounted inside a small piece of PVC pipe
which is connected to about 50 feet of shielded cable. This is placed
out in the field. The transmitter pages out via an antenna which is
mounted in the orchard. The whole business is powered by a 12.0 volt
wet cell. This antenna was described to me as "being about ten feet
tall". So I'm assuming its a CB unit of sorts but until I run an SWR
test I won't know for sure. The temperature control unit has a switch
with positions from 26 to 35 degrees F. You select the temperature you
want to alarm at and a contact closure occurs. This powers the
transmitter and sends a signal to the pagers. The critical temperature
is 31 degrees F. This control unit was damaged by water and is
useless. I tried to trouble shoot it but there are house numbers on
the semiconductors and the manufacturer is out of business. I took a
chance and connected the transmitter to a CB antenna and I have
determined that the RF portion of this Rube Golgberg works. However,
since I didn't like the idea of hours of transmitter on time when I
rebuild it I intend to use a one shot timer after the new temperature
control unit so that I'll get a 10 second closure and then power will
be removed from the transmitter. When the temperature warms the timer
will reset and be ready for the next shot. Apparently during this two
to three week period a dip to 31 degrees requires they get some huge
fans out to warm the trees. This is the reason for the page. As much
as I'd like to spend the time designing a system to do this the
customer is a commercial orchard and probably will be willing to spend
a few bucks to just have the situation resolved. I hope this clarifys
things a little. Thanks, Lenny.



I doubt you need accuracy of 1 degree, but you should be able to achieve
something close. Have a look at this part
http://datasheets.maxim-ic.com/en/ds/DS1821.pdf. It's a complete
programmable thermostat on a chip.

If you want to avoid programming, have a look at this one
http://www.national.com/pf/LM/LM34.html 1 degree accuracy, linear output
in degrees F, all you need then is a comparator, some precision
resistors and a precision voltage reference.