Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 5/3/83; site heurikon.UUCP
Path: utzoo!watmath!clyde!floyd!harpo!seismo!uwvax!heurikon!jeff
From: jeff@heurikon.UUCP
Newsgroups: net.misc,net.physics
Subject: Re: Why don't thermostats work?
Message-ID: <194@heurikon.UUCP>
Date: Sat, 4-Feb-84 03:28:07 EST
Article-I.D.: heurikon.194
Posted: Sat Feb  4 03:28:07 1984
Date-Received: Wed, 8-Feb-84 08:24:30 EST
References: <877@ihuxl.UUCP>
Organization: Heurikon Corp., Madison WI
Lines: 55

I'm not a thermostat expert, and it was a long time ago that I took
control theory, but I'll take a stab at answering your questions:

1)  The bi-metal strip thermostats have something called an "anticipator"
    in them.  It's a little heater element which runs off the 250 ma or so
    of current used by the furnace relay.  When the thermostat is calling
    for heat, the heater element warms up the strip, thus anticipating
    the warming of the room.  If this were not done, there would be wild
    changes in temperature - it would get much warmer in the room before
    the thermostat shut off.  You'd be uncomfortable because of the wider
    temperature swings.  (Look inside your thermostat and you should find
    a little calibration screw or lever, labeled in milliamps.)

(For you Californians who - through lack of use - don't know what a
thermostat is, it's that little box with levers under your doorbell.
Sorry, I just *couldn't resist that.)

2)  A closed loop control system (such as a room-thermostat-furnace-room)
    regulates the "process" by creating an error term.  A thermostat
    is a relatively dumb control unit.  There must be an error in room
    temperature in order for the thermostat to call for heat.  The
    thermostat does not know how much heat is leaving the room, it only
    knows that there is an error in desired the temperature.  The average
    error will be proportional to the rate of heat loss in the room.  So,
    on colder days, the average room temperature will be lower.

    Look at it this way:  If a (dumb) thermostat *was* able to get the
    room temperature *exact*, the error would be zero.  If the error were
    zero, the thermostat would not call for heat.  But if it doesn't ask
    for heat, the room temperature will fall.  So the error *can't* be zero.
    
    For the system to reduce the error to zero, there must be additional
    intelligence.  The system must be able to compute the heat loss.  When
    you manually increase the setting on a cold day, you are adding that
    "intelligence" by adding a constant to the error term to compensate for
    the extra rate of heat loss.  There are some uproc based thermostats on
    the market.  I saw one (Heath?) which is smart enough to realize that
    the error term isn't going to zero and increase its call for heat.

    In control theory, the extra feedback term which is needed to completely
    reduce the error to zero is one which integrates the error signal over
    time.  So, to regulate the error to zero, you need two feedback terms:
    one which is proportional to the error itself, and the other which is
    the integral of the residual error.  Thermostats lack the latter.

Your idea of using a simple knob instead of a thermostat would allow you to
get the exact desired temperature as long as the room's heat loss rate was
a constant.   Simply raising the thermostat setting does the same thing
and also compensates *some* for changes in the loss rate.

Now, is there a control theory expert out there who can confirm any of this?
-- 
/"""\	Jeffrey Mattox, Heurikon Corp, Madison, WI
|O.O|	{harpo, hao, philabs}!seismo!uwvax!heurikon!jeff  (news & mail)
\_=_/				     ihnp4!heurikon!jeff  (mail - fast)