The Cyber-Spy.Com Usenet Archive Feeds Directly
From The Open And Publicly Available Newsgroup
This Group And Thousands Of Others Are Available
On Most IS NNTP News Servers On Port 119.
Cyber-Spy.Com Is NOT Responsible For Any Topic,
Opinions Or Content Posted To This Or Any Other
Newsgroup. This Web Archive Of The Newsgroup And
Posts Are For Informational Purposes Only.
From: email@example.com (john jardine)
Subject: Re: Heated seat controller, open or closed loop temp.control
Date: 1 Nov 2002 12:27:23 -0800
References: <firstname.lastname@example.org> <email@example.com> <4Diw9.11247$I6.firstname.lastname@example.org>
NNTP-Posting-Date: 1 Nov 2002 20:27:23 GMT
"Jeroen" wrote in message news:<4Diw9.11247$I6.email@example.com>...
> > That's right but ... Essentially, what I'm suggesting, is that
> > sticking a probe somewhere in the seat and allowing the setup to cycle
> > within an undefined temperature window without PID control is
> > actually still more complicated (even less comfortable!)than having a
> > completely open loop setup.
> Why is there an undefined temperature window? I will use a uC with a
> tempsensor. When the temp is lower than the setpoint-1 cdeg the heater is
> turned on and when the temp reaches setpoint+1cdeg it is turned off again.
> The ripple in temp. will be defined by the amount of hysteresis I define
> (which can actually be quite small, nothing wears out by rapid cycling). And
> I will check the temp and act only once a second, any faster is not
With 'real world' closed loop control systems, the physical properties
of the non electronic variables, gum up the electronic works and act
as a grade #1 bitch!. It can't be half answered so this here is a bit
Yes ... In an ideal situation, what you say is absolutely correct.
From the electronics POV, they will switch power on when then sensor
says I'm too cold and switch power off when the sensor says I'm too
hot, and would be happy to control the seat temperature with even
+/- 0.01degC heating/cooling limits. Simple power On, power OFF,
couple of resistors to set the hi/lo limit trips, a comparitor +Fet,
et voila, no problem!.
It's easy to do, so build a unit up and try it!. You will actually
find the seat temperature cycling at much,much greater limits than
what it was designed for. (the seat temp' would, more likely run
rampantly out of control).
There is a very real, very practical, problem lurking in the depths of
this closed loop system you have just created and concerns **how
quickly** that seat temperature sensor can notice any temperature
changes. Get a mental model or 'handle' on why this 'speed of
response' can be a problem and you will know the very foundation stone
that ALL electronic/electrical/mechanical, control sytem and feedback
theory, sits on. (The 'delay' in mechanical systems is usually called
response time, in purely electronic systems it can be called 'phase
Take a worst case example. Say the temp sensor is one of the LM???,
T05, linear two wire types and you insert it in the seat fabric at the
body point where the most comfort is wanted. Assume that in this
position the sensor is say 10mm away from the surface of the nearest
embedded heater wire. The PIC (say) will digitise the sensor's
voltage, compare it against the hi/lo limits and switch power on and
off as needed.
Ok, let's go ...
 Turn ignition ON and power up the seat heater control system.
 Sensor says too cold so PIC switches power to the heater.
 Seat warms up
 Sensor at some time later will say "I'm too hot".
 Power switched off and seat cools
 Sensor at some time later will say "I'm too cold".
 Power switched on and seat heats again
 Blindly repeat  to .
The killer, is in lines  and . The "some time later" could be 3
seconds or even a few minutes. It all depends on how fast the heat can
get through the local path from the heater wire to the sensor body, or
(on cooling how fast the heat can dissipate from the sensor itself and
region around the sensor.
[nb, If you think about it you'll also notice that these two 'time
constants' can also be widely different just like a refrigerator motor
cycles only occasionally.]
So ... The heater's just been switched on. The heat manages to
percolate through to the sensor in say 10 seconds and the heater is
then switched off as the sensor has now reached the upper limit. The
problem now, is that the sensor has happily said that it's temp' is at
the upper limit but in actual fact most of the surrounding seat area
material and fabric is at a much higher temperature as it has had 10
seconds worth of heating and is still conveying these higher
temperatures through to the sensor **even though the power has been
This leads to massive temperature 'overshoots' which can only be
reduced (not eliminated!) by making the thermal paths, 'time
constants' shorter. E.g using a very low mass sensor attached directly
to the resistance wire in the heater cabling. If this is done, then
when the sensor says "I'm at the higher limit" then it is probably
more nearer to the truth. Although in this case it may have very
little relationship to the actual general seat temperature.
If you think this heat/distance problem through you maybe can see why
a situation can develop where the heater is asked to switch on even
when the seat temp' is already way too high. This is then a system in
To (realistically) get control of this heat lag/lead problem you need
as a first step, to ignore the upper/lower limit action and move to
continuous on line on/off duty cycle control. This is easily done via
the PIC say. Regularly monitor (eg once per second) how much the seat
temperature has changed (+/-) over the last second. Compare with the
setpoint and find how far it is from the required temp' and make
adjustment to the on/off duty cycle such that the heating/cooling
effects slow down as the setpoint is reached.
This method is the (D)erivative or 'rate of change' part of a full
PI(D) control setup.
Getting this kind of control system up and running can be a black art
and there is NO guaranteed method of finding a solution beforehand.
Numerous ways and means can be used to get a 'fix' and it is down to
the designer to dream up something suitable.
Ignore the books and software that tell you they will provide
solutions to these problems(they cost money and are mostly bollocks)
as they require a full mathematical description (transfer function) of
the the seat/body heatflow dynamics before they can calculate an
optimum control setup. In your case you would need the convective and
conductive heat xfer data for the laminar seat materials and a
clothed human body and then the application of a finite element 3D
solution using the Fourier heat transfer equation.
Setup these one off systems as you find them!. As long as you are
aware of the pitfalls (the basics) beforehand they can be quite fun
from a progging POV.
Go Back To The Cyber-Spy.Com
Usenet Web Archive Index Of
The sci.electronics.design Newsgroup