--===============0172514961==
Content-Type: text/plain; charset=GB2312
Content-Transfer-Encoding: quoted-printable
X-MIME-Autoconverted: from 8bit to quoted-printable by pch.mit.edu id n1L3JsQu028138

Hi,

Good temperature control is not quite as easy as you make it out to
be. The biggest problem is compensating for the time lag between your
heater and your temperature sensor, while still achieving adequate
rejection of external disturbances, system stability and lack of
overshoot, and adequate response time. Also, whatever you place in the
chamber will alter both the time lag and the time constant of your
system.

Usually the temperature sensor's response time is negligible compared
to the time it takes for the heater to affect the sensor. Even if the
overall thermal time constant of the system is relatively low, it can
still take a while for convection to carry heated air from the heater
to the sensor. You can often help by adding a fan, but then you may be
controlling the air temperature but not the temperature of the walls
of the chamber or of objects within it very well.

The system itself is stable (i.e., when it is open-loop and not under
feedback control) but you can certainly make it unstable with the
addition of the feedback loop. This is because it is not really a
first-order system but actually an infinite-order system because of
the time delay from heater to sensor. So, stability is still a concern
if you use feedback.

Usually PID works fairly well for temperature control, but the optimal
tuning will depend on what you have in the chamber. Change that and
you may significantly alter the response. Sometimes the chamber can
try to perform some "system identification" as it is heating up to
estimate how much different than expected the properties are, and then
adjust its own PID terms to better suit the actual present situation.
Of course, this is actually putting a feedback loop around a feedback
loop and has its own stability concerns if not done properly.

As for analog versus digital, I would suggest going digital. It makes
it a lot easier to adjust the PID coefficients. It also would enable
adaptive control if you desire to explore that in the future. You
probably do not need a D to A for the output - just a PWM generator.

Sean




On Fri, Feb 20, 2009 at 7:52 PM, gardenyu <gardenyu2004@hotmail.com> wrot=
e:
>
>
>  I started looking at
> temperature control inside a thermal chamber.  That tank is actually a
> first-order system with some possible disturbances from different input=
 and
> ambient temperatures. Also a temperature sensor can be considered to ha=
ve a
> linear, very small time constant (or: high bandwidth). Thus, the system=
 should
> be stable by its nature. Since not much compensation might be needed, m=
any
> people just use PID to tune up the time response of a temperature contr=
ol system.
>
>
>
>   Recently I read a book
> written by Robert Pease, called "analog circuits, world class
> designs". The front chapters talked about how to realize feedback contr=
ol
> loops by using a series of opamps, caps and resistors. It makes so much=
 sense
> so that I found some link between all the control courses I took to the=
 real
> world controls.
>
>
>
>   As to the PID control, I
> found some modern designs simply use A/D, D/A, and a microprocessor, an=
d let
> the software guys to do the math (which takes time since microprocessor=
 cannot
> do calculus that easy, also discrete control involves different math). =
There is
> also a more analog, or old-fashioned way, like Bob did in the book, was=
 just to
> use many discrete semiconductors. (Although I guess a micro is always n=
eeded
> somewhere)
>
>
>
>   I wonder which way might
> be more appropriate? In an analog way, certainly a hardware guy can tak=
e
> control of most things; he can also get an easier and more direct conti=
nuous
> system model. But a digital way seems like  easy-configurable and
> adjustable, but the credit won't be mostly mine anymore :)
>
> _________________________________________________________________
> MSN=D0=C2=B0=E6=D4=DA=CF=DF=C9=E7=BD=BB=CD=F8=C2=E7=A3=AC=D3=EB=C4=FA=B5=
=C4MSN=BA=C3=D3=D1=B8=FC=C8=AB=C3=E6=BB=A5=B6=AF=A3=A1
> http://im.live.cn/msn9/
> --
> http://www.piclist.com PIC/SX FAQ & list archive
> View/change your membership options at
> http://mailman.mit.edu/mailman/listinfo/piclist
>
>


--===============0172514961==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Disposition: inline
Content-Transfer-Encoding: 7bit

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist

--===============0172514961==--