Student Question on using a PID as a Dimmer to Control a Light Bulb!!! This is a fun one!!...

Here are 2 pages for dimmer circuits that may of help:

http://www.epanorama.net/documents/lights/lightdimmer.html

http://www.discovercircuits.com/L/lite-dimmer.htm

deriso@gmail.com wrote:

You can purchase PID controllers with continuous analog outputs for
less than $200, and much less for used ones on eBay.  you convert the
analog output to phase control (lamp dimmer type) with a solid state
relay made for that purpose.

Here is a page from the Mouser catalog that lists some, including
suffix "C" that are controlled by 0 to 10 volts.
http://www.mouser.com/catalog/625/1309.pdf
You may find something similar for about half that, rated for the low
current you need, if you shop around.

So, I would:

1) find a PID that has a continuous analog output (0-10v dc)
     (can you explain what that is a little more :) ...
     does it output a voltage from 0-10v dc according to what it thinks
my lamp needs to
     heat to?
     ex. set temp is 200deg C
          current object temp is 150deg C

          the PID will give a voltage output of lets say 8.789 volts
(my guess) and decrease
          the voltage to say 5.565 as the temp nears the setpoint (or
however the graph
          goes...). The idea is that it gives its approximation of how
far my temperature is
          from my set temp in terms of an analog (dc?) voltage. i might
be very lost here...

2) use an SSR that converts the analog voltage output to a phase
control (dont really
    know what that is either... but it looks expensive) What is a phase
control??

3) take that phase output and put it into some kind of dimmer circuit
like this one
http://www.epanorama.net/documents/lights/lightdimmer.html#advanced

What goes to the lamp? the phase output? is that a voltage potential?
please explain...

AND THANK YOU!!!!!!!!!!!!!!!!

deriso@gmail.com wrote:

Yes, the controller has to have either a voltage output or a current
output (that can be converted to a voltage with a load resistor) that
is compatible with the phase power controller relay.  It also has to
accept the type of thermocouple you have or can get.


Yes.  0% output is zero volts, and 100% output is 10 volts.


The PID algorithm compares the present value of the input
(temperature) with the stored setpoint value and computes the three
terms (proportional, integral and derivative) and combines the three
factors to produce the output.  You give the controller a model of the
response of the system, so it uses the correct amount of each factor
to push the measured value toward setpoint without overshooting or
oscillating.  If you want to read a nontechnical tutorial about how to
tune a PID controller and how the three factors act, one I wrote is
available at:
http://www.tcnj.edu/~rgraham/PID/popelish.html


It is a solid state switch that holds back the line voltage for some
part of each half line cycle and then turns on and conducts for the
rest of the half cycle.  The fraction of each half cycle that line
voltage is applied to the load is controlled by the DC input voltage.
  So, if the power comes from a 60 Hz line frequency, you get 120
power pulses per second, which is fast enough for the lamp to have
very little flicker.


If you get one of the phase controlled relays, it controls the power
to the lamp, directly.  you just wire its two output terminals in a
break in the lamp circuit.

That link shows you approximately what is inside the phase control
relay, if you would rather buy than build.

deriso@gmail.com wrote:


Often with temperature control problems you can get pretty good control
by driving your heater with a  very slow PWM signal -- I've personally
seen one with a period of two seconds, and periods over a minute
wouldn't surprise me.  The two limiting factors are what will it do to
the lamp to be turned on and off continually, and how rapidly will your
plant (the thing you're heating) respond?  You want a speed that will
keep the lamp happy yet not induce too much variation in the plant --
and if the slowest speed that will not damage the lamp is faster than
the highest speed you can tolerate in your plant, then you need to find
another method.

You can implement the slow PWM method by buying an off-the-shelf
temperature controller -- this is where my knowledge and experience
stops helping.  I know that nearly anyone who makes industrial control
modules makes them, and I know that Omega is a good source for
onesie-twosie lab-type controllers.  Expect to pay $100 to $200, but I
think you could get one that would handle a J-type input and 110V output
off the shelf.

If the slow PWM approach isn't going to work for you then you may need
to drive the lamp with a variable voltage.  You can still use fairly
slow PWM here, but drive it fast enough to keep the lamp on.  At 35W I
expect that it's a 12V lamp -- if this is the case then you should be
able to take a 12V supply and PWM a MOSFET driver at speeds as low as
50Hz.  The controller could be nearly anything, depending on what you're
good at implementing -- in this day and age a Basic Stamp controller
from Parallax would probably be best.  You'd need input conditioning for
the thermocouple, and a MOSFET output, but there's a good chance you
could put the whole thing together with parts from Radio Shack.  The
Basic Stamps work with BASIC, so they'll be easy to code -- you should
be able to port the code in my article
(http://www.wescottdesign.com/articles/pidwophd.html ) pretty easily.

No matter what method you use to drive the lamp, you are correct that a
PID controller should work.  The whole point of integral action in a
controller is to hold the average error to zero.  This does mean that
for every undershoot there's an overshoot, but once the temperature has
stabilized that won't be a big issue.  If you can't have big over- or
under-shoots, let us know -- someone will help you out with at least one
solution.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google?  See http://cfaj.freeshell.org/google/

I just wrote a great response, and lost it. Damn.

ANYWAY...

I am most intreseted in the MOSFET. I did a little research and found:

"Pronounced MAWS-feht. Acronym for metal-oxide semiconductor
field-effect transistor, a common type of transistor in which charge
carriers, such as electrons, flow along channels. The width of the
channel, which determines how well the device conducts, is controlled
by an electrode called the gate, separated from channel by a thin layer
of oxide insulation. The insulation keeps current from flowing between
the gate and channel." taken from
http://www.webopedia.com/TERM/M/MOSFET.html

Is the MOSFET like a PWM, except continous. In other words, instead of
keeping the same voltage and releasing it on intervals to create a
wattage, does the MOSFET just limit the voltage without inturrepting
it?

Is this more practical?

And as for eliminating over and undershoots, I would just like a
gradual continuous increase until the setpoint is reached and then
corresponding increases if the object drops in temperature. Any ideas?

One again, Thanks Tim!!!

deriso@gmail.com wrote:

It can be, but it's inefficient. A switch has low resistance, and
therefore low power dissipation when current flows. It has no
dissipation in its other state when no current flows. Both MOSFETs and
bipolar transistors cam be used as switches, and both can be used for
proportional control.

Consider what happens when a switch reduces a constant-resistance load*
to quarter power. The switch is on for a quarter of the time and off for
three quarters. The power in the switch is small. Now let's do that as
you suggested. The effective resistance of the control transistor is set
equal to that of the load, so that half the voltage appears across each.
The current flows through both, so that the power to the load and the
power dissipated in the transistor are the same. Half the power is
wasted, but worse, the transistor needs to get rid of a lot of heat.

With cycling on-off control, you'll be able to run your heater from AC
using a triac. With proportional control of current, you'll need a DC
power supply.


Good control systems needn't overshoot. The more rapidly they respond to
changes, the harder it becomes to keep them from overshooting. Tim has a
new book on the subject. I haven't seen it yet, but it may be a tad too
technical to be useful to you.

Your choice of Type-K thermocouple may prove troublesome. Any
thermocouple needs cold-junction compensation; it measures not
temperature, but temperature /difference/ between two points. Roughly, a
Type-K delivers about .05 volts at 2200F (I can look it up if you need
me to) and the temperatures you want to measure are much lower. To
discern one degree, you will need to measure (and distinguish from
noise) a difference of .000022 V (22 microvolts).

Thermistors are non-linear, but might be useful with proper calibration.
I've had good success with ordinary silicon diodes -- the forward drop
is a function of temperature -- but careful design is needed and
replacement usually requires recalibration in high-precision applications.

Jerry
_______________________________________________________________
* The resistance of a light bulb is changes with the power level.
--
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Jerry Avins wrote:


-- snip --


-- snip again --

I generally recommend that book every chance that I get, but I was
refraining for just that reason.  If you're motivated, and if you got
through your first few weeks of calculus, it may be of help.  The title
is "Applied Control Theory for Embedded Systems", the publisher is
Elsevier, the author is yours truly, the intended audience is embedded
hardware/software engineers who only learned digital circuits or none at
all, and the book doesn't assume that you know anything about control
theory except that you think you need to use it.  I did keep the theory
to what I felt was a necessary minimum -- but there's still a good slug
of mathematics in there to be gotten through.

Here's Elsevier's page on the book:
http://www.elsevier.com/wps/find/bookdescription.cws_home/707797/description#description

There's a blurb about it on my website, including this link:
http://www.powells.com/partner/30696/s?kw=Wescott+Tim
which will let you buy the book from my favorite bookstore.  Of course
Amazon has it as well, and your local bookstore can probably get it.

As I mentioned in a recent posting, the book doesn't directly answer the
question of reducing overshoot, although it gives you a pretty complete
toolbox to do so.  I've already marked up my development copy, so should
the book ever get to a 2nd edition there will be something in there
then, and something will be popping up on my website soon.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google?  See http://cfaj.freeshell.org/google/

deriso@gmail.com wrote:
(snip)

The response to error of the control system is set with the three
constants that configure how much P, I and D terms.  You can set them
to provide the fastest possible response, but with decaying cycles of
over and undershoot, or an extremely slow creep toward setpoint, or
anything in between.  Making the decisions about how to set the three
gains is called loop tuning.

I just wrote a great response, and lost it. Damn.

ANYWAY...

I am most intreseted in the MOSFET. I did a little research and found:

"Pronounced MAWS-feht. Acronym for metal-oxide semiconductor
field-effect transistor, a common type of transistor in which charge
carriers, such as electrons, flow along channels. The width of the
channel, which determines how well the device conducts, is controlled
by an electrode called the gate, separated from channel by a thin layer
of oxide insulation. The insulation keeps current from flowing between
the gate and channel." taken from
http://www.webopedia.com/TERM/M/MOSFET.html

Is the MOSFET like a PWM, except continous. In other words, instead of
keeping the same voltage and releasing it on intervals to create a
wattage, does the MOSFET just limit the voltage without inturrepting
it?

Is this more practical?

And as for eliminating over and undershoots, I would just like a
gradual continuous increase until the setpoint is reached and then
corresponding increases if the object drops in temperature. Any ideas?

One again, Thanks Tim!!!

Tim,

I have a few questions the PWM signal sounds like it may burn out most
bulbs with all the flickering if its less than standard 60hz. Then
again I wouldnt know. However, you mentioned a variable voltage method.
Does the variable voltage method keep the bulb on the whole time and
just limit the voltage instead of the wattage? How does current play a
role in all of this??? The PWM sounds good, but if you drive it fast,
doesnt that increase the wattage? From what I understand, its the same
voltage just being averaged over a time interval to create a specified
wattage.. am i correct?


Is there a -- I've personally
seen one with a period of two seconds, and periods over a minute
wouldn't surprise me.  The two limiting factors are what will it do to
the lamp to be turned on and off continually, and how rapidly will your
plant (the thing you're heating) respond?  You want a speed that will
keep the lamp happy yet not induce too much variation in the plant --
and if the slowest speed that will not damage the lamp is faster than
the highest speed you can tolerate in your plant, then you need to find
another method.

You can implement the slow PWM method by buying an off-the-shelf
temperature controller -- this is where my knowledge and experience
stops helping.  I know that nearly anyone who makes industrial control
modules makes them, and I know that Omega is a good source for
onesie-twosie lab-type controllers.  Expect to pay $100 to $200, but I
think you could get one that would handle a J-type input and 110V
output
off the shelf.

I looked up MOSFET, and found the following:

"Pronounced MAWS-feht. Acronym for metal-oxide semiconductor
field-effect transistor, a common type of transistor in which charge
carriers, such as electrons, flow along channels. The width of the
channel, which determines how well the device conducts, is controlled
by an electrode called the gate, separated from channel by a thin layer
of oxide insulation. The insulation keeps current from flowing between
the gate and channel." Taken from
http://www.webopedia.com/TERM/M/MOSFET.html

Is a MOSFET like a variable voltage controller that can increase or
decrease the voltage supplied to a device on a continuous level? or are
ther on-and-offs like the PWM.

Also, I cacan't have big over- or under-shoots. I want a gradual
increase and then a stop at the set temperature and then an increase
when the object temp decreases.

Thanks again Tim! This project is nuts.

deriso@gmail.com wrote:

The current into the bulb will depend on the voltage and the filament
temperature.  Cutting the voltage will reduce the current, although not
as much on average as would be the case for a resistor.

You could just use a honking big DC amplifier to drive your bulb (I
think this is what was suggested to you in the first place).  The bulb
would get dim, as desired.

The PWM drive that I was suggesting was one where you drive a switch at
50-500Hz.  The voltage to the bulb will be a series of rectangular
pulses.  During a pulse the current will be high, but on average the
power to the bulb will diminish as the duty cycle goes down.  You
wouldn't be able to drive a 'specified' power, but it would be a
controllable one.

Yes and yes.  A MOSFET is a transistor.  You can either choose to drive
it in a way that regulates the current to the bulb in a continuous
manner (assuming a DC supply to the bulb) or you can choose to turn it
on hard, then off hard.  If you choose the continuous regulation method
then you will have to dissipate significant power in the MOSFET --
choosing PWM allows you to significantly simplify both your circuitry
and your heat sinking.

Don't we all.

If you can get by with moderate over- and under-shoots then a 'straight'
PID controller should work for you.  You can make the increases and
decreases more gradual by increasing your derivative gain, but you can't
completely stop the overshoot, and the system as a whole will settle slowly.

You _can_ play a trick with a PID controller by rearranging it to some
extent.  You can take the difference between your actual and target
temperature and run it through a block to generate a desired temperature
ramp.  Then you take the derivative of the actual temperature and
compare this to your desired ramp, and apply it to a PI controller.  If
you limit magnitude of your desired temperature ramp then you'll
significantly limit the overshoot.


                 .---------.              .------------.
          _      |      -- |       _      |            |
  Set    / \     |     /   |      / \     |       ki   |  Drive
  ----->| + |--->|    /    |---->| + |--->| kp + ----- |------->
  Temp   \_/     |   /     |      \_/     |      z - 1 |
          A      | --      |       A      |            |
          |      '---------'       |      '------------'
          |                        |
          |                        |
          |                        |    .-------.
          |                        |    |       |
          |                        |    | z - 1 |         Plant
          |                        '----| ----- |<------o------
          |                             |   z   |       | Temp
          |                             |       |       |
          |                             '-------'       |
          |                                             |
          '---------------------------------------------'
(created by AACircuit v1.28.6 beta 04/19/05 www.tech-chat.de)

I suggest that before you try this, you first get a regular PID loop
running, and play with it until it works except for overshoot.  Once you
have that going, then investigate making this one work.

Note that this loop doesn't absolve you from anti-windup limiting in
your integrator -- that's still necessary to keep large temperature
excursions in the plant from making the loop misbehave, but the
limited-ramp method does decouple the ramp-up behavior from the
integrator limiting.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google?  See http://cfaj.freeshell.org/google/

deriso@gmail.com wrote:

I don't like being the bringer of bad news, but it seems often to be my
role here. This thread is no exception.

You don't want to run a halogen bulb on a dimmer of any kind. It's not
bad for the bulb to run it very dim, but a little bit of dimming will
burn it out quickly. A halogen-cycle lamp is simple in concept, and not
even tricky when it's understood, but it's subtle.

Light bulbs burn out because the substance of the filament sublimates.
Tungsten has a very low vapor pressure, but it isn't zero. At
incandescent temperatures, the vapor pressure increases somewhere
between the fifth and seventh power of absolute temperature. You can see
the darkening of the inside of an ordinary bulb from deposited tungsten
vapor. The bulbs are made large to spread the tungsten over a large area
and thus maintain transparency. As tungsten sublimates, some parts of
the filament become thinner than the rest, thus becoming hotter and
losing material faster. Then poof!

Tungsten-halogen bulbs take a different tack. The envelope is made very
small, so that it gets very hot. (Originally, only fused quarts could
withstand the temperature, but special glasses can now be used.) The
temperature is high enough for tungsten deposited on the envelope to
combine with the bromine or iodine sealed in. The tungsten halide is a
gas that decomposes at the filament temperature, restoring the tungsten
to to the filament and the halogen to the void. It gets even better. The
decomposition rate is also temperature dependent, so tungsten deposits
preferentially on the thin spots. As a result, the filament can be run
much hotter, which results in higher luminous efficiency.

When the bulb is run very dim, sublimation is nil in any event. When it
is dimmed only a little, the regeneration cycle ceases. Sublimation
continues only slightly abated, and the bulb burns out relatively
quickly. If the bulb has been run at reduced temperature only for a
relatively short time, such as might happen in a brownout, subsequent
full-output operation usually repairs most of the damage.

Conclusion: choose a different bulb or use a slow on-off cycle.

Jerry

--
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Jerry Avins wrote:

Wow Jerry.  That's good to know.  This must by why I told Dave to post
his question to the group.

Am I correct in concluding that if he runs his bulb with some guaranteed
on time (i.e. a modified form of the 'slow PWM' that I suggested in my
reply) that it would work OK?  Is there any way to know just how long
this on time would need to be?

On choosing a different bulb:  It's hard to say without knowing just
what you're trying to heat, but if you switch to a 'normal' 35W
incandescent light bulb it'll be dimmer, but it'll generate more heat.
If you focus it with a reflector that reflects IR (like shiny metal),
and if you don't put an IR absorptive material in between the bulb and
the thing you're heating, it should be just as effective as a halide light.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google?  See http://cfaj.freeshell.org/google/

About the Bulb...

Jerry, wow, in a word. That was a very educational. Thank you, I would
have definatley encountered some problems down the road. So, to find a
replacement, I have sent the following message to several bulb experts:

"I am trying to build a heater that runs off of a 35w halogen bulb (the
small, common ones). To adjust the temperature of the bulb, I am using
a dimmer circuit. However, it has come to my attention that dimming a
halogen bulb will make it burn out much quicker. I like the halogen
bulbs because they get hot quickly and are small and inexpensive to
replace. Is there another kind of bulb that I can use that gets just as
hot, is just as small,is as easy and inexpensive to replace, and can be
dimmed????

I know its an odd question, but I am a student, and any help would be
greatly appreciated!!!"

So, the bulb problem will be handeled shortly!!

deriso@gmail.com wrote:

Well...I didn't know about this dimming halogen bulb thing

So my house is full of halogen bulbs on dimmers...that last just as well
as bulbs not on dimmers

Fulliautomatix wrote:

Then you must dim them deeply, or not for very long. Dimming a little --
running a 120V bulb at temperatures equivalent to 105 or so shortens
their life most drastically, to less than an ordinary incandescent's
approximately 1,000 hours. Do you get the rated hours from your dimmed
bulbs? At 90 volts, any 120V bulb lasts nearly forever. Some low-voltage
tungsten-iodine lamps are rated 10,000 hours and last longer, but most
"halogen" nowadays is bromine and some such bulbs are rated as low as
1,500 hours.

This is from http://en.wikipedia.org/wiki/Incandescent_light_bulb :

The halogen lamp

One invention that addressed the problem of short lamp life was the
halogen lamp, also called the tungsten-halogen lamp, where a tungsten
filament is sealed into a clear "capsule" filled with a halogen gas such
as iodine or bromine. This type of incandescent lamp creates an
equilibrium reaction where the tungsten filament that evaporates when
giving off light is chemically re-deposited at the hot-spots, preventing
the early failure of the lamp. This also allows halogen lamps to be run
at higher temperatures (which would cause unacceptably low lamp
lifetimes in ordinary incandescent lamps) allowing for greater
brightness, whiter color temperature, and efficiency.

Because the lamp must be very hot to create this reaction, the halogen
capsule is often made of fused quartz, instead of ordinary glass which
would soften and flow too much at these temperatures. Thus, halogen
lamps are sometimes called quartz-halogen lamps, or tungsten-halogen
lamps (the filament is tungsten). They were once called quartz iodine
lamps. Modern halogen lamps are made of 'doped' quartz with additives to
reduce the UV output. Halogen lamps with integrated reflectors often
include a transparent UV filter to seal the lamp.

A further development that has added to lamp efficiency is an infrared
coating (IRC). The quartz envelope is coated with a multi-layered
coating which allows visible light to be emitted while reflecting a
portion of the infrared radiation back on to the filament. The result is
that less power is needed to produce an equivalent light output. This
efficiency increase can be as much as 40% when compared to its standard
equivalent.

Perhaps the most significant side effect of using quartz instead of
ordinary glass is that the lamp becomes a source of UV-B light, because
the quartz is transparent to this spectral range and ordinary glass is
not. Quartz halogen lamps are thus used in some scientific instrument as
a UV-B light source. One consequence of this is that it is possible to
get a sunburn from excess exposure to the light of a quartz halogen
lamp. To mitigate the negative effects of UV exposure, some
manufacturers add a coating of UV inhibitors on the capsule that
effectively filters UV radiation. When this is done correctly, a halogen
lamp with UV inhibitors will produce less UV than its standard
incandescent counterpart.

Because the halogen lamp is hot, and poses a danger of fire or burns,
and because of the risk from UV exposure, these lamps are usually
protected by a lens of ordinary glass, which, as noted above, absorbs
most of the UV-B light.

The quartz capsule can be damaged by any oils or residue from
fingerprints. These lamps should be handled without touching the clear
quartz, either by using a clean paper towel or carefully holding the
porcelain base. If the quartz is touched, it must be cleaned with
rubbing alcohol.

The incandescent lamp is still widely used in domestic applications, and
is the basis of most portable lighting, such as table lamps, some car
headlamps and electric flashlights. Halogen lamps have become more
common in auto headlights and domestic situations, particularly where
light is to be concentrated on a particular point. The fluorescent
light, has, however, replaced many applications of the incandescent lamp
with its superior life and energy efficiency. LED lights are beginning
to see increased home and auto use, replacing incandescent lamps. Newer
headlights are often high-intensity discharge lamps, such as metal
halide lamps, which produce purple-tinted light instead of the usual
yellowish color of a standard incandescent bulb.

http://tinyurl.com/5jql7  is also interesting reading.

Jerry
--
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Jerry Avins wrote:


Yes, they are either very dim fully on