Proportional Control Circuit w/ Thermocouple

[Dacr0n]

Hello everyone.

I am trying to build a proportional temperature controlling circuit that is similar to a PID controller, in that I can set a desired temp and the circuit will instantly adjust the heater to the desired temp.


Now the heater is going to be controlled by a PWM circuit powering a MOSFET.

The PWM circuit I have so far is controlled by adjusting a potentiometer to adjust the pulse width.


So this proportional controller is going to have to control the PWM circuit some how. Unless you have a better idea.


Any ideas about the whole project?

Ohh yeah and the range is kind of wide... 80-200C

 

[PGrogan]

Use a microcontroller to make the P controller. It can easily output a PWM signal able to control your mosfet. It can also be used to make an user interface to set temp.

For your sensor, you said that you will used thermocouple. You should look for thermistor also. (I am building a project where I need to control liquid temp in a vat and we will use thermistors to sense the temp).

 

[steveB]

So this proportional controller is going to have to control the PWM circuit some how. Unless you have a better idea.

I would recommend that you plan to use a PI controller. This has always worked better in any temperature control circuits I've developed. I think that thermal systems typically have significant delay that makes stabilization difficult. A proportional controller will probably need low gain to remain stable, and then your temperature error is larger than desired.

The recommendations to use a microcontroller are definitely a good idea. However, in a pinch, a simple analog loop works well too. A PI filter can be made with one OPAMP. Just take your standard inverting amplfier with gain -R2/R1 and replace R2 by a series resitor and capacitor.

 

[Leftyretro]

I would recommend that you plan to use a PI controller. This has always worked better in any temperature control circuits I've developed. I think that thermal systems typically have significant delay that makes stabilization difficult. A proportional controller will probably need low gain to remain stable, and then your temperature error is larger than desired.

The recommendations to use a microcontroller are definitely a good idea. However, in a pinch, a simple analog loop works well too. A PI filter can be made with one OPAMP. Just take your standard inverting amplfier with gain -R2/R1 and replace R2 by a series resitor and capacitor.

Sometimes you can find good prices on E-bay for industrial PID temperature controllers. Examples:

**broken link removed**

**broken link removed**

**broken link removed**

Lefty

 

[Dacr0n]

okay well....

Here's the catch... I need the response rate to be in the millisecond range... I purchased a PID controller and the reason it will not work for me is that I need it to give almost realtime feedback to the PWM portion of the circuit so that the temperature is right on regardless of how much heat it being drawn from the heater. The PID controller I have at present has a very slow response cycle. I believe it is 1 second.

A little bit about the element and its application. The heating element is going to be in a chamber, heating the air going through it. The exit temperature needs to be fairly precise +/- 3 degrees.

So... when the temp is set on the controller, regardless of how cold it is or at what speed the air is entering the heater, the exit temperature is quickly detected and the element instantly adjusts as needed to give the desired temp with precision.


Thats what I need...

Im not sure that the controller I have can do this. The manual to it is here

Its a Toho-TM 002 I believe. Its too slow unless there is another way to use it... I've been trying the SSR output but it is, like i said, too slow for what I need. There is another output labeled EV1 (event 1).. Maybe that has a faster feedback method?

That idea for the micro-controller and the PI filter sound like its what I need.

Can anyone point me into the direction of where I could find a circuit that applies both of those things?


Thanks in for all the help everyone!


Ohh and I am almost certain I will be using a type T thermocouple.

I hear they're more precise than thermistors.

 

[Dacr0n]

Yeah... I checked the specs of my current PID controller and there were 3 output options:
1.relay contact
2. SSR drive
3. Current 4-20mah

The PID i have is SSR drive.

Im guessing option 3 might have been better but I am not certain that even then it could give me the response time I need.

 

[Analog]

okay well....

Here's the catch... I need the response rate to be in the millisecond range... I purchased a PID controller and the reason it will not work for me is that I need it to give almost realtime feedback to the PWM portion of the circuit so that the temperature is right on regardless of how much heat it being drawn from the heater. The PID controller I have at present has a very slow response cycle. I believe it is 1 second.

A little bit about the element and its application. The heating element is going to be in a chamber, heating the air going through it. The exit temperature needs to be fairly precise +/- 3 degrees.

So... when the temp is set on the controller, regardless of how cold it is or at what speed the air is entering the heater, the exit temperature is quickly detected and the element instantly adjusts as needed to give the desired temp with precision.


Thats what I need...

Im not sure that the controller I have can do this. The manual to it is here

Its a Toho-TM 002 I believe. Its too slow unless there is another way to use it... I've been trying the SSR output but it is, like i said, too slow for what I need. There is another output labeled EV1 (event 1).. Maybe that has a faster feedback method?

That idea for the micro-controller and the PI filter sound like its what I need.

Can anyone point me into the direction of where I could find a circuit that applies both of those things?


Thanks in for all the help everyone!


Ohh and I am almost certain I will be using a type T thermocouple.

I hear they're more precise than thermistors.

Its unlikely to get air temperature to settle within milliseconds, in the range you specified. It will likely oscillate. If you then dampen it severely to get rid of the oscillation, your back to a slow response.

Most applications for fast temperature regulation are for radiant heat like in a IR oven application. I did one for a wafer annealing process years ago.

 

[Dacr0n]

I've started looking into Thermocouple amplifiers such as the AD595-AQ

So in theory.. I could hook the thermocouple up to that and then take the output from the IC and feed some sort of control circuit to regulate the PWM?


Would the control circuit be the "PI filter that was spoken of earlier?

 

[Dacr0n]

Its unlikely to get air temperature to settle within milliseconds, in the range you specified.

Actually... the the exact temp it needs to accurately maintain is 140c I believe...

Would that still be difficult ?

Lets just pretend it is quite possible

 

[steveB]

That idea for the micro-controller and the PI filter sound like its what I need.

Can anyone point me into the direction of where I could find a circuit that applies both of those things?

If you choose to use a microcontroller, then, for this application, you can implement the PI filter and feedback control digitally in the microcontroller. The response times for thermal control are relatively slow and any modern micro can handle it. A micro can also help you optimize the response time more easily. You can quickly tune the P and I gains, and even develop an adaptive gain where the gains depend on both the external temp and the control temperature.

 

[PGrogan]

In our app, with are developping a feed-forward controller. The system's tau is really high (for a 30 gal fill up with water it is about an hour). With air, the constant would be little higher.

Question : how will you use your thermocouple (in what circuit)? You can't use it alone.

 

[Dacr0n]

Thanks for the info steveB.

What would be the first step to getting this going... Figuring out how to program a microcontroller? I would imagine I need a programer board or something right?



Ohh and regarding the thermocouple..

It is a special one that is specifically designed to respond quickly to changing air temperatures and be non effected by any electro magnetic interference (Type T) that may be caused by any nearby heating elements.

 

[steveB]

Thanks for the info steveB.

What would be the first step to getting this going... Figuring out how to program a microcontroller? I would imagine I need a programer board or something right?

Yes, that would be a good first step. I'm not the best person to recommend microcontrollers. There are many people here that could guide you better. I've built mostly analog temperature controllers for laser diodes. However, I did use a TI MSP430 micro (the smallest one, I can't remember the exact number: maybe 12xx or something) for a temperature control circuit and it worked very well.

Ohh and regarding the thermocouple..

It is a special one that is specifically designed to respond quickly to changing air temperatures and be non effected by any electro magnetic interference (Type T) that may be caused by any nearby heating elements.

It's good to have a fast temperature sensor, and circuit, as this allows the best control. Also, a fast drive circuit helps. Ultimately, the slow dynamics of the thermal system is what limits you, if you do a good job on the circuit and control design.

 

[Rolf]

Mass

okay well....

Here's the catch... I need the response rate to be in the millisecond range...

{snip}

Ohh and I am almost certain I will be using a type T thermocouple.

I hear they're more precise than thermistors.

From what I have seen, a standard thermocouple can not respond in milli seconds because has way to much mass. And I have not seen any made of the thin wire gauge your application would require. Maybe you could get one custom made. It would be very fragile.

 

[Andrew Leigh]

Hi

If you need accuracy then a thin film PT100 is the route to go. They have a linear response and you can get them in very small sizes (2mm x 2mm). Google MINCO and search for their thin film elements.

Andrew

 

[Leftyretro]

Hi

If you need accuracy then a thin film PT100 is the route to go. They have a linear response and you can get them in very small sizes (2mm x 2mm). Google MINCO and search for their thin film elements.

Andrew

We used PT100 RTD sensors a lot at the refinery I worked on. They are easy to work with, very stable. In the process control industry they are considered the gold standard. Here is a link to an affordable one if anyone would like to play around with one.

**broken link removed**

Lefty

 

[Dacr0n]

Well I am pretty sure the TC I have will cut it as far as speed goes....

Its the smallest TC I have ever seen...

Its like a tiny little ball, maybe 1.5-2.0mm in diameter and its supposed to go right where the air flows so I am willing to bet that it can give me close to what I want.


I've looked on the minco site and the special TC I have looks better than most of their high accuracy fast response stuff.

SO I guess my next step is.. I am going to buy one of those linear TC amp chips..

Then I am going to need to incorporate that somehow with a proportional control circuit that in turn is going to control this PWM load driver which is going to control the heating element...

The PWM circuit looks like this.
**broken link removed**


Any ideas?

I really appreciate everyones help!

Thanks again.

 

[Andrew Leigh]

Hi

Thermocouples are fairly old technology and are not noted for their accuracy. I speak under corrrection but would doubt if your thermocouple is able to offer ±0.1°C tolerance. If you T/C is located someway from you circuit you will also need to use compenating cable. PT100's or Thermistors are in my opinion a more elegant solution.

I have just completed a project which controls an element and uses a thermistor. It consists of a LM741 configured as a comparitor, it works very well. The circuit is equipped with a pot to set the hysteresis, a tighter tolerance up to ±0.1°C is possible. As mine drives a relay I kept the tolerance wider to avoid excessive switching. I don't know if it can be adapted to drive your PWM module, I simply don't have skill.

Have a look at the attached schematic and see if it gives you ideas.

Regards
Andrew

PS: I am curious as if it works would like to adapt mine accordingly.

 

[Dacr0n]

You know what...... I am getting mixed info then... Some people say TCs are superior and more accurate than Thermistors and then I hear vise versa.

In any case....

Its a good thing I have collected a range of parts for this project.

I have also purchased some very small/quick responding thermistors as well and am more than willing to play with that circuit you posted... Thanks much for that..

I will let you know how things go.

 

[Rolf]

Hi

{snip}

Thermocouples are fairly old technology and are not noted for their accuracy.

Regards
Andrew

PS: I am curious as if it works would like to adapt mine accordingly.

They might not be the absolute best but when you consider that they are cheap and has an useful range of over 2000°F.
The calibration sheet for my thermocouple pyrometer shows only a +- 3°F non linearity from -50 to 1800°F.
So that means that over a narrow span it is for most practical use perfect.
What standard do you think the OP should use to calibrate his controller against to make sure he has met his engineering goal? Like most of us he will probably check it against a non calibrated instrument.

What is the purpose of talking about this kind of accuracy if it is not calibrated against a standard traceable to a national laboratory?

Let us take an example, my digital outdoor thermometer it has a 0.1°F resulotion, what good is that, when the basic instrument might have a +- 1°F accuracy or worse. But to the average consumer it sure looks good.

By the way I was fortunate to work in the electrical department of a calibration laboratory for many years. And of cause we had to send our standards to a national laboratory for calibration.

I guess the whole purpose of this diatribe is to illustrate the necessity for using restraint when using the word "accuracy".