Proportional Control Circuit w/ Thermocouple

[blueroomelectronics]

I remember when fuzzy logic was all the rage for temperature control.

 

[Leftyretro]

I remember when fuzzy logic was all the rage for temperature control.

It was a marketing buzz for awhile, later adaptive tuning was more excepted in that while it was still basically good PID control, the extra adaptive tuning algorithms help automatically arrive at a good set of tuning variables for the PID. It could save days in tuning a loop, especially slow process with long time lag like some temperature loops. Pressure, flow and level control is much easier to tune.

A lot of real advance control these days just rely on well tuned PID controllers with a higher level computer program that fiddles with all the setpoint values to optimize and drive towards system wide objectives, while letting the PID worry about each individual control loop.

Lefty

 

[Rolf]

Physics......

{snip}
Rolf: Its not overkill and quite necessary as one of the requirements is that the heating system is that it needs to turn on, be able to deliver the desired heat(with decent precision) with in 5 seconds of operation signal, maintain the target temp range while the flow rate of the air changes slightly, and immediately begin cooling when the operating signal stopped. I highly doubt a house thermostat can do that..
{snip}

Any heating element starts cooling immediately after power is interrupted, that is a given. To what degree depends to a large extent of the mass of the heating element and the air flow and temperature of its environment.
Everything has mass even your thermocouple, what is its response time in the range you are using it? I see it has a time constant of 6 seconds, is that fast enough?
This has to be taken into consideration or your circuit will over shoot.
I am repeating myself but the reason a simple thermostat doesn't overshoot is because it has a heat rise antisipator.
As far as I can tell you have not factored this into your design.

 

[Dacr0n]

Ugh... okay... let me try to explain one more time how this thing works. What the air is heating isn't really important.... All that is important is that the temperature is close to target ( +/- 10 degrees of target)


I have a 2" ceramic tube with a nichrome coil in it. The air that is to be heated flows through that. The coil is hooked up to a 12VDC PWM control circuit. I take the temperature reading 4" away from the coil to get the exit air temp. The unit is more than capable of heating the air to my desired temp which as of right now is 140 Celsius. I am bouncing between 120 Celsius and 160 Celsius.

I've almost got what I need so far. I just need to add a loop to that PWM circuit that proportionally reduces the duty cycle when the target temp is passed.

I would like to go this route unless anyone knows of an already built controller that will do just that for less than $40.


Rolf: are you asking how long it takes from the moment the coil is energized on to heat up?

If so... its less than 3 seconds because its very thin nichrome that has the ability to heat up quickly.


Im trying to keep this as simple and cheap as possible..... so anyone have any input for my current setup? Is there no way to add the loop mentioned earlier?

 

[Leftyretro]

"I would like to go this route unless anyone knows of an already built controller that will do just that for less than $40."

This is a useful inexpensive controller that should be able to handle the I/O needs. It's open source hardware and software project using C and C++ as the user programming language. There are several clone manufacture, this one is real good.


**broken link removed**

Here is a user contribuited PID software libary:

Arduino playground - PIDLibrary

Lefty

 

[duffy]

Is there no way to add the loop mentioned earlier?

Maybe, but it's going to take a little putzing around. Your thermistor is near the value of the pot: 250k at room temp. The resistance goes DOWN when it heats up, so if you disconnect the leg of the pot that's on the side you turn it to go LOWEST on the pwm, connect the thermistor between the 1k resistor and the wiper on the pot, so it works like that side of the pot... See what I'm getting at? Instead of lowering the resistance with the pot, you now lower it with the thermistor. PWM goes down, heater cools off, thermistor cools off, PWM goes back up, etc.

Ugh... okay... let me try to explain one more time how this thing works. What the air is heating isn't really important....

Actually, that's probably the most important thing when the question is "What kind of controller should it be?" It's like asking whom you should hire to spray water, but not explaining if they're to be a fireman or gardner.

Nobody's trying to weasle secrets out of you, and you might even be shocked at the level of disinterest guys like us have in wanting to hear about another "big idea". To speak in general terms about the dynamics of the system will not give anything away, and the larger issue of P vs PI vs PID cannot be determined without it.

 

[Dacr0n]

Maybe, but it's going to take a little putzing around. Your thermistor is near the value of the pot: 250k at room temp. The resistance goes DOWN when it heats up, so if you disconnect the leg of the pot that's on the side you turn it to go LOWEST on the pwm, connect the thermistor between the 1k resistor and the wiper on the pot, so it works like that side of the pot... See what I'm getting at? Instead of lowering the resistance with the pot, you now lower it with the thermistor. PWM goes down, heater cools off, thermistor cools off, PWM goes back up, etc.

Cool...

Ill tinker with that a bit...

Thanks for the links Lefty... Ill check those out too..


Much thanks for all the help guys.

Be back soon with some results..

 

[Rolf]

Physics II

{snip}
If so... its less than 3 seconds because its very thin nichrome that has the ability to heat up quickly.
{snip}

Now you just have to figure out how long it takes for the ceramic tube to reach thew same temp!

Because the air is being heated by the nichrome wires and at the same time being cooled by the ceramic tube.
Unless the ceramic tube's temp is close to the set point, then I can visualize stability problems. You didn't forget to insulate the tube did you?
In any oven or heater, whether the air is flowing or not, you can not treat the heating element as an entirely different element from the the chambers walls. They both has to be the same temperature before the air therein will be at that same temp.
But if the air is flowing then only the last section of the tube (chamber) has to meet that criteria. But that is even more difficult to achieve.

 

[Dacr0n]

Actually....

Here's a bit more info about operation.


A typical example of the typical work routine for this component would be:

30 Seconds on, while ideally providing the target temp.

Followed by maybe..... 5 minutes of non-operation time.

Followed by another 30 seconds, et cetera, et cetera, and so on and so forth following that pattern.

So... I would imagine that tube wouldn't ever really get hot... just warm...

Plus, if I have a semi decent control system... the controller should be able to compensate for the heating element enclosure being cool by bumping up the juice... and when it is already warm not requiring as much power.

I mean.. that is supposed to be the whole idea behind proportional control right?

 

[Andrew Leigh]

Hi

I still follow this thread with great interest but I am unable to build a full picture in my head as I think we are short on a couple inputs, or perhaps my lack of understanding is not allowing me to comprehend what is going on. My thoughts;

We are trying to create an intellgent heater that will control to ±10C?

1. With such a wide tolerance why then the need for such a responsive sensor with such little thermal mass?
2. The desired temperature of ±10C is easlity attainable through simpler means but for some reason you have need to narrow the duty cycle through PWM. Would it be correct to assume therefore that the temperature is not as important as the method of heating, PWM in this case?
3. Why is the substance being heated not important? Surely the thermal mass has a massive role to play in the design your element and circuit.

Heres a blue sky thought to try stimulate new ideas.

As your tolerance is very wide, 7% of your target temperature, you can consider a less elegant solution of a heat exchanger consisting of multiple zones independently controlled with seperate and elements.

Zone 1 - Lift the ambient temperature of the air through a permanently on element by say 120C. (This assuming our ambient temperature will not lift by any more than 20C.
Zone 2 - Have an element to account for a very coarse change in ambient temperature say ±20C.
Zone 3 - Have an element to control the temperature by ±5C.

Zone 1 stays on permanently, Zone 2 less so and Zone 3 hopefully even less. Providing the zones are baffelled correctly, insulated then this should achieve very similar results to PWM but with inexpensive controllers. The combined duty cycle may come close to what you want.

Perhaps this can be achieved with 3 windings in one element in one heat exchanger, but with two controllers.

Cheers
Andrew

 

[blueroomelectronics]

I doubt we'll find out what the OP is building. He doesn't seem to understand it himself.

 

[steveB]

SteveB: it must be some sort of proportional control. ON/OFF is unacceptable....

I would just make a quick comment here. You may be right that only proportional control will do the job. However, it might be better to design with a PI controller, and if you you don't need it, just set the integral gain to zero.

I say this because you can probably get better performance with PI, and adding the integral term is not much more difficult than just doing proportional gain. For example, with an analog contoller (with op-amps) you would just add a capacitor. Also, with digital control (with a micro) the integral term is just a counter (i=i+1) times a gain, added to the proportional term. This is very easy to implement.

The main issue is that most thermal systems have delay. Delay will result in instability with proportional control, unless the gain is set low (somtimes < 10). Lower gain implies larger steady state temperature error. There can also be overshoot problems. With PI control you have more flexibility to control the overshoot, and the steady state temperature error goes to zero.

It really all depends on the details of your system, and also on how stringent your performance specifications are. So perhaps proportional or on/off type control is just fine, but I just thought I'd throw out these ideas.

 

[Rolf]

{snip}
So... I would imagine that tube wouldn't ever really get hot... just warm...

Plus, if I have a semi decent control system... the controller should be able to compensate for the heating element enclosure being cool by bumping up the juice... and when it is already warm not requiring as much power.

I mean.. that is supposed to be the whole idea behind proportional control right?

"Bumping up the juse" wont do it! The tube HAS to be at the same temp as the outgoing air!
(maybe not the whole tube but definitely the end portion where the air excites)

And using a proportional controller doesn't even enter into the picture!

As an example take a cooking oven, set the temp to 250°F and turn it on. After a about 90 seconds the cal-rods are red hot. But it takes several more minutes for the INTERIOR (walls, door etc.) to heat up. Finally when the interior is at 250° then the air is at 250°, and the thermostat opens.
The ovens difference in size and shape from your apparatus does not change how the air is heated, the principle is exactly the same.
You can not change physics, no matter how hard you try, so you have to design around the obstacles instead.
Maybe you need to contemplate how to get the air out of the tube without contacting the walls; a neat but impossible goal.

 

[Dacr0n]

Hi


We are trying to create an intellgent heater that will control to ±10C?

1. With such a wide tolerance why then the need for such a responsive sensor with such little thermal mass?
2. The desired temperature of ±10C is easlity attainable through simpler means but for some reason you have need to narrow the duty cycle through PWM. Would it be correct to assume therefore that the temperature is not as important as the method of heating, PWM in this case?
3. Why is the substance being heated not important? Surely the thermal mass has a massive role to play in the design your element and circuit.

Thanks for the input Andrew.

Actually, when I said ±10C, I was being very generous as to what the acceptable tolerances because I was because I felt that I was not making much ground as far as advancing on the circuit design, so I loosened the parameters a bit.

Now that I am making advances and ideas are flowing in my head and yours , I would say that the error margin must be a little less... I mean.. If I could do it ±5C I can consider my goal met and application of a smart temperature controller complete. Ideally I wish the controller to be able to integrate the feedback from the sensor 4" away from the element immediately and respond quickly and accordingly to achieve target. Being that all parts of this system are very fast to heat up and it detects the output temperature almost immediately.

In regards to the temp not being as important as the method of heating, I stress again that tight temperature control is the primary objective of this project.



And lastly, in response to your post Andrew, the substance being heated is solely, good old fashion, air.


I remember Rolf mentioning that the temperature of the ceramic tube be added to this equation so ill just say a few things about that. Since the heating element contained in the tube is like the coil elements inside a hair drier except, symmetrically covering the entire inner "area" of the tube so no air flows by without being subject to heating... I have not really worried about the tubes temperature. Being that is is very thin AND only about 4" long. I don't think that tube temperature is going to be a very heavily influencing factor. Recall.. The temperature feedback device(either Thermistor or thermocouple) is at the end of that 4 inch tube so the air doesn't really have much time to be affected by the tube temp... It simply gets heated---->heated air goes straight to the sensor---->out at the target temp with out being affected by much else.

I could understand how the tube temp could be a problem say on a run of a couple feet, but this is such a short run, thats why my first inclination was to ignore it all together until the time comes to really fine tune.

Hey blueroomdude thanks for the utterly useless criticism. It motivates me even more.

 

[Rolf]

Actually....

Here's a bit more info about operation.


A typical example of the typical work routine for this component would be:

30 Seconds on, while ideally providing the target temp.

Followed by maybe..... 5 minutes of non-operation time.

Followed by another 30 seconds, et cetera, et cetera, and so on and so forth following that pattern.
{snip}
Why not keep the tube at the target temperature with both external and internal heating elements? Then you could turn the air flow on and off at the inter-wall needed.

 

[blueroomelectronics]

Hey blueroomdude thanks for the utterly useless criticism. It motivates me even more. **broken link removed**

Go for it. Best I can figure is you're building a dirigible or a welder. Browns gas is not so good for cars.

Is there a reason you won't tell us what its supposed to do?

 

[duffy]

And lastly, in response to your post Andrew, the substance being heated is solely, good old fashion, air.

You can pretend it works like that, but the thing that's using the air is part of the heat flow.

Unless you're just blowing hot air.

 

[Dacr0n]

Unless you're just blowing hot air.

Thats pretty much sums it up duffy. Air blown at a fairly specific temperature that can be slightly increased or decreased depending on the needs.

I am curious to try add the thermistor to the loop like you mentioned back on page 5. Ill do that when I have some time to tinker and post some results.

 

[Rolf]

{snip}
I remember Rolf mentioning that the temperature of the ceramic tube be added to this equation so ill just say a few things about that. Since the heating element contained in the tube is like the coil elements inside a hair drier except, symmetrically covering the entire inner "area" of the tube so no air flows by without being subject to heating... I have not really worried about the tubes temperature. Being that is is very thin AND only about 4" long. I don't think that tube temperature is going to be a very heavily influencing factor. Recall.. The temperature feedback device(either Thermistor or thermocouple) is at the end of that 4 inch tube so the air doesn't really have much time to be affected by the tube temp...

Don't you think the surface area of the tube is at least a thousand times greater than the heating element? And the air flows by both at the same speed.

I don't know about your hair dryer, but the one my wifes uses has be ON for over 30 seconds before the temperature stabilizes at its max.
What could cause that?

BTW what is the air flow velocity in your apparatus?

 

[Dacr0n]

Don't you think the surface area of the tube is at least a thousand times greater than the heating element? And the air flows by both at the same speed.

I don't know about your hair dryer, but the one my wifes uses has be ON for over 30 seconds before the temperature stabilizes at its max.
What could cause that?

BTW what is the air flow velocity in your apparatus?

Well... I agree with ya rolf that the surface area of the tube is greater than that of the element, but here's my thinking..

Lets talk about temperature transfer effectiveness or "influence" on the air.

Say... the surface area of the small tube is 30x the area of the heater.

Per cubic centimeter of area, the element has a much greater influence on the air passing through than the air that contacts the walls of the tube.

I don't know physics. Im not even the best at math... So I cant write down how it works..

All that I know is, I don't have to really worry about the tube temp.

Why?
#1. Well... first of all... You're hair dryer takes 30 seconds to get to its max temp? Well this thing blows air HOTTER than target within 3 seconds of operation.

#2 Since it can do that, then all is needed is a simple proportional control to not let it go over during the initial start up and keep it near target.. It is obvious that the element will be working a little harder at the beginning when the tube is at RT to get the target temp... but the power delivered to the element is calculated by the circuit and the target temp is held.

As the tube gets a little warmer with periodic operation, of course the target temp will be reached with less work required by the element, that amount of work again is determined by the controller.

So without really understanding the physics of it... I have witnessed, first hand, that it is totally possible to achieve what I am talking about here.

Ohh and the air flow is produced by a box (computer type) fan.

Volume or air and rate of flow? I have no ****ing clue...

Like I said... physics isn't my thing.... But I am a specialist at ghetto improvisation.