Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Proportional Control Circuit w/ Thermocouple

Status
Not open for further replies.
So then the best way to be accurate is to calibrate any "projects" with a precisely calibrated piece of lab equipment.

But having the temperature scale right on isn't really necessary or a particular goal for this project... Even if its +/- a couple degrees or so...




It's the sampling rate, response time, steadiness of control that is the goal here.

Obviously the electronics are going to have to be tweaked and tuned a little to achieve this. So shall it be done.
 
Another question for all the master circuit readers in here:


Is there a way to adapt that 555 PWM circuit so that it varies the pulse width in response to a voltage change instead of a change in the 250k Ohm potentiometer?
 
Hi Rolf, you make some interesting points

however, for any given component, for the same price, surely it is prudent to use the one that is the most accurate? Using components with wide tolerances increases the risk of compounding inaccuracies. To my mind one can remove these potential inaccuracies from the project at the outset.

Accuracy and calibration are entirely different in my mind. Calibration assures that the said piece of kit is able to deliver what it claims. You can calibrate a controller of ± .01°C and you can calibrate a controller of ± 10°C. Accuracy is not determined by calibration, calibration varifies accuracy. My point is that even though the controller may not be calibrated, its accuracy can still be excellent but you just won't be able to quantify it.

I am not too sure what you mean in your point about resolution? In my experience, when one speaks of "resolving" in measurement terms it means that you are able to accurately determine to that level of detail. i.e. resolving to 0.1°C would mean that the equipment is able to accurately provide results to 0.1°C. Are you refering to the "resolution" of the display only as being different from the instruments resolution.

One important reason to use the most accurate sensor is that it improves hysteresis which is vital to being able to control within a very narrow band.

I was not knocking the virtues of thermocouples, each device has its niche and will perform well. I have used all types.


Cheers
Andrew
 
Hi Dacr0n,

should you fiddle with the circuit you will need two mods;

1. A tweak for your temperature range. Thermistors are normally rated at 25°C, their resistance alters obviously as they warm or cool. In your instance you need to buy a thermistor that is as close to 15k when at your desired temp of 140°C. If the thermistor ends up at a different value than ≈ 15k @ 140°C then R1 must be altered to the same value. i.e. your thermistor is 1650 Ω then put a 1k5 in at R1.

2. The circuit is set as a"cooling" thermostat. Mine heats an element in an absorption fridge in order to cool it. Yours is a true heating application, the inputs to pins 2 and 3 on the 741 must be swapped.

Here is the original article that I used, it has all the theory. Craig's Thermostat Circuits

I have the PCB as well in Eagle format.

Cheers
Andrew
 
Calibration II (last installment)

So then the best way to be accurate is to calibrate any "projects" with a precisely calibrated piece of lab equipment.

{snip}


If a project is doing its "thing" I would not bother to calibrate it, as a mather of fact I don't think I have ever seen a DIY project with a calibration sticker. But I have bought used equipment that at one time was calibrated. It is presumed to be out of calibration when the date expires. In practical therms that is of cause seldom true.

IMHO what most DIY's call calibration is simply verification of operation and rudimentary comparison with other available uncalibrated equipment.

Very few places that have calibration service available us DIYs and it is horrendously expensive.
To get a Tektronix oscope calibrated to factory specifications, would probably cost more than the scope cost you got it the first place, providing you bought it used.
And then it would only be within +- 3%, at least that was the major specs when I did them.
 
Hi Rolf, you make some interesting points

however, for any given component, for the same price, surely it is prudent to use the one that is the most accurate? Using components with wide tolerances increases the risk of compounding inaccuracies. To my mind one can remove these potential inaccuracies from the project at the outset.

Accuracy and calibration are entirely different in my mind. Calibration assures that the said piece of kit is able to deliver what it claims. You can calibrate a controller of ± .01°C and you can calibrate a controller of ± 10°C. Accuracy is not determined by calibration, calibration varifies accuracy. My point is that even though the controller may not be calibrated, its accuracy can still be excellent but you just won't be able to quantify it.

I am not too sure what you mean in your point about resolution? In my experience, when one speaks of "resolving" in measurement terms it means that you are able to accurately determine to that level of detail. i.e. resolving to 0.1°C would mean that the equipment is able to accurately provide results to 0.1°C. Are you refering to the "resolution" of the display only as being different from the instruments resolution.

One important reason to use the most accurate sensor is that it improves hysteresis which is vital to being able to control within a very narrow band.

I was not knocking the virtues of thermocouples, each device has its niche and will perform well. I have used all types.


Cheers
Andrew

In some parts of some circuits like voltage divders and feedback in fixed gain amps were the gain accuracy is important. then you need to use the best you can afford. And of cause in hi gain circuits low noise components are a must.

My point is that even though the controller may not be calibrated, its accuracy can still be excellent but you just won't be able to quantify it.

Why not just say it is working within the required parameters.

Are you referring to the "resolution" of the display only as being different from the instruments resolution.

The temperature readout is digital (and probably 100% accurate) But the sensor is analog and has a certain nonlinearity so the two are not a perfect match. Also you have to take into the equation the accuracy of the calibration standard used and how diligently it was used. What I am saying is that the instrument's (thermometer) accuracy is problay no better than +- 1%, no mater what the resolution of the readout. I am almost certain they are just given a rudimentary check on the production line, how else can they sell them so cheap. When you look at the thermometers readings in the store you will see what I mean.
I have newer seen any type of commercial thermometer that has any accuracy claim on the instrument itself or in the owners manual.

One important reason to use the most accurate sensor is that it improves hysteresis which is vital to being able to control within a very narrow band.

Sorry but I don't think hysteresis has anything to do with sensor accuracy.
Sensor accuracy is usually given over a large temperature span where as hysteresis is either buildt into the instruments electronics or it is a function of the sensors physical mass (including wiring and mounting), I am fairly certain it is not in the spec sheet, at least I dont remember seeing it.

I realize most of this discussion is about semantics' and I don't like to be picky but as DIY's, technicians and engineers we should strive to use the correct terminology so we can understand each other without to much confusion.

P.S. I am confused enough already, my wife tells me!
 
Last edited:
Okay... now that we've got the semantics out of the way..


Thanks for the tips Andrew.

The thermistor I have is made for high temps. Its i believe its 1,000 ohms @ 200C.

@ like 150C it would be less...

So I guess the resistance range going to come from the sensor is going to be around 1Kohm.

Thats not what I am worried about at this moment though.

Andrew i've seen Craig's thermostat circuits, but the problem I see with those is that they're ON/OFF.

Okay... here's my latest idea for this DIY unit.


I am going to use the PWM as the driver for the heater.

There are going to be two modes once I turn the circuit on...

-Warm mode

and

-Operating mode.

*Warm Mode*
When turned on, by default, the PWM driver is going to be in warm mode. This will keep the coil preheated to say..... about 65 degrees celsius, just so it heats quicker and does not fully cool off causing rapid fluctuations in the temperature of the heater coil and excess wear.

*Operating mode*
At the press of a button, a proportional controller will engage and take the element temperature to the SET temperature.


Now.. before that can be done.. I will run a few trials with the PWM driver and the heating element to see what PULSE WIDTH gets the element to what temperature.

Then I am going to have to incorporate some kind of limiter in the PWM driver that restricts the output PULSE WIDTH to something just above the width that gets me to my desired temp.


Okay... how does that sound so far?





I am not going to need to worry about overshoot or integral or derivatives because the boundaries that I will have set on the PWM driver will be so narrow that all the proportional control circuit will have to do is very small corrections within the accepted output parameters.

So thats my story and Im sticking to it.

Now I just need to make it a reality..

So now all I need is for someone to show me where I can find just a simple proportional control circuit...

I think I can adapt the rest.
 
A 1ms response time off s 1.5-2mm "ball" with leads attached doesn't sound right. I've tried to get kilohertz-range frequencies out of incandescent lamps with filaments amounting to a total thermal mass less than that with no success, and the filament has a larger surface area to volume ratio.
 
Last edited:
Semantics may get in the way but I think I missed anyone using the word we used often in process control in a refinery and that is good repeatability.

Absolute accuracy is often just a pissing contest and is never really resolved in the real working world, just a time and cash sink. What is required for good control is good repeatability. That is if a temperature sensor is saying the process is at 400F today, what does it say in a week, at a different time of day, with a different ambient temperature, but assuming the process is at the exact same temperature as the prior reading. Most process control uses PID controllers to maintain a process at a desired 'setpoint'. As long as the sensor has good repeatability then there is the possibility for good control.

Lefty
 
Last edited:
I am not expecting the response time to be 1ms...

I just need the circuit to be taking readings and adjusting accordingly in "realtime"

I'm not sure if Im making this too clear..


Okay.... for example.. I have a PID controller...

It takes readings once a second then adjust its PID calculations accordingly.

I need something that just has a continuous correction with out that delay

What I need is a circuit with :

-A input that reads voltage from the sensor.

-Another input to the circuit that reads voltage and is the basis for the TARGET voltage on the first input.

-Then an output that simply puts out a varying voltage depending on the differential from the sensor voltage and the set target voltage...

Is this circuit really so hard to build??...

I've searched the net a bit but cant seem to find any schematics for proportional circuits that just do those few simple things outlined in "What I need"

I've just thrown out the whole temperature topic for now until I get this circuit designed.


Are there any circuit designing/simulating programs anyone can recommend for this?
 
I am not expecting the response time to be 1ms...

I just need the circuit to be taking readings and adjusting accordingly in "realtime"

I'm not sure if Im making this too clear..


Okay.... for example.. I have a PID controller...

It takes readings once a second then adjust its PID calculations accordingly.

I need something that just has a continuous correction with out that delay

What I need is a circuit with :

-A input that reads voltage from the sensor.

-Another input to the circuit that reads voltage and is the basis for the TARGET voltage on the first input.

-Then an output that simply puts out a varying voltage depending on the differential from the sensor voltage and the set target voltage...

Is this circuit really so hard to build??...

I've searched the net a bit but cant seem to find any schematics for proportional circuits that just do those few simple things outlined in "What I need"

I've just thrown out the whole temperature topic for now until I get this circuit designed.


Are there any circuit designing/simulating programs anyone can recommend for this?

I think if you look at some classic control references they will show that often the process time lag is the limiting speed factor. Temperature control is typically very slow as thermo lag is usually the limiting factor. That is unless you have a 10kw heating element mounted inside a shoe box size oven :p

Lefty
 
Absolute accuracy is often just a pissing contest and is never really resolved in the real working world, just a time and cash sink. What is required for good control is good repeatability.

Wisdom.


Dacron, I looked through your posts and didn't see anywhere that you mentioned the wattage of the heating element, voltage or anything about what you're driving.

I wouldn't suggest the 555 for this. Although you can make a PWM out of it, it's too awkward for what you're doing. I think the TL494 chip would be much better. It's a PWM controller, has two separate differential input amps (you only need one), two uncommitted drive transistors that can be configured several different ways, a voltage reference and an onboard clock.
https://www.electro-tech-online.com/custompdfs/2009/02/tl494.pdf

I've used this chip many times with excellent results.
 
That is unless you have a 10kw heating element mounted inside a shoe box size oven :p

Lefty


Actually Lefty.... I kinda have something like that. Trust me... On a much smaller scale though.

The heat from the element will reach the sensor almost instantly.


Ohh and by the way... in regards to the heating system details.

Its going to be on a 12v DC system

with a mosfet to drive the heating element

thats as much as I know this far.
 
Last edited:
I have a small part in a large project.

The project is going to be a 100% hydrogen fueled electricity generator.

I don't know the exact physics of the engine yet, but I am pretty sure its not piston based.

My assignment in the whole deal is the intake.

I am to construct an intake with a heating element inside (most likely a coil) that is controlled proportionally to the volume of air flowing (the demand) and must be fairly accurate.. Within +/- 5 degrees.

Maybe now you can have a better understanding of why, for this particular application, the sensor will almost immediately detect the temperature that is being produced by the element. The sensor will be just on the other side of heating element, monitoring the air that travels by.

As long as the sensor has good repeatability then there is the possibility for good control.

Lefty


That is why I have spared no expense when buying the sensors (Thermistors & Type T Thermocouples). They are all of the highest quality and of the fastest response rate (Fastest that I was able to find and purchase on the net). So I need a fast and tight circuit to go with those high quality components.

I specifically bought Type T to eliminate the possibility of any interference on the sensor feedback caused by the heating element :D ,as the Type T thermocouples are unaffected by magnetic fields.
 
Last edited:
AirFLOW? That opens another can of worms, now you're dealing with fluidics and mixing. Air doesn't behave like you might think, the streams shift and stay shifted sometimes, keep that in mind.

Hope your airflow is pretty low volume, otherwise it's tough to heat, especially at 12V. If you tried to operate a little 1200W space heater at 12V it would need to pull 100 amps.

Please tell me this is a for-real project and not another damn HHO or overunity BS machine.
 
AirFLOW? That opens another can of worms, now you're dealing with fluidics and mixing. Air doesn't behave like you might think, the streams shift and stay shifted sometimes, keep that in mind.

Hope your airflow is pretty low volume, otherwise it's tough to heat, especially at 12V. If you tried to operate a little 1200W space heater at 12V it would need to pull 100 amps.

Please tell me this is a for-real project and not another damn HHO or overunity BS machine.

I agree, short transit time due to the air flow at any reasonable velocity means that any heat transfer is going to be minimum and very difficult and control probably impossible. Large heat exchangers are the normal method used to heat gas flow, not sticking a heating element in the air flow path. This is not going to be a very promising way to proceed.
 
The project is real man... Im shocked you even asked that.

If you think I am full of it then I guess you think I am full of it. :(

Its not going to require a lot of air flow. A small amount. So small that I am not even worried that the heater will have a problem heating it.

I know it will work.

I have already built the trial tube that will contain the element.

Late next week I will be receiving my PWM driver kit. For the preliminary run I am going to be running it with a 75 amp mosfet.

My power supply is 12 Volts DC---- can deliver 400AMPs if necessary, so there's no lacking there. However, I suspect that I will not need much more than 20-30 amps.

After assembling the kit and hooking everything up, I am going to be testing the heating capabilities by manually adjusting the PWM driver while there is air flowing. The thermocouple will be hooked up to a meter that displays degrees Celsius so I know where I am at.

After that is tested and my theory confirmed, i will still pursue some sort of proportional control.

I hope you guys haven't given up on me yet!
 
The project is real man... Im shocked you even asked that.

If you think I am full of it then I guess you think I am full of it. :(

Its not going to require a lot of air flow. A small amount. So small that I am not even worried that the heater will have a problem heating it.

I know it will work.

I have already built the trial tube that will contain the element.

Late next week I will be receiving my PWM driver kit. For the preliminary run I am going to be running it with a 75 amp mosfet.

My power supply is 12 Volts DC---- can deliver 400AMPs if necessary, so there's no lacking there. However, I suspect that I will not need much more than 20-30 amps.

After assembling the kit and hooking everything up, I am going to be testing the heating capabilities by manually adjusting the PWM driver while there is air flowing. The thermocouple will be hooked up to a meter that displays degrees Celsius so I know where I am at.

After that is tested and my theory confirmed, i will still pursue some sort of proportional control.

I hope you guys haven't given up on me yet!

Not yet. ;) And your proposed steps to proceed are proper and logical. If you can reach the desired temperature manually via PWM and get your desired temperature within the current limits of your MOSFET, only then does it make sense to proceed on the control algorithm. During testing be sure to measure the inlet gas temperature to find out how high a delta increase your heating system can obtain over it's 0-100% manually adjusted duty cycle of your PWM signal. The tricky part will be measuring the actual heated gas temperature independent of any direct but non-representative reading because of mounting the sensor close to the heating element. You might want to take temp readings several diameters downstream of the heating element.

Lefty

Lefty
 
Last edited:
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top