Fine temperature control (Triac + temperature sensor)

[KrisBlueNZ]

Right. You'll need a more complicated control circuit, I think.

Assuming that the control circuitry is a separate transformer-powered circuit, you will probably want to use an optocoupler to drive the gate of the triac - something like the MOC306x series. These include a zero crossing detector and are connected between MT2 and the gate, with a low value series resistor. Switching at zero crossings produces less interference than using phase control. Instead of turning the triac ON part-way through each half-cycle, you turn it ON for whole half-cycles, and use the thermistor to determine the ratio between half-cycles for which the triac is ON, and half-cycles where the triac is OFF.

So you drive the MOC306x with a relatively slow signal (e.g. 0.2~1.0 Hz) whose duty cycle is determined by the temperature detected by the thermistor.

 

[Gryd3]

So you drive the MOC306x with a relatively slow signal (e.g. 0.2~1.0 Hz) whose duty cycle is determined by the temperature detected by the thermistor.

That sounds pretty low. If I have a 60Hz sine wave, would I not have 120 zero crossings per second?
If I understand correctly, you're talking about triggering the triac at the zero crossing mark and using the entire half-wave right?

...You know what? I think I answered my own question thinking about it. A 1Hz 'pwm' used with a 15% duty cycle will use 30 of the available 120 half-waves (positive and negative)

 

[KrisBlueNZ]

Yes that's right. We're talking about a heating load, not a light bulb, so it's fine to switch it ON and OFF with a cycle time measured in seconds.

 

[Gryd3]

So.. I hit another wall..
I have been digging into PID, but am unsure how to interface it to my application without the result being the same as a simple temperature switch like those on waffle irons...
This is a sample circuit I have found:

I understand enough to use it in a positioning system, or a balancing robot because when the circuit is at it's set value there is no output... but I am not familiar enough to make it 'hold' a set value...

I'm thinking that to be able to maintain an accuracy of at least 0.5 degreed, I will need a microcontroller.
More tips are greatly welcome at this point...

 

[Arouse1973]

I might not have this quite right but it looks like the proportional control is the actual temperature + the gain of this stage. This would be compared to your required temperature and create an error voltage which causes the heater to switch on. The integral averages out the temperature rise over time until the error is zero and this would turn off the heater. The differential part monitors the rate of change of temperature rise and makes sure the temperature doesn't rise too quickly. So I would say you are going to need a temperature sensor and a micro of some kind to monitor the output and compare it to a temperature setting. But I am no expert on this.

Steve might know more?

Adam

 

[Gryd3]

I might not have this quite right but it looks like the proportional control is the actual temperature + the gain of this stage. This would be compared to your required temperature and create an error voltage which causes the heater to switch on. The integral averages out the temperature rise over time until the error is zero and this would turn off the heater. The differential part monitors the rate of change of temperature rise and makes sure the temperature doesn't rise too quickly. So I would say you are going to need a temperature sensor and a micro of some kind to monitor the output and compare it to a temperature setting. But I am no expert on this.

Steve might know more?

Adam

I know a micro would work very well... but I would like to attempt to do this without...
Kind of a challenge for myself.
I know that circuit will present me with the 'error' rate, which would be a voltage offset around 0V depending on if my current temp is too high or too low. This makes sense to me for something like a balancing robot or positional control... because the output will move either + or - to move the device, once in position the output would be 0V. I'm scratching my head thinking of how I could translate this to a different output.
For example, I have an output voltage currently of 3V, and the temperature is too low, I want the output voltage to climb, and once at the desired temperature, have the output voltage hold at the current level. (which could be 4.5V)

I am getting ready to say screw it and use a micro though, but I won't learn anything new that way

 

[(*steve*)]

I think that if the input of this circuit is proportional to the difference between the measured and the desired difference, you'll get something like the desired behaviour.

I say "something like" because tuning a PID controller can be an exercise in frustration. Getting the balance right between the three of them can be time consuming.

 

[Gryd3]

I think that if the input of this circuit is proportional to the difference between the measured and the desired difference, you'll get something like the desired behaviour.

I say "something like" because tuning a PID controller can be an exercise in frustration. Getting the balance right between the three of them can be time consuming.

My thought was to use a potentiometer as a voltage divider to set the desired temperature (Using a 0-5V scale)
I would then hook up the temperature probe in a second voltage divider and use that for feedback.
The part I am fiddling with is that the circuit input will be 0-5V, the output will be a PWM signal, and the feedback should be 0-5V. The PWM output should increase duty cycle while the feedback is below the input, and decrease while the feedback is higher than the input, but I would like the PWM duty cycle to remain constant when the target is reached... Or am I misunderstanding that it will never be constant?