PWM And Heating Elements

[Andrew Leigh]

Hi,

I was reading another post which interested me.The poster was looking for a rapid response temperature control circuit to set the temperature on a PWM contolled heating element.

What would the benefits of a PWM heater be? The only one that I can think of is that the heater would appear to be "on" permananetly unlike with a normal thermostat where the intervals between on and off are much longer. Would PWM therefore have almost a 0° hysteresis.

Would PWM be more efficient than a conventional controller element arrangement?

Cheers
Andrew

 

[dknguyen]

The advantage is variable power control which allows you to react sooner to temperature changes that are too high or too low without making the system undershoot or overshoot by as much as a thermostat. WIth a thermostat, you would have to wait until the temperature got far enough away from the desired temperature before reacting so you do not end up overshooting/undershooting constantly.

It's just as efficient as on/off thermostats- they use just as much power for a given temperature.

It's not entirely correct to think of a PWM as "always on" compared to a thermostat. A thermostat is either on/off. A PWM heating control is somewhere between 0% and 100% duty cycle (0% being off, and 100% being full power).

 

[Leftyretro]

Hi,

I was reading another post which interested me.The poster was looking for a rapid response temperature control circuit to set the temperature on a PWM contolled heating element.

What would the benefits of a PWM heater be? The only one that I can think of is that the heater would appear to be "on" permananetly unlike with a normal thermostat where the intervals between on and off are much longer. Would PWM therefore have almost a 0° hysteresis.

Would PWM be more efficient than a conventional controller element arrangement?

Cheers
Andrew

PWM is more efficient because there are less losses in the electronics switching device controlling the current to the heater. The load (the heater) doesn't care if it's DC controlled or PWM controlled as it responds to the average power it consumes. A thermostat is a very efficient form of very slow PWM but is not a precise device (it's an on/off gap controller that cycles around a specific set temperature) and requires a relatively high amount of hysteresis to operate correctly.

If one needed a temperature control loop to be able to maintain say a +/- 1 degree temperature then a more sensitive temperature sensor is required, like a thermocouple or RTD sensor. How the load (the heater) is driven really doesn't determine the quality of the control, as a linear controlled current could work just as accurately as a PWM output, it just would not be as electrically efficient.

Make sense?

Lefty

 

[dknguyen]

PWM is more efficient because there are less losses in the electronics switching device controlling the current to the heater.

??? What are you talking about? Less energy getting to the heater = less losses in the switch. More energy getting to the heater = more losses in the switch. I don't see how PWM has anything to do with that.

 

[Leftyretro]

??? What are you talking about? Less energy getting to the heater = less losses in the switch. More energy getting to the heater = more losses in the switch. I don't see how PWM has anything to do with that.

It's the same thing as with DC regulator circuits, a load (say a lamp) doesn't care if the regulator is a linear type or a switching type, but there is a big difference in electrical efficiency between a linear regulator and a PWM regulator. Make sense?

Lefty

 

[dknguyen]

It's the same thing as with DC regulator circuits, a load (say a lamp) doesn't care if the regulator is a linear type or a switching type, but there is a big difference in electrical efficiency between a linear regulator and a PWM regulator. Make sense?

Lefty

Yeah, but thatis because a linear regulator is burning off excess voltage as heat. A thermostat does not do that because it is either fully off or fully conducting.

Would PWM be more efficient than a conventional controller element arrangement?

I just noticed the last question from the OP. Yes, with PWM you pretty much need no hysteresis compared to a thermostat since you can change the output to react almost instantly to change without significant overshoot/undershoot for the reasons already described. "Almost instantly" because a PWM hardware has limits to it's duty cycle resolution and cannot produce infintesmally small changes to the duty cycle. They also do not react instaneously (depending on what the thing is controlling due to things like thermal capacitance or a motor's intertia). In this case, your hysteresis would be called a deadband- a small range centered around the desired value (ie. temperature) in which the system does not change it's control output (ie. heater's PWM duty cycle) because the change is below the accuracy that the system is able to measure, the system does not have enough resolution to effectively react to the small change, and/or the system cannot react fast enough to prevent significant overshoot when correcting such a small change). So the current value is "close enough" to the desired value and the system just maintains it's current control output...until current value travels out of the deadband.

 

[Andrew Leigh]

Hi and thanks for the responses,


Reason for the questions. I have just completed a temperature control circuit that utilises a thermistor to control the element in my absorption fridge. The exisitng thermostat has a hysteresis of ± 7,5°C which is unacceptable for me.

Today I was going to send out the artwork to get the boards made (one for me and one for a friend). The new circuit circuit allows one to set the minimum temperature required, the temperature will then drift upwards (the degree to which is determined by the hysteresis pot) and will switch on until the lower limit is reached. I can get about ± 0.1°C which is more than I require, mine is set to 3°C with a tolerance of +1.0 -0.0

I know I don't need it but this is part of learning. From the posts I am led to believe that using PWM on the element it will give me an infinitely more control?

So with my exisiting existing arrangement when the controller see 4°C it switches on and the element will reach maximum sheath temperature and hold it for the duration until 3° is reached when it will switch off.

Would I then be correct in stating that with PWM that the element will probably never reach it maximum sheath temperature. Depending on the setting, the element will be cycling more frequently giving to my mind that is, a more constant but lower heat. This should extend the life of the element as the thermal shocks are significantly reduced?

So rather than sending the output from my circut to control a relay coil I could send it to a PWM controller for nth degree control?



Cheers
Andrew

 

[Leftyretro]

Hi and thanks for the responses,


Reason for the questions. I have just completed a temperature control circuit that utilises a thermistor to control the element in my absorption fridge. The exisitng thermostat has a hysteresis of ± 7,5°C which is unacceptable for me.

Today I was going to send out the artwork to get the boards made (one for me and one for a friend). The new circuit circuit allows one to set the minimum temperature required, the temperature will then drift upwards (the degree to which is determined by the hysteresis pot) and will switch on until the lower limit is reached. I can get about ± 0.1°C which is more than I require, mine is set to 3°C with a tolerance of +1.0 -0.0

I know I don't need it but this is part of learning. From the posts I am led to believe that using PWM on the element it will give me an infinitely more control?

So with my exisiting existing arrangement when the controller see 4°C it switches on and the element will reach maximum sheath temperature and hold it for the duration until 3° is reached when it will switch off.

Would I then be correct in stating that with PWM that the element will probably never reach it maximum sheath temperature. Depending on the setting, the element will be cycling more frequently giving to my mind that is, a more constant but lower heat. This should extend the life of the element as the thermal shocks are significantly reduced?

So rather than sending the output from my circut to control a relay coil I could send it to a PWM controller for nth degree control?



Cheers
Andrew

Good temperature control has more to do about the control electronics algorithm then what type of output drive circuitry is used. Industry uses PID controllers for precise temperature control and RTD or thermocouple sensors.

PID controller - Wikipedia, the free encyclopedia

 

[dknguyen]

Yes. Same average, but the average is done with a bunch of real times values that are closer to the average, rather than having a bunch of really low and high peaks that compensate each other to average out to the same thing. Which usually also means less overshoot/undershoot.

PWM is kind of like a really fast thermostat so it has higher bandwidth and can react to things faster producing better results if given the proper control. Except that PWM is fixed frequency/varying duty cycle, while a typical on/off thermostat just kind of turns on when the temperature is too low and off when too high resulting in no identifiable frequency or duty cycles while operating.

Like lefty said, going from a on/off thermostat to a PWM controlled switch is kind of like giving yourself better hands to work with...but to get good results you need a good brain to control them properly

 

[stevez]

Will your absorption unit operate properly at reduced input to the heating element?

 

[Andrew Leigh]

Hey Steve, awesome question.

Had not considered that ...... ah these clever forums ..... I have absolutely no idea. I am led to believe that the temperature of the element is inversely porpotional to the cooling?


You got me

Cheers
Andrew

 

[Andrew Leigh]

Blonde moment for me,

it will operate correctly. It is a three way fridge 220 /12 V / Gas. On the gas setting the flame intensity is settable so I have to assume it will.

Andrew