Modulating RMS value with PWM

[maxeen]

Hello.

So, here's the problem: I have a project where I need to control water's temperature with a PID controller. My heater is 1kW/127V(rms), so I cannot conect the PID controller made of AmpOp directly in 127V. My question is: the controller's maximum V will be about 1V, where 1V would mean 127V. How can I amplify such signal?

I've thought of using a simple transistor, but the heater requires at most 12A, and I would need to rectifie the AC signal. I've thought of using a solid-state relay, but not sure how the PID can control the AC signal.

Any tips?

 

[Warpspeed]

Many PID temperature controllers have an on/off output that you can program the time interval.

For instance,if you program the time interval for 100 seconds.
If the PID output commands 37% heating power, the output switches on for thirty seven seconds and stays off for sixty three seconds.

This is often very useful for say a gas burner controlling the temperature of a big oven or swimming pool, where the gas burner can only be fully on or fully off.

For heating water with an electric element, it should work perfectly well even if the time interval is set to be fairly long.
Or the time interval can be set quite short, say every second or two, and you can switch individual mains cycles on and off with a solid state relay.

 

[maxeen]

Thanks for your reply

The problem is that I cannot use a PID controler, I have to build one with AmpOps..

 

[Warpspeed]

I have a project where I need to control water's temperature with a PID controller.

The problem is that I cannot use a PID controller

You certainly have a problem there.

 

[maxeen]

:lol::lol:

Ok, you got me wrong. When you said "Many PID controlllers.." it seemed to me that you meant industrial PID controllers, like you can buy one.
What I mean is that I will in fact use one, but one made of operational amplifiers, resistors and capacitors.. not one that you can buy.

Still, the problem is that I need to control the RMS value at a heater.. and I can't figure any way to do that..

 

[Warpspeed]

Still, the problem is that I need to control the RMS value at a heater..

You could always use phase control, as with a light dimmer.
https://en.wikipedia.org/wiki/Phase-fired_controllers

 

[maxeen]

In this case, the PWM signal has to be syncronized with the AC signal.. any tips on that?

 

[Warpspeed]

Safest way to do this is to first isolate the mains connected part of the circuit from the low voltage dc control section.

You need a triac to switch the ac through your heating element on and off each half cycle, and an opto isolator which contains another much smaller triac to trigger the big one.
Something like this:
https://encrypted-tbn1.gstatic.com/...NbPq5Fc7bBYI-HkWO3KkT0VotZlPfzd5WEIGGcGv92DLb

You can then work much more safely on the low voltage control part of the circuit.

A small low voltage transformer and a rectifier can then supply both dc power and be used to detect mains zero crossings for your PWM control system.

The most important thing is to completely isolate what you are working on from the mains supply, and a transformer and opto isolator can do that.

Then you can experiment further in reasonable safety.

 

[vGoodtimes]

In this case, the PWM signal has to be syncronized with the AC signal.. any tips on that?

Does it? if you can do a slow enough PWM, you could probably get away without using phase-control.

 

[FvM]

You could always use phase control, as with a light dimmer.

According to power quality standards (e.g. IEC 1000-3-2) phase angle control isn't allowed for heating applications. You can decide to ignore the rules for a hobby project, but I would at least try to design state-of-the art full wave switching.

I understand that you want to make an analog controller. That's of course possible, it's the way how most temperature controllers have been designed 30 or 40 years ago. The method is very similar to high frequent PWM. Have a ramp generator with e.g. 1 to 10 seconds period and a comparator that compares the ramp with the output of your PI or PID circuit. The comparator output goes to the soldid state relays with zero crossing control respectively an optotriac.

 

[maxeen]

Thank you for all your answers.

Warpspeed, isolation was intended, using optocouplers.. Thanks for the tip

vGoodtimes, but if the signals are not synchronized, would I have unwanted triggering?

FvM, your answers seems to best fit my problem. I have thought of an IC which genererates a triangular wave to compare with the PID signal, generating a PWM signal. But how the PWM controls the relay? I think I got it! Say I have a triangular with 1s of period. The PID controls the PWM duty cicle, which in turn controls the relay's conduction. I have a solid state relay, its datasheet says that whenever the signal control is high (>3V), the relay will "conduct" when the AC signal reaches zero. But since the period is 1s, the relay will take at most 8.3ms to conduct, which won't affect much of my RMS signal... Am I thinking correcty? Thank you very much sir!

 

[vGoodtimes]

@maxeen -- this would be similar to how microwaves and ovens work. On for 3 seconds, off for 1 second. adding or missing a cycle adds to noise, but isn't significant on that scale.

 

[maxeen]

Of course! I see it now.. thanks!

 

[FvM]

The triangle wave can be generated with a popular two OP circuit, an integrator + a comparator with hysteresis.

It's true that the full-wave switching logic quantizes the output power, for highest accuracy requirements you would use a circuit that corrects the error on average. These operations are easily implemented in a digital controller, I must admit that I never did in an analog design.

 

[Warpspeed]

For heating water, a bit of random time jitter is not going to have any significant effect averaged over minutes or hours.
Tuning the whole thing should be fairly non critical, this is an great learning project that has many very interesting aspects to it.

I suggest at least initially, to keep all the time periods fairly short, it will be a lot easier to to see things happen that way.
Some kind of data logger (with printout) would be a big help too when you get around to tuning the response of the PI loop.

Its very likely to respond faster in one direction than the other, all really interesting stuff.

 

[maxeen]

I have done some simulations, for the water to heat really fast (2 minutes, for instance) I would need about 700V, with this signal decreasing exponentially. But I realized that this is not critical, whenever the output should be higher than 127V, it will be 127V.
It is a project for control system class, and it was and still is a challenge to get everything working together..

Thanks for all your replies, I may post the final result...

 

[vGoodtimes]

Sounds good. You might want a quick-shutdown function for the no-water case. If you have a slow control loop, it might be ok with burning up the heater (if possible).

You might also look at integrator wind-up -- integrating error when control is maxed out and then having to compensate later.