what kind of filtering (if any) should a mains connected heater controller have?

[SArepairman]

I intend to control a heating element by using low frequency high res PWM (~200Hz) into a NMOS transistor.

Mains -> fuse -> Rectifier -> Capacitors -> Mosfet -> heater (resistive)


Should this setup have some kind of filter prior to the rectifier? Not exactly a switch mode supply.

[Whales]

If you're oscillating the power draw, it's a switch mode supply   ie the same problems with noise over the power lines will exist.

Is this for personal use?  Most modern electronics are very resilient to line noise and you might be fine.  YMMV, and you might have legal requirements on noise to meet (esp if you live in a shared building).

[Phoenix]

What power level are we talking here, I assume it's considerable if it's a heating element? There may also be better options for topology if the load is non-polar.

[tszaboo]

Why would you rectify mains? Why would you heat with PWM? Just use a triac and a bimetal.

[Alex Eisenhut]

I intend to control a heating element by using low frequency high res PWM (~200Hz) into a NMOS transistor.

Mains -> fuse -> Rectifier -> Capacitors -> Mosfet -> heater (resistive)


Should this setup have some kind of filter prior to the rectifier? Not exactly a switch mode supply.

http://www.ebay.com/itm/350897868118?_trksid=p2059210.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

[Alex Eisenhut]

Why would you rectify mains? Why would you heat with PWM? Just use a triac and a bimetal.

"and"? Can you explain that please?

[Whales]

TRIACS are silicon switches that happily switch AC.  You can send the PWM to that and skip the whole rectification stage.

Do you need to sync these to the waveforms to be effective, or is that optional?

[tszaboo]

Why would you rectify mains? Why would you heat with PWM? Just use a triac and a bimetal.

"and"? Can you explain that please?

Yes, the sentence could have been "triac or a bimetal". But I prefer not sending high currents through the bimetal. After seeing cheap heaters making scary sparks every few minutes, I prefer solid state mains switching.

[David Hess]

The off-line heaters I have seen in the past used a triac and optically isolated zero crossing driver with pulse modulation synchronized to the AC line.

[eneuro]

Full bridge rectifier loses at 10A to get 2.3kW at 230VAC without ideal bridge drivers might require fan to cool things down, but maybe you plan use fan to improove heating anyway, so it will not be a problem.

If I wanted lower power loses for some reason in driver circuit I would consider AC mosfets (a few in pararell to lower its series RDSON) switch  with current sensor to be able make zero crossing group driving of this heater-mains current sensing and switch on/off in current minimum for a desired number of full 50Hz periods 20ms each.

[Niklas]

Is this a one of or something that will be produced in volume? Heating of for instance water is quite slow even with a 3 kW (13 A @ 230 Vac) element. Is there really a need for that high frequency PWM or can you slow it down? We used a 3 kW heater for a jacuzzi filled with 300 liters of water and had to slow down to approx. 1 switching per minute to comply with the CE standard for flicker. That solution used a BTA25-600B power triac from ST that we bolted to the seating plate of the heating element for cooling purposes. Even with a MOC3063 zero cross detection optocoupler, a 3.3 uF X2 cap was needed between L and N to suppress emission. A snubber network of 100nF X2 cap + 330 ohms resistor was also used.

[SArepairman]

I thought a triac had to switch at zero crossing. I wanted very fine adjust ability of ramp rate on a high voltage heater (400W)

Like, 400 watts, 0.1c / 10 min, relatively linearly, small thermal load (no more then a few hundred grams, well insulated)

If a triac can manage to heat something from 351 to 351.4 degrees c over a long time then it may be a good choice.

I just figured a PWM would be the best way to do it.

*i don't care about a set point. I just want it to move up slowly. the idea is a temperature sweep, not a temperature hold.

[Alex Eisenhut]

I thought a triac had to switch at zero crossing. I wanted very fine adjust ability of ramp rate on a high voltage heater (400W)

Like, 400 watts, 0.1c / 10 min, relatively linearly, small thermal load (no more then a few hundred grams, well insulated)

If a triac can manage to heat something from 351 to 351.4 degrees c over a long time then it may be a good choice.

I just figured a PWM would be the best way to do it.

*i don't care about a set point. I just want it to move up slowly. the idea is a temperature sweep, not a temperature hold.

**and its important that it does not go hottter-colder-hotter-colder, like, the derivative of the graph would never cross zero. If the heating slows down or speeds up a bit, ok, so long there is no cooling.

http://www.ebay.com/itm/220888955443?_trksid=p2059210.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

Don't know if it can do the ramp up, maybe it can be controlled from outside but just for the temperature display and SSR it's fun.

[SArepairman]

I thought a triac had to switch at zero crossing. I wanted very fine adjust ability of ramp rate on a high voltage heater (400W)

Like, 400 watts, 0.1c / 10 min, relatively linearly, small thermal load (no more then a few hundred grams, well insulated)

If a triac can manage to heat something from 351 to 351.4 degrees c over a long time then it may be a good choice.

I just figured a PWM would be the best way to do it.

*i don't care about a set point. I just want it to move up slowly. the idea is a temperature sweep, not a temperature hold.

**and its important that it does not go hottter-colder-hotter-colder, like, the derivative of the graph would never cross zero. If the heating slows down or speeds up a bit, ok, so long there is no cooling.

http://www.ebay.com/itm/220888955443?_trksid=p2059210.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

Don't know if it can do the ramp up, maybe it can be controlled from outside but just for the temperature display and SSR it's fun.

no ebay solution please, its part of a system

[IuriC]

I have an air heating system using a TRIAC and opto isolator, I can set temperature with at least 1ºC step.

Maybe it can be done too. It will rely on your feedback sensor and control algorithm.

[ajb]

I thought a triac had to switch at zero crossing.

Once switched on, the triac will continue to conduct until the current falls to zero, so with a resistive load it always switches off at zero crossing.  You can switch it on wherever you like in the waveform, with the caveat that switching on near peak voltage with a resistive load will lead to large inrush currents and possibly significant acoustic/electrical noise.  A large choke in series with the load would likely be in order if you need the ability to switch anywhere in the waveform. 

[SeanB]

Would not bother smoothing it, just use a mains filter and rectifier, then control the unsmoothed DC with the switch of choice. If the heater is 400W and your load is 100g then you probably will need to use PWM and PID as well, as you will definitely have overshoot with a low mass system and such a high power input. Using a SCR probably would lead to overshoot, so use a high voltage MOSFET and PWM to do the control, trating the incoming as if it is DC. You probably would want a 4u7 400VAC class X capacitor in the input filter as well before the bridge rectifier to handle the noise.

[David Hess]

Using a SCR probably would lead to overshoot, so use a high voltage MOSFET and PWM to do the control, trating the incoming as if it is DC.

How much overshoot can you possibly get in a large heater using an SCR or TRIAC when it will shutoff in a maximum of 1/50th or 1/60th of a second?  Or half of that time if you allow half cycles.

[SArepairman]

Yea, I guess the filter cap is kinda silly with a resistive heater.

So mains -> fuse -> filter -> rectifier -> 16 bit pwm mos @ 200Hz -> heater

again, I would like 0.1C rise over the passing of minutes.
I wanted it to rise much slower then last digit of my thermometer can measure, which is 0.01C

so, this filter, 4.7uF? Common mode choke too?

[SeanB]

How much overshoot can you possibly get in a large heater using an SCR or TRIAC when it will shutoff in a maximum of 1/50th or 1/60th of a second?  Or half of that time if you allow half cycles.

Depends on thermal lag, and just what degree of control you want. For 5C variation it does not matter, but if you want like the OP a 0.1c max variation then you really do want to control the amount of energy you put in to a very fine level.

[SArepairman]

how about 0.001c overshoot. 0.1c is butchery 

again, super slow thermal sweep .

I need a slow sweep so the thing I am measuring will equalize in temperature with a thermometer.

chokes? y capacitors too?

[SeanB]

The whole enchilada of input filter before the bridge, and preferably a large common mode choke and even a line reactor as well to control the impedance of the supply. You will be making a lot of mains noise down the line and preferably you want to reduce it a little.

Though if you are wanting high resolution you probably will have to use a linear amplifier to drive the heater, or use a 1kHz or higher PWM on the drive side.

[SArepairman]

why the need for a higher frequency PWM?

[SArepairman]

oh, do you think a sag of temperature will happen in the down time of 200Hz cycle?

Can anyone help me do basic thermodynamic calculations to figure out how many bits of resolution I need for PWM control to maintain a certain temperature? Point me in the right direction?

Like, a 16 bit PWM @ 1KHz would require external PWM hardware. I figure a double 16bit MDAC, one to generate a triangle wave and one to generate a DC output, fed into a comparator to make a variable duty cycle wave ?
this way 1 reference and common IC, so there should be no drift of duty cycle with temperature.

At low frequencies or less bits I can use the PWM feature of a dspic.

[ajb]

To at least get in the ballpark, you should look for thermal resistivity figures for the materials your setup will be fabricated from.  Thermal resistivity combined with the geometry should give you a thermal resistance (just like electrical resistivity+geometry=electrical resistance).  The complexity of the resistivity to resistance conversions will depend on the geometry of your parts and how accurate you want to be.  Once you have thermal resistance values you can roughly calculate the rate of heat flow through your insulation--in other words, how quickly you lose energy.  This is directly equal to the amount of energy you'll have to put into the system to maintain a given temperature.  If you find the specific heat for the various parts of your system, you should be able to figure out roughly how much energy is stored in the system at a given temperature.  Combine that with your aggregate thermal resistance and you've got the thermal equivalent of an RC circuit, and should be able to figure the temperature decline with no heat input as you would plot the discharge of a capacitor.  You can also calculate how the temperature will change as you put more or less energy into the system.

Of course the devil is in the details, and the validity of any modelling you do will depend on how well you capture those details.  Things like convection in any air spaces in the setup can have substantial effects, for instance.  The simpler solution may be to get someone to model it in SolidWorks for you.