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

[SArepairman]

I have access to the full version of solidworks but I have never played with it much. maybe its time.

[Phoenix]

Do you really need 16bit PWM? For a 340Vdc input (approximate rectified mains) that's a resolution of 5mV (340/65536). The incoming mains or your rectified DC is going to fluctuate far more than that over one period.

[qno]

You want to control the energy the heater is making.

Energy is Power x Time

You can either control time or power.

If you want to control the power you need a lot of hardware.

If you want to control time you need an on off switch.

To heat 1 qubic meter of air on degree Kelvin or Celcius you need 710 J.

[SArepairman]

so how should I go about designing this filter, if I need the whole enchilada?

[qno]

Usually a fliter for switching on ohmic loads is not required.
There are no inductive spikes to filter on a pure resistive load.

If you use a relais for control you might need a small capacitor to filter sparks.

If your controller switches on or off every few seconds a small spike is allowed.

[Niklas]

Switching a few times per second or minute is probably seen as a repetitive action. Compare to a thermostat with a more random switching. Flicker rules might apply, both time and current counts. Also consider the rectified current waveform. Typically not sinusoidal and with peaks

[salbayeng]

I'm not 100% convinced you have a good grasp of the physics of thermal time delays, for example a 400w lightbulb has the smallest thermal inertia I can think of , and a bulb this big has negligible 100Hz flicker when operated on mains AC. 
Thermal diffusivity http://en.wikipedia.org/wiki/Thermal_diffusivity is what smooths out temperature variations as a thermal pulse propagates through a medium. 
If you put your hand in at one end of a half-full bath, while someone else turns a hot tap on and off at the other end, all you feel is a gradual increase in temperature , there is no "negative slope" .

The 100Hz pulsations of the mains would still come through and dominate over any 400Hz pulsations.

Realistic reasons to use PWM on a (large) heater would be to reduce thermal fatigue or minimise buzzing noise (due to Lorentz forces) , and typically 1kHz would be used.

Getting back to your original filter anyway,  I would just use an off the shelf 6A 250v mains filter.
If you really wanted to build your own , then something with typically 1mH to 5mH inductors on each phase and maybe 1uF X class cap across the line on input and output, and maybe 1000pF Y class to gnd on both input lines,  It would cost more to procure these parts than buying an off the shelf filter.

[SArepairman]

I got plenty of those filters around. But I did notice there are like 50 varieties of mains filters to choose from.

What do you suggest? Also, the PIC I am going to use can do 16 bit PWM @ 1KHz by default.

I figured if its only a 100 grams of metal @ 400 degrees C it might cool a bit, but I don't know, I guess I would need to do the drain gain calculation but I am not sure how.

And a light bulb element is in a vacuum, I am not sure how that effects the rate of heat loss. I have a bare light bulb, it's too bright for me to notice any flicker. Maybe I will try with a welders mask, but I know that if its visible then it has to be changing by quite a bit (after all going from white to red hot is not a fine line) I wonder if I can measure it by connecting a photodiode/resistor to a oscilloscope, the change in temperature of a light bulb should be determinable from some equations

[eneuro]

I wanted very fine adjust ability of ramp rate on a high voltage heater (400W)

Resistive 400W @ 230VAC means about 1.7A < 2A current, so easy task for triac or... AC mosfets switch with mains current & voltage sensing.
Why AC mosfets switch not triac... because of AC mosfets switch can be turned ON/OFF when I want, while triac or thyristor when current drops to some level.
So It looks like very easy task to limit this output power to very low levels when you have mains current & voltage sensing while you can estimate in MCU power consumed for heating and you are able to triger switching ON this AC mosfets switch at the end of 10ms half perdiod in the case of 50Hz 230VAC mains, so I do not think that any PWM is needed to controll this heater, but simply monitor mains current & voltage and input only required amount of power by adjusting phase when in each 10ms period AC mosfets switch is turned ON/OFF.
I sucessfully used before custom light dimmer (but classic with diac only higher current rating of used BTA triac) to limit heat power of 2kW 230VAC 50Hz oil heater, but in recent spot welder project I will test also this AC mosfets switch with mains monitoring aproach too, while it does not require any rectifiers and by putting many mosfets in pararell I can get better RDSON than with triac I hope.
I think you could simply use:

mains-fuse-filter-current&volatge_sensor-ac_switch-rectifier-capacitors-heater-temperature_sensor.

Whatever you do still PID controll paremeters or other fancy methods including determined heat transfers models might be needed fro such precise temperature control, but it looks like PWM after rectifieed DC is not needed, while we can use AC phase shifting and predict/set needed amount of input power to heater simply by adjusting delays time.

The main problem might be to know how much energy is needed to do not overheat this thing in changing ambient temperature, I guess 

[Richard Crowley]

How much thermal mass are you talking about here?  Unless you are trying to control the temperature of something the size/mass of a match-head to within 0.01 degrees, using PWM, at high frequencies just seems like gross overkill, not to mention the additional risk of failure from all those extra components.  We typically see simple SCR/Triac switching of the direct mains current/frequency with zero-cross switching.  The thermal mass of the "load" performs all the final "filtering".

[SArepairman]

How much thermal mass are you talking about here?  Unless you are trying to control the temperature of something the size/mass of a match-head to within 0.01 degrees, using PWM, at high frequencies just seems like gross overkill, not to mention the additional risk of failure from all those extra components.  We typically see simple SCR/Triac switching of the direct mains current/frequency with zero-cross switching.  The thermal mass of the "load" performs all the final "filtering".

100-200 grams, probably some kind of copper alloy, fine control to 450C.

does the max temperature change peoples opinions of PWM being unnecessary? I know at elevated temp it will lose heat alot quicker.

[Richard Crowley]

Clearly, you would need to scale the capacity of the heating element to the mass of the "load". A heater of insufficient power will not be able to keep up with whatever losses from the mass, and one that is too large will possibly overheat even with short pulses, and will create an uneven temperature over time.  I still don't see why ordinary mains-frequency, zero-crossing switching wouldn't work.  Not at all clear why PWM or HF would be necessary here?  The thermal mass itself if a fine integrator.