Dimming Thermostat - general questions

[Scampy]

Hi,

Thanks for the reply. Dimming stats used for keeping reptiles tend to drive either normal incandescent bulbs, or ceramic elements like these



Yes the intention would be to switch 240v AC either through SSR's or zero crossing driven triacs.

So If I follow your drift, a PID routine would still be required, but rather than a low frequency pulse, the filter is used to determin the pulse width in a high requency PWM ?

[HenrikOlsson]

Yes the intention would be to switch 240v AC either through SSR's or zero crossing driven triacs.

Won't work very well, if at all, with "high frequency" PWM. For "dimming" AC you would need to create a phase angle control.

So If I follow your drift, a PID routine would still be required, but rather than a low frequency pulse, the filter is used to determin the pulse width in a high requency PWM ?

First part yes, second part no.
Using proportional control only will most likely give you the result I explained in my previous post. You will need at least PI.
As before, you can not use a normal SSR to power a AC device and simply feed a "high frequency" PWM signal to it to "dim" it, it's not like dimming a LED with PWM. You need either a voltage controlled phase angle controller and drive that with a voltage created by low pass filtering the PWM output. Or you create a LOW frequency PWM signal with a period in the seconds range (which I believe is exactly what I beleive Darrel created for you in the version you have already) - and it's like that for a reason.

If you where to rectify and filter the incomming AC you COULD then use a high voltage MOSFET or IGBT to PWM the load but, well....

/Henrik.

[Normnet]

I did a similar project; A PID algorithm temperature controlled laminator.
It used HPWM to operate a zero crossing solid state relay.
The PIC HPWM at lower settings operates the SSR more off than on and visa versa at higher settings.
Darrel's low frequency PWM would be good however mine worked with regular HPWM.
I included the light in the rocker power switch with the heater to give a visual indication of the on/off pulsing of the SSR.

Norm

[Scampy]

Don't light dimmers work by PWM ?

[HenrikOlsson]

Normal dimmers for incandescent lamps doen not work with "normal" PWM, no. They work with phase angle control.

They're using a triac. A triac, once turned on, doesn't turn off untill the current thru it goes to zero - which it does each zero crossing - you can not turn it off randomly. If you run a "high frequency" PWM signal into to the gate of such a device the lamp would flicker like crazy because the PWM signal isn't sync'ed to the AC mains. Sometimes it turns on at the end of the AC cycle, some times it turns on the start and some time it turns on the middle.

The way they work is that they detect the zero-crossing and then waits for a certain amount of time untill it sends a trig pulse to the triac which then turns on and conducts untill the next zero-crossing. For 60Hz mains each half-cycle is 8.333ms. If the delay between the zero-crossing and triggering the triac is 4.16ms the lamp would be at 50% (well not really since it's not linear but you get the idea).

Your typical AC SSR contains a triac on the output-side and works the same way. Once triggered it doens't turn off untill next zero-crossing.

The dutycycle of the PWM signal feeding your SSR would not relate to the relative power sent to the heaters because the current to the load would be turned ON at the rising edge of the PWM signal but not turned OFF at the falling edge.

/Henrik.

[HenrikOlsson]

Hi,
I created the following picture to further illustrate the issue with running a "normal" PWM signal into a triac or a random firing or random turn-on SSR:



I choosed a random PWM frequency (which was easy to draw) relative to the 60Hz mains but it doesn't really matter.
The triac/SSR is ON during the purple sections which, as you can clearly see, is a lot more than the 12% dutycycle of the PWM signal.

I hope this helps illustrate my point.

/Henrik.

[amgen]

you have 120- half cycles per second that you can cycle....... you can break that down to 60/half second or 30 per quarter second....... if you consider controlling for 4 times a second, then you can go from 1 cycle on and 29 cycles off up to the opposite on/off ratio.......
that's 250 milliseconds each control loop.....so say 10 milliseconds on pulse and off 240 milliseconds for minimum......
that's just pin hi, pin low functioning with some kind of counting loop and changing the ratios accordingly....
That would use a zero-crossing SSR for ease of use.
There could be a I cycle error unless you use some kind of zero cross detector to sync up...... probably not necessary because the error would probably average out.