Lamp Dimmer - MOSFET based

[Hell-e-Guy]

Hi Guys,

Currently I'm starting to design smart light dimmers for my house (433Mhz operation). I know the conventional circuits use triacs and zero detect methods for rectification.
I was thinking to fully rectify the mains voltage and then use a MOSFET with free running high frequency PWM for dimming.
This will eliminate the need for a zero crossing circuit and the power needed to open a MOSFET is less then a triac...a zener circuit as a simple power supply will dissipate less power.
Am I missing something or is this feasible?

Hell-e-Guy

[jeffs555]

The reason triacs are used is because of cost. HV mosfets are generally more expensive than triacs. Since you said "mains voltage" I suspect you are in a 220v country, so you would need 400 volt mosfets.

You might want to take a look at this app note from microchip that uses a PIC and a triac. http://ww1.microchip.com/downloads/en/AppNotes/91094A.pdf

[Hell-e-Guy]

Thanks jeffs555,

Can you please explain how the zero crossing is done on an ordinary GPIO in the circuit?

Hell-e-Guy

[jeffs555]

I haven't used that circuit, just ran across it. They don't explain what is happening very well. Basically, the junction between d1 and d2 varies between 0 and 5 volts, and changes at the zero crossing point.

[ice1234]

U Can Use Op Amp To Detect Zero Cross Pt.

[GeneSS]

The zero crossing circuit is used to minimize RF interference not necessarily for rectification. Without it, when ever you turn on the mosfet or triac, and the current is not at a zero crossing, you generate a very steep wavefront, similar to a square wave. This generates tons of harmonics. The early light dimmers did not incorporate this feature (zero crossing detection), and they would render AM radios, within 100 ft, useless.

Gene

[andrewm1973]

In the case of dimming lamp you don't switch the active device at the zero crossing point. You dont get any dimming if you switch full cycles. (or you get dimming with a 50Hz step).

Zero crossing switching is used to reduce EMI and phase problems in other cases though.

[Hell-e-Guy]

So do you think that my proposal is feasible?
Turn the mains AC to DC with diode bridge, use a 400V mosfet to PWM at variable duty cycle in order to get the dimming effect. This way no zero crossing is needed to time the triac triggering.

Hell-e-Guy

[jeffs555]

It could be made to work, but if it is asynchronous, your switching frequency would have to be high compared with the line frequency. This would multiply your problems with switching noise. Higher switching frequencies would also increase the power disipation in the mosfet.

I don't know why you are so worried about a zero crossing detector. You did say "smart light dimmers", so I assume you are going to use some type of microprocessor. If you don't need the micro isolated from the AC, like the microchip app note, a zero crossing detector is as simple as a resistor and zener. If you want the micro totally isolated from the ac, a dual opto-isolator with one section driving the triac, and the led of the other section driven with a high value resistor from the AC line will give you isolation and a zero crossing detector.

Here is another good article to read about dimmers. http://www.epanorama.net/documents/lights/lightdimmer.html

[Comatose]

If you're willing to invest some engineering time into the plan, then running a high switching frequency with a MOSFET is no big deal at all. You'll need a big enough capacitor on the input after the bridge to smooth out your current draw. This is larger and moreexpensive, which is why its not the typical way to do these things, but it'd work fine.

[Hell-e-Guy]

Is it true that DC current shortens the life span of incandescent light bulbs?

[andrewm1973]

For low wattage resistive loads there is no legal problem with phase angle firing an scr/triac. It is very cheap and easy to do. That is why everyone does it that way.

However large loads (particularly inductive ones) you can not do it - so they do oft employ FETs/IGBJTs to PWM. So in answer to your "is it feasable" - Yes. Is it practicle and ecconomical for small resistive loads like lamps - No.

In answer to the "Does DC shorten the life".

Theoretically yes. Electromigration would make the fillament thinner at one end. How this works out in a real life situation would require the input of a lightbulb expert.

[Dan Baldwin]

Running light bulbs on a filtered DC power supply such as the one you are planning on building could shorten the life of the bulbs. I'm not sure where you are, but as for here in the states, our AC is 120 volts (more or less) RMS. That means that to get the equivalent amount of power through a resistive load such as a light bulb, you would use 120 volts DC. If you rectify and filter that 120 VRMS AC, you will end up with the peak voltage, which is about 170 volts. When you put a load on your filtered DC supply, the voltage will drop a bit, but it will likely still be considerable above the 120 volts you are looking for.

Bulbs are used on DC all the time, such as automotive bulbs.

Dan

[GlowFly]

Is it true that DC current shortens the life span of incandescent light bulbs?
I've been through this several times before with industrial equipment I've designed. Low power incandescent bulbs (i.e. thin filaments) DO seem to suffer from considerably reduced life over that achieved on AC. This has been measured over the life of thousands of replacement bulbs. It got to the point where we designed a test rig to establish our own reference for AC vs DC operation. Later the bulb manufacturers confirmed our diagnosis.
Eventually we designed a new system that used LEDs! :D
--
Steve

[andrewm1973]

Running light bulbs on a filtered DC power supply such as the one you are planning on building could shorten the life of the bulbs. I'm not sure where you are, but as for here in the states, our AC is 120 volts (more or less) RMS. That means that to get the equivalent amount of power through a resistive load such as a light bulb, you would use 120 volts DC. If you rectify and filter that 120 VRMS AC, you will end up with the peak voltage, which is about 170 volts. When you put a load on your filtered DC supply, the voltage will drop a bit, but it will likely still be considerable above the 120 volts you are looking for.

Bulbs are used on DC all the time, such as automotive bulbs.

Dan

I would have assumed that the person with the original intent to PWM the light bulb would have been doing a maximum duty cycle near 0.707 if using filtered DC or more sensibly not filtering the DC and leaving the peaks and dips (which would averave 0.707)

It's a bit obvious that overdriving a light bulb by that amount would shorten the life.

So I think the problem being refered to may have been what happens when them little electrons race around really fast and bump into things in the filament.

There is debate over if this is real in things as large as light bulb filaments

[Hell-e-Guy]

Hi Guys,

Thanks for all the wise answers. I Thought I can outsmart them electrons, but they are around a bit longer then I am...
I'll go with the ordinary triac based design, with a zener diode zero voltage crossing circuit and just add soft on/off function and a remote control function.
I want to use a zener diode based transformer less power supply but my 433MHz receivers sink around 20mA... This may push the power supply to the limit but I think this is possible.

BTW, do you know of any through hole zeners with 50uA zener curent?

Hell-e-Guy