Mains zero crossing.

[orolo]

I've been reading the depletion mode mosfet section on The Art of Electronics, and decided to place some of these devices (LND150) in my next order to Mouser. As a first experiment, I'm thinking about a zero cross detector for mains:



Here the mosfets play the two roles from TAoE, input protection for the comparator and discharge for the mains capacitor. Additionaly, they provide a neutral referenced bipolar power supply which, if I pulled it right, is an interesting twist.

Is there something I've overlooked that could result in pirotecnics? I haven't found any spice models for small depletion mosfets, so I'm basing myself in calculations and a very limited experience with these devices. Of course, in case I build the circuit, I'll use an isolation transformer. Given the nature of the power supply, all the components should be very low power. With the comparator there's no problem; I wonder if the optocoupler could be replaced by magnetic coupling, to reduce power.

[ebclr]

You can add 2 more ic's on your mouse list for research zero crossing

http://www.mouser.com/ds/2/427/il4116-241328.pdf

http://www.mouser.com/ds/2/149/MOC3043M-195994.pdf

[BrianHG]

If you want a 0 crossing detection circuit for timing or directing a microcontroller, why not just use an AC input optocoupler with 2 back-back series zener diode to control the 0 crossing window size with a series resistor and pull-down on the opto-input.  The output is still a transistor out.  You wont need any op-amps or mosfets or anything else.

Something like a LTV-814 or FOD814.

[Dubbie]

If you want a 0 crossing detection circuit for timing or directing a microcontroller, why not just use an AC input optocoupler with 2 back-back series zener diode to control the 0 crossing window size with a series resistor and pull-down on the opto-input.  The output is still a transistor out.  You wont need any op-amps or mosfets or anything else.

Something like a LTV-814 or FOD814.

Thats how I did it on a dimmer bank I built once.

Worked fine.

[orolo]

Wow, there seems to be an IC for every conceivable task. I'm doing this to learn about depletion mosfets, not to build a zero crosser, but I'll get some of these optocouplers as reference to test how precise the circuit is, if it works. Thanks for the info!

[james_s]

In many cases a simple zener diode and resistor is sufficient.

When I built a DMX dimmer pack several years back I used a comparator and got a very clean and dependable zero cross signal. I don't recall the exact circuit offhand but I can dig out my schematic. No need for a special purpose IC.

[frog]

Interesting use of depletion-mode there.  So as I see it the FETs are switched off when significant current starts to flow, so at high voltages you have more or less a current source/sink.  If the resistor between the two FETs is the right value then you should be able to dissipate most of the excess voltage here so the FETs aren't working too hard.  Depending on your priorities it might be preferable to grab all the charge you need when the AC voltage is relatively low and then switch off altogether for the rest of the half-cycle, this could give you better efficiency for some extra complexity in the design.

Oddly you don't see depletion mode FETs that often.  I suppose an Rds of 1000 ohm could be off-putting if you're used to something less than an ohm.

As a cautionary note, I'd look at the startup conditions if you happen to switch on in the middle of an AC cycle.

[Cerebus]

Oddly you don't see depletion mode FETs that often.

You don't see depletion mode MOSFETs all that often; of course, you only see depletion mode JFETs.

[orolo]

Interesting use of depletion-mode there.  So as I see it the FETs are switched off when significant current starts to flow, so at high voltages you have more or less a current source/sink.  If the resistor between the two FETs is the right value then you should be able to dissipate most of the excess voltage here so the FETs aren't working too hard.  Depending on your priorities it might be preferable to grab all the charge you need when the AC voltage is relatively low and then switch off altogether for the rest of the half-cycle, this could give you better efficiency for some extra complexity in the design.

That's the idea, exactly. The resistor between the FETs is to define gate bias, and could be maybe omitted; most of the dissipation work will be by the FETs themselves. As long as it's only 1-3mA, I guess it should be ok. I'm beginning to like the idea of the power supply, in a Big Clive kind of way: transformerless and capacitorless. My idea was to "improve" the design, poundland style. Turning off the FETs at the right moment is a magnificent idea, though: if the voltage of the supply rails is high enough, the FET is turned off for the rest of the half cycle, saving a lot of power. Very good.

As a cautionary note, I'd look at the startup conditions if you happen to switch on in the middle of an AC cycle.

Oops, I didn't think about this, and it's a classic failure mode. I didn't think about the turn-off trick, neither. Thanks a lot for the ideas!