Mains voltage zero crossing detector with optical isolated output

And then some EPA dude makes a new law that outlaws your dropper because it guzzles 50mW too much :-)

Very brazen folks use a (small!) Y-rated cap and use that to do an el-cheapo transfer. Sans optocoupler. But one has to know what one is doing or there'll be grief.

One approach, where accuracy is important (such as phase control) is to create a DC supply on the 'hot' side and pass the zero crossing as a pulse straddling the zero crossing, using a high speed optocoupler such as the venerable H11. You don't need much in the way of dropping resistor since the LED drive current is only drawn for some tens of microseconds at the zero crossing (while the input voltage is within a few volts of zero). Takes a few more components though (on the hot side).

The simple-simon cheap approach is to use an AC-input optocoupler and a dropping resistor, and a pullup at the micro. But speed of the opto vs. watts in the dropping resistor vs. pullup vs. accuracy becomes a consideration.

In a consumer app I am sure they'd do it with a safety-rated Y-cap. For countries with polarized plugs such as the US you only need one, otherwise two. The rest is super cheap. The Y-cap is the most expensive part, at around a dime in qties. Of course the circuit ground must be tied to PE for this to work unless you do it differentially.

For higher precision this would either require some gain on the line side or some uC-smarts on the system side to find the middle of the window.

John S has Spiced the totem-pole opto thing. With 100K to the AC line, it outputs a square wave that lags the sine wave zero crossings by about 400 us. Adding a simple RC phase lead thing on the input side can nail the zero crossings.

Or the Tektronix heat shrink trick we talked about a few months ago.

Cheers

Phil Hobbs

Hmm ... ... can't remember. For LF signal transfer?

Here's a refresh (/CAS before /RAS) for your memory:-

Best regards, Spehro Pefhany

aka "gimmick" ...Jim Thompson

is

riac

me so

type

nt

.com

r...

dra...

Can you use spice to understand? Personally I liked JL's circuit. You'd want to add some serious filtering to the AC input.

George H.

Yeah, why not a transformer?

George H.

is

riac

me so

on

se

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 ... to power

Must be wire wrapped on input line, (a gimmick) Heat shrink for 'isolation'.

George H.

For "Is AC there or not" it's ok but I would be careful using that for precise zero-crossers. On single-phase line-neutral type setups it'll probably work, on 240VAC or some European systems maybe not so well.

For detecting the presence of power. You heatshrink a piece of wire to the power cord inside the case, and use (say) a two-section RC lowpass to get rid of the junk and EMI. It's easy to control the phase shift, and you don't need to bring the mains voltage onto your board. Vladimir hated it, but I thought it was pretty.

Cheers

Phil Hobbs

We power our benchtop delay generator from an external 24VDC wart, so we don't get a line trigger. We considered all sorts of radical ways to do that (phase lock to wart ripple, electrostatic antenna, whatever). We would up requiring a second small AC wart to do line trigger.

Modem technology to the rescue.

Well, so it has two warts then. ;) It's a pretty nice box otherwise though--much better than my old SRS one.

Cheers

Phil Hobbs

That's mostly useful for slow events, such as telephone ring voltages. I used that very part in my 800-number killer. ...Jim Thompson

We've been left alone quite nicely since the Do-not-call registry came into place. The occasional "poll" or something, but that's it.

Here we've been plagued with "polls" using local numbers, so I'm contemplating a white-list approach to call-blocking. If you're not on the white-list you'll have to enter a roving code to even get to voice-mail ;-) ...Jim Thompson

Screening all calls? Yuck.

Next step would be that nobody can enter your house unless fingerprinted :-)