Sensing mains presence as a logic signal

[ppppenguin]

I need to be able to sense the presence of mains as a 5V CMOS logic signal. Here's my first thoughts. Any comments?

The class Y cap and 22M resistor (it will be a VR37 series, rated at 2500V) are really belt and braces isolation, the current should be around 10uA with mains present. Since it will be feeding a CMOS load the high impedances won't be a problem.

Obviously if mains is reversed it won't work. This isn't a problem in my intended application but I suppose another 1n +22M to the other side of the mains would solve it.

[Refugee]

I would just use a 5V wall wart.

[Mark Hennessy]

If you wanted to be "belt and braces", you could add a series 10k resistor at the input of the gate, but after the catch diodes.

This limits the current that could flow in the intrinsic diodes should they have a lower Vf than the IN400x you've got (not impossible), and should they switch faster than the 1N400x (quite likely).

I've never looked up the current rating of the intrinsic diodes for regular 4000-series - I prefer not to rely on them for anything. On many ICs, specific ratings and guidance is given, but even so, it strikes me as dubious practice to include them as part of the circuit.

I'd probably use 1N4148 for the catch diodes (faster).

Good call on the VR37. I suppose a MOV between L and N might be worth considering if any doubt lingers...

At 50Hz, 1n looks like 3MΩ; 25M and 470k gives 6V peak, so you should be OK after a fair few cycles.

An easier way to control time constants might be to let the first gate ackle at 50Hz and use that to discharge a C - said C can charge in the absence of cycles, and can do so quicker if desired. That's the sort of thing that's needed for power-fail detect in audio amps, where you need to release the loudspeaker relay before any pops and clicks are heard. I usually just have a simple transistor (fed from a sniff of a secondary winding) to discharge the C every cycle...

Presumably the CMOS IC has Schmitt inputs

[boater sam]

LED and photocell?

[Mark Hennessy]

Yes, I've used optocouplers for this sort of thing. The LED in even the most basic types is reasonably sensitive, so you can get away with just a couple of milliamps. For simplicity, I'd just use a diode and resistor (~150k) on the mains side (and an inverse parallel diode across the LED in case the series diode leaks). The transistor will be switching off 50 times a second, but a cap in parallel with that works well. A pull-up resistor charges the cap in the absence of mains, and a Schmitt CMOS input takes it from there.

I use optocouplers in unexpected situations - sometimes they are just can be the easiest/cheapest way to level-shift...

[ppppenguin]

Thanks Ref, hadn't thought of using a wallwart. That's a nice KISS solution for what will be a "one off". It also means I don't have to make sure the mains side wiring is safe. I'm not worried about response time in either direction so the decay time of the wallwart output wouldn't matter. I'd use a series resistor to limit the current in the CMOS catch diodes. possibly have my own catch diodes too.

Mark, your thought of a 10K series resistor before the CMOS is a good one. The diodes could easily end up as 1N4148 anyway. I really should have a reverse polarity diode from the 22M resistor to 0V. This means that on -ve half cycles the input to the rectifier is defined, rather than possibly up to full mains. In practice it won't get anywhere near full mains because of stray C but better to be certain.

In case you're wondering what all this is about, it's for the AV router at the Museum. When the head end system is switched to standby the outputs need to go to defaults. The standby switch removes mains from the COW, DVD player and monitoring, leaving the freeview boxes, RF distribution and Aurorae active. The defaults are currently off-air on all 3 channels. The new system will be switchable from off-air to testcards by an extra switch on the control panel.

Assuming I use the Invotron routers I've devised a cunning circuit using a couple of 4051 analogue 8:1 muxes to do the defaults. Soon I'll draft the schematic properly in Orcad. Makes it easier to build when you've got a properly annotated schematic in front of you and also makes sure it's documented.