AC Zero crossing detect, H11AA1 troubleshooting

[mrflibble]

Don't you see how LEDs are drawn in the datasheet?

Everybody see this, but this is what is seen on its input-each diode inside must have some reverse voltage and probably is low.

Okay, another way you can think of this ... forget the optocoupler. Just take 2 regular boring diodes in anti-parallel configuration, similar to those two LEDs.

When diode A is forward biased, it will clamp it's voltage to let's say 0.7 V. Or maybe with a larger current through it 0.8 V. Then yes, the other diode is reverse biased and will see that -0.8 V. And in case you are worried about reverse breakdown voltage, then yes both diodes will have a reverse breakdown voltage. But you will never get close to that due to the voltage clamping done by the forward biased diode.

And the exact same thing is going on in that H11AA1 optocoupler, only it's a bigger voltage drop. Quick look at the datasheet: voltage across led is something like 1.0 V - 1.4V, depending on temperature. And if you check Figure 1 in the datasheet you can even see precisely what wraper is talking about.

[eneuro]

And in case you are worried about reverse breakdown voltage, then yes both diodes will have a reverse breakdown voltage. But you will never get close to that due to the voltage clamping done by the forward biased diode.

It is clear from the begining how it work for low frequency, but now lets take crappy LEDs in the same configuration and connect it not to 50Hz AC, but.... hundreds  kHz
Then diode recovery time doesn't count ?
http://www.diodes.com/pdfs/apd/AN-1012_TRRComparison_v0.2.pdf

If this Vf of one diode will protect another no problem, but at higher frequency who knows -real electronic component is not ON/OFF device and many nonlinear regions....so not everything which seams to work at low frequency will work at much higher frequency...
It is interesting if this H11AA1 has such frequency at witch those diodes will fails , eg. do to recovery times?   

[bktemp]

Why should the diodes fail due to recovery time? They will simply keep conducting and slow down the other diode turning on. Since the current is limited nothing bad will happen. It may only stop emitting light, because the parasitic capacitors will take all the current and prevent the diodes from conducting.
If the forward recovery time would be high, then the reverse voltage could reach high values, but on almost all diodes the forward recovery time is negligible. Only high voltage diodes show significant forward recovery effects.

[wraper]

And in case you are worried about reverse breakdown voltage, then yes both diodes will have a reverse breakdown voltage. But you will never get close to that due to the voltage clamping done by the forward biased diode.

It is clear from the begining how it work for low frequency, but now lets take crappy LEDs in the same configuration and connect it not to 50Hz AC, but.... hundreds  kHz
Then diode recovery time doesn't count ?
http://www.diodes.com/pdfs/apd/AN-1012_TRRComparison_v0.2.pdf

If this Vf of one diode will protect another no problem, but at higher frequency who knows -real electronic component is not ON/OFF device and many nonlinear regions....so not everything which seams to work at low frequency will work at much higher frequency...
It is interesting if this H11AA1 has such frequency at witch those diodes will fails , eg. do to recovery times?   

Oh my, recovery time doesn't matter at all in this situation. You do care if it opens fast enough (and that time is much faster than recovery time), not stops conducting as one LED protects another led while conducting, not being in closed state. Secondly, most of the optocouplers are no good after few tens of kHz, therefore not usable at such frequencies. Thirdly, with reverse protection diode in series as you suggested, is when you will get reverse recovery time issue with fast rise time AC voltage and LED can be damaged with reverse voltage.
With diode being in reverse-parallel you can consider that if rise will be so high that reverse recovery comes in play. One diode reverse recovery time will be much longer than time at which another diode starts conducting, therefore overlaping. Because of that always protected from overvoltage.

[aeberbach]

Although, it is not clear why you need ZC detection at all? Its not as if you are dimming lamps here.

I might one day. This thing is controlled by a RFDuino, it's a box with a mains socket and a plug to the wall. Control of temperature (or light) is done by an iPhone app through Bluetooth LE. That's the easy part and it's done already.

[aeberbach]

And then ofcourse in the software you have to compensate the delay between detecting the pulse and switching the optotriac else you are still to late with the switch on.

I'm expecting to lose a very small part of the cycle but that won't matter much, just reduce the speed that the heater can come up to temperature. The PID control algorithm could take care of maintaining temp if that is ignored because it's getting constant feedback.

Depending on how this transformer is wound I expect either 180 degree phase shift or none, so I should be able to ignore that - the zero crossing point remains the same.

[mrflibble]

And in case you are worried about reverse breakdown voltage, then yes both diodes will have a reverse breakdown voltage. But you will never get close to that due to the voltage clamping done by the forward biased diode.

It is clear from the begining how it work for low frequency, but now lets take crappy LEDs in the same configuration and connect it not to 50Hz AC, but.... hundreds  kHz
Then diode recovery time doesn't count ?

Does reverse recover time count? Sure it does, the diode is still the same. It just doesn't have all that much influence here.

Take two red leds from the crusty old parts bin, put them anti-parallel, and add a suitable series resistor. Then connect to a sine wave of lets say 5 Volt or 10 Volt amplitude. Then try it for 50 Hz and for 100 kHz. Do you see anything that you attribute to reverse recovery time of the led?

And fail due to recovery times? Yeah sure, no doubt. Pump 1 GHz into it and it will probably fail. But that's not all that relevant to the use case here. Especially because the use case here involves an optocoupler, and the photodiode part of it usually is bloody slow anyways. So no worries about that reverse recovery time on the LED side. You won't come anywhere near the frequency that reverse recovery becomes a problem. The photodiode will have given up long before that.

[mrflibble]

Just thought of something. I mentioned that 1 GHz jokingly, but even with that kind of frequency reverse recovery will not be an issue. You won't even get any mobile carriers at that frequency for a random crap led from the parts box. Probably something like parasitic capacitance will have spoiled the party already. 

[aeberbach]

Mystery solved! 



My RCD tripping saved my new oscilloscope, and Dave has saved me from doing this again.

[eneuro]

Take two red leds from the crusty old parts bin, put them anti-parallel, and add a suitable series resistor.

I've soldered something like this with random 3mm yellow diodes:
230VAC(L) -> 2x100k + 2xLEDs + 2x100k <- 230VAC(N)


"Unfortunatelly", no mains fuses fired and no smoke smell, so you must be right  and  it passed 230VAC 50Hz test


This configuration above means only about 2.3mA current with 100k resistance, since only one diode (50% "PWM") at average 1.1mA at day light those diodes not such visible, but at dark environment this simple 230VAC power indicator might be usable sometimes (power loses on each of 0.25W 100k resistors only 0.121W, so not so bad).

Anyway, this current is quite small and does H11AA1 works as detector with output pulled up to 5Vcc at direct  input 230VAC (240VAC close to this) and around 100k series resistance?
Such small curents (range 1.6mAmax -1.1mA avg) are fine to triger and detect those zero crossing using H11AA1?
Did you tried to count lets say 1 hour zero crossing count and compare with expected number of crossing if perfect sine 50Hz wave AC power supply was?
This test should give around 360k /hour such zero crossing. It will be interesting to compare it to Hall based methods of mains current monitoring 

[mrflibble]

Mystery solved! 



My RCD tripping saved my new oscilloscope, and Dave has saved me from doing this again.

Good thing that you solved it, and thanks for sharing the problem/solution. Incidentally, this is a good example of why it helps to show a photo of the measurement setup. Chances are pretty good that if someone sees a mains connected <whatever> being measured with a scope they go "Heeeey waitaminute, how is that connected exactly?"

[mrflibble]

I've soldered something like this with random 3mm yellow diodes:
230VAC(L) -> 2x100k + 2xLEDs + 2x100k <- 230VAC(N)


"Unfortunatelly", no mains fuses fired and no smoke smell, so you must be right  and  it passed 230VAC 50Hz test

Bonus points for doing the experiment. If you happen to have something that can generate a similar waveform at higher frequency (100 kHz) then you should see the same boring effect of no explosions.

Such small curents (range 1.6mAmax -1.1mA avg) are fine to triger and detect those zero crossing using H11AA1?

Are you making me open up the datasheet again? From a quick read yesterday I recall the answer to that is yes. After all you have ~ 1 mA of photocurrent going into the base of that photodiode/phototransistor, get some conversion losses, and then apply the transistor gain ... yeah, 1 mA should be enough to make that transistor pull the output low. Looks like it will be switching fairly slow at 1 mA led current though, but for 100 Hz fast enough (I think). To know for sure one would have to check the datasheet. Well, or build the circuit and just measure what it does.

Datasheet attached for convenience.

[eneuro]

Looks like it will be switching fairly slow at 1 mA led current though, but for 100 Hz fast enough (I think). To know for sure one would have to check the datasheet.

Maybe missed something reading its datasheet, but wasn't able to figureout its cutoff frequency or turn on/off times  which is easy to find for eg. for slow PC817 or fast 6N137 

Looking for something very fast to disable Enable pin in 6N137 ( http://www.fairchildsemi.com/datasheets/6N/6N137.pdf) in the case of active/shorted half part top/bottom  of 1 of 3 half-bridges in 3 phase optoisolated from MPU motor controller and it could be nice to have such AC capable input like in cheap H11AA1, but have no idea for the moment which can be its switching frequency limits, or will find fast double input optocoupler and make pcb connections to have those diodes in parallel like in H11AA1  and make tests and compare those switching limits  while need 6 such AC detectors per controller to disable without MPU knowledge 6N137 enable pin to switch off N-channel mosfets 3A driver to prevent short cut, so need to know safe delay time in AC current detection circuit, but in my case 10mA current is not a problem since I need detect not zero  but current maximum   

[wraper]

What hard to find?

[eneuro]

What hard to find?

It looks like it depends on manufacturer and in ISOCOM datasheet this Tr~4us and Tf~3us are easy to find 
http://www.tme.eu/en/Document/9d740051576413a595b2de55776108d8/H11AAx-I.pdf
Those times at given output current looks fine for me 

[mrflibble]

What hard to find?

To be fair, that "Switching characteristics" section you show is from the H11A1 datasheet, found here: http://www.vishay.com/docs/83730/h11a1.pdf

A similar section is not listed in the H11AA1 datasheet, found here: http://www.vishay.com/docs/83608/h11aa1.pdf

In the H11AA1 datasheet more reading between the lines is required, which usually doesn't help. Anyways, good thing you posted this, because with some luck you can use the H11A1 datasheet to fill in some blanks on the H11AA1 datasheet.

And while browser Mouser I noticed that for example Fairchild also produces the H11AA1, so you can shop around to 1) find a cheaper option and 2) find a manufacturer with more complete datasheets (which helps in design IMO).