Precision mains zero cross detector

[cupoftea]

Hi,
Any improvements on this precisions mains zero cross detector.
We read the pulse into a micro, time it..then half the time is the zero cross.

LTspice and pdf schem attached

 

[KlausST]

Hi,

What are the requirements?

To the circuit:
* The inputs have not identical input impedance, thus you get bad common mode rejection ratio.
* U4 at this place is useless.
* Better place it after the rectifier... to compensate for it's jump in output impedance
* There is no filter in the whole circuit, thus you amplify all noise, the output is fast, but jittery.
* What's the use of U3 circuit
* U5 circuit is nonsense. It has no output. And the diodes makes no sense.
* it does not switch very close to zero cross, I guess rather with 5V offset.
* it's a fast 7ns comparator, but the Opamps are slow with 0.4V/us (or 2.8mV/7ns)

So it may work. It depends what you want to do with the signal.
But you should check if the function (including errors like offset, drift, noise...) meet your requirements.

Did you simulate it? Can you show the simulation results?
Although the simulation tools don't need power supply capacitors,
I recommend to add them in a schematic just not to forget them in the real circuit.
(The real circuit won't work properly without them)

Klaus

 

[cupoftea]

Thanks, yes the sim works well (the sim is attached with the top post) but thanks for good advice as caps will be needed.
Woops, sorry about U5, i forgot to delete it from my mains peak detector cct, which is what this is re-hashed from.

--- Updated Aug 25, 2021 ---

The inputs have not identical input impedance, thus you get bad common mode rejection ratio.

...thanks, so you mean use an INAMP instead of a diff-amp?

Do you think there will be big common mode noise, and it will end up ruining the zero cross signal?
The "amplifier" actually attenuates by some 100x, rather than amplifying, as you will know.

 

[KlausST]

Hi,

...thanks, so you mean use an INAMP instead of a diff-amp?

Do you think there will be big common mode noise, and it will end up ruining the zero cross signal?
The "amplifier" actually attenuates by some 100x, rather than amplifying, as you will know.

There is no connetion between AC input and your circuit GND, thus they float with respect to each other.
It´s not a little "noise" I´m concerned of. It can be anything from HIGH DC in any direction (electro static voltage) to HF.

An INAMP will improve common mode rejection, a filter attenautes HF.

No requirements, no validation. Just guessing.

Klaus

 

[cupoftea]

Hi,
Can you think of any improvements on this zcd cct for accuracy of zero crossing and cost?
LTspice sim and jpeg schem provided.

 

[wwfeldman]

assuming i read your schematic correctly, you have the inverting input of the op amp
1 diode drop from 5V and 1 diode drop from ground (BAT54 in both cases)
what am i missing?

according to digi-key, BAT 54 is obsolete

 

[KlausST]

Hi,

You surely know that you still did not give any hint what "accuracy" means for you...nor what the ZCD signal is used for.

Klaus

 

[cupoftea]

Yes thats right, though the SI8610 can apparently be OK with this....?....according to datasheet....and presumbaly no long term effects not mentioned in the datasheet?

 

[wwfeldman]

assuming i read your schematic correctly, you have the inverting input of the op amp
1 diode drop from 5V and 1 diode drop from ground (BAT54 in both cases)
what am i missing?

according to digi-key, BAT 54 is obsolete

much to my chagrin and embarrassment, i realize what i missed
the diode only conduct when forward biased.
if the sample voltage goes above 5V + a diode drop, that diode conducts, clamping the voltage to one diode drop above 5V
likewise, if the sample voltage goes below 0V - a diode drop, that diode conducts, clamping the voltage to one diode drop below 0V

 

[betwixt]

Just an idea - not tried but very simple to implement:
There are devices designed specifically to look at polarity reversal at their inputs and to produce a rapid change at the output at TTL levels. They are RS485/422 line receivers. Maybe worth experimenting because they cost pennies and typically have 200mV voltage detection levels.

Brian.