zero cross pulse circuit sometimes goes wrong

[cupoftea]

Hi,

We have a zero cross detector , one on each phase to phase, for a 3 phase delta supply. We are just operating it with one singe 60Hz phase (85-125vac) at the moment. We are using a single phase AC power source. (with settable frequency and amplitude, etc)

The ZCD is a very common one that I have seen before, but I can’t show it here. It is all discrete, using BJTs, diodes, Zener, res and caps.

It gives a zero cross pulse just before the zero cross.

On our other product, the 3 phase star 240VAC phase system, this zero cross detector seems to operate fine, giving a pulse some 400us away from the zero cross.

But when we put in the single phase input at 85-125vac, in the 3 phase delta product, it sometimes goes wrong. The expected (acceptable) operation, which we usually see, involves the rising edge of the zero cross pulse getting nearer to the zero cross as the mains is increased from 85vac to 125vac. This is also the operation that the simulator shows for it. However, sometimes a particular channel will show the reverse of this, ie, the zero cross pulse getting further from the mains zero cross as the mains is incremented from 85vac to 125vac. Also, the rising edge of the zero cross pulse ends up too far from the zero cross when at 125vac (about 600us instead of 300-400us).

So anyway, sometimes the circuit works properly, and sometimes not. On the 240VAC 3 phase star system, we haven’t yet seen it go wrong on that.

We are thinking its possibly one of the following…

1…Something to do with the diff probes leakage current to earth.
2…Something to do with the single phase AC power source
3…Something to do with the input common mode choke , in that with just one phase into it, it will show more inductance, and there is ringing at the zero cross which affects the zero cross detector.


Do you have experience of the same problem?, a zero cross pulse sometimes going wrong like this?

 

[D.A.(Tony)Stewart]

I have yet to see any evidence of capacitance sensitivity with voltage in film caps.

They are known to be made in wide tolerances such as 10% due to processing method and sorted to better tolerances.
There is also a voltage-sensitive self-healing mode where a contaminant particle in the gap. That may reduce the voltage breakdown in one of many layers in a tiny spec of the total area. These materials are rated for very high E-field strength 150kV to 300kV/mm.


ref info https://www.nrel.gov/docs/fy19osti/71386.pdf

 

[BradtheRad]

Even with plain resistive (ohmic) loads, all 3 sources must be equal amplitude when Amperes are measured for zero crossings. This simulation has everything balanced except one AC voltage source is greater by a mere 1 percent. (Dots confirm AC generators are oriented properly).

Notice Ampere waveforms are not in phase with voltage waveforms.
If a supply is 3% low, then its Ampere waveform is in opposite phase to its voltage!

Link to animated Falstad simulation running above schematic:

https://tinyurl.com/ynhnahqd

 

[KlausST]

Notice Ampere waveforms are not in phase with voltage waveforms.
If a supply is 3% low, then its Ampere waveform is in opposite phase to its voltage!

I´m having difficulties with accepting these statements.

Imagine a pupil takes this sentence to his electronics teacher. Could the teacher agree?

It´s clear what you mean.
But somehow it´s comparing apples with pears.

In school we learn: with resistive loads voltage and current is in phase. Always.
And that´s true.

But here you talk about a single voltage, but not the current caused by this single voltage,
Instead you indeed talk about two combined currents.

***
What does "opposite phase" mean? 180°?
I doubt one can get 180° with a 3% voltage difference. But indeed I did not do the math.

Klaus

 

[BradtheRad]

I myself don't fully understand what I'm seeing. Very well, picture Bradley as a pupil bringing simulation results in front of teacherss. The simulations are built with simplified components, so perhaps 1 percent does create a difference among 3 ideal generators. However in real-life I don't suppose it makes sense because 3-phase is desired in industrial installations for its stable performance.

The yellow scope trace is current through a generator. It aligns with voltage waveform in a 1-phase by itself with resistive load. Furthermore for the longest time I've drawn demo 3-phase systems where 3 generators matched and each generator's current aligned with its own voltage.

However it didn't occur to me to set up an imbalanced 3-phase system, and see whether 3 generators jostle each other voltage-wise or current-wise. I did so because our experienced OP reports that his workable technique for sensing zero-crossings in 1-phase, has a problem when applied to a 3-phase system. I agree that resistive loads have current waveforms matching the voltage waveform, yet current in the generators might throw off voltage-sensing devices if the devices depend on waveforms within the source generators.

 

[KlausST]

Hi,

I guess we are talking about a single source system with 3 phases (nominally 120° phase shifted each)

***
somehow off topic:

A totally different problem is when 3 generators (each 3 phases) are connected in a system.
Here a voltage mismatch will cause a lot of phase shift.
(when a synchronous generator is connected to the grid, you can adjust V-I phase shift by modifying the stator field which causes the generated voltage to change)

Klaus

 

[D.A.(Tony)Stewart]

When I define ZCD signals, I assume we are talking about the voltage zero-crossing-detection relative to Neutral, used as a timing or phase reference. Some regulators may use "zero-valley switching" and also SCR, Triac switches shutoff at zero-crossing current.

One must realize that modelling Delta R loads will cause the other phase currents unlike Wye or "Y" loads relative to Neutral. These sources still see linear resistive loads but the Delta reference is shifting for voltage, so we must assume Phase to Neutral voltage for sensing ZCD as a reference.

 

[std_match]

I need a simple block diagram with the single phase AC supply, all phases of the device, the common-mode choke, the switching elements, the ZCDs, and any ground or neutral connection.
Post #1 says the the ZCDs are "phase to phase".
Post #26 assumes that they are phase to neutral, and it got "thumbs up" from the OP.
Very confusing.

 

[cupoftea]

We have developed a better zero cross detector, but it depicts pretty well bang on the zero cross....but only every other zero cross...ie, every 20ms for 50Hz. It is said that you cannot process it and say the midpoint zero cross is half way between the samples...because the mains may siimply not have constant frequency, is this correct?....ie, a mains zero cross detector which only detects every alternate zero cross is useless, as it cannot be relied upon for detecting the inbetween zero cross?

 

[KlausST]

It is said that you cannot process it and say the midpoint zero cross is half way between the samples...because the mains may siimply not have constant frequency, is this correct?

Not constant frequency:
Let's say the mains frequency varies between 45Hz and 55Hz.

Which of these frequencies does not have the rising ZC at the midpoint of two falling ZC?

So the "varying frequencies" is no valid argumment for me.

Klaus

 

[cupoftea]

Thanks, so is there any situation, where assuming that a "inbetween" zero cross is not exactly half way between two 20ms zero cross points is valid?

 

[KlausST]

Hi,

You know I've designed mains monitoring devices.
And with the evaluated data I can definitely say there are some deviations.

I've experienced
* uncoordinated jitter (phase noise)
* but also unsymmetries caused by loads.

Many of our coffee machines use some kind of vibration pumps. They work with a magnet powered from mains using a diode in series.
So they draw current only of one half wave. Clearely unsymmetric.
And they casuse a bit of DC on the mains ..... in all outlets of the house (using the same phase).

Before doing this measurements I thought there is about zero (low millivolts) DC on mains. The first time I saw DC in the low Volts I doubted my measurement. Soon got the link to my coffe machine. And cross checked by rotating the plugs.
(I even were able to validate DC in the millivolts caused by a diode in series with 10k)

Back to the zero cross timing.
Although I experienced DC on mains I never did a check on how much this affected the zero cross timing unsymmetries. But I wouldn't be surprised if they cause several 10us.

Klaus