3 phase zero crossing detection

[Bob Edge]

Hi,

I am currently using pic microcontrollers to detect the zero crossing point of the 3 phase supply. It works great until the machines are used with a generator, usually thousands of miles away. I think most of the problems are to do with detecting the zero crossing point.

What is the industry standard way of detecting the zero crossing point?

Best regards
Bob...

 

[luben111]

The schematic depends on how precisely you want to detect the zero crossing point.

One simple solution used in many printer/faxes is to put resistors 100k-200K in serial with bidirectional optron (two LED in opposite directions). If resistors are SMD it is recommended to put several in serial to increase the breaking voltage. On the other side of the optron you get short pulses on the collector to HIGH level when the input voltage is around zero.

In case you need more precise zero crossing detection you need more complex schematic.

 

[chuckey]

I tink that if you use a crude low pass filter on the mains feed, say use Luben111's 100k and then put a 1000pF capacitor across the opto. The capacitors should be closely matched in value. Their purpose is to shunts out spikes and general noise on the mains input.
Frank

 

[luben111]

Hi Chuckey,
The dark side of such capacitors is that they shift the phase of the zero crossing by slightly delaying the pulses (RC filter delay). For sure they help but if you keep in mind that LEDs can accept huge short overcurrents their precense is not neccessery. I have seen many schematics of printers and faxes with such zero crossing detection but I never had seen capacitor in parallel to the diodes, which doesn't mean that it can't be useful in some applications

 

[chuckey]

Yes, Luben111 that why I said the caps should be matched (and so should the resistors). The problem is the noise and spikes which if occuring at the mains zero crossing will artificially "move" the apparent zero crossing. The posh way would be to do a phase lock on to the mains so you can get a very clean sinewave and rely on the LPF within the phase lock loop to remove any phase noise.
We have use a mains spike detector in what can be described as a clean laboratory condition, no multi horse power motors within the building and it registered spikes every few seconds over 500 V and every few minutes it went off scale (>7KV!!)
Frank

 

[luben111]

Hi Chuckey,
Well, such strong spikes are really scary and need careful design of the schematic. The faxes and printers need to know only if there is main power or not, so my proposed schematic will work well. The optron is quite slow device so it will not react on short spikes but I absolutely agree with you that for prtecise zero crossing it is not suitable.
BTW, what could be done with PLL could be imitated with microprocessor - the output from the optron is connected to some uP pin to trigger high priority interrupt. The optron events are used to calculate the time between two zero crossings and the measured time is strongly digitally filtered. The zero crossing will be generated through some uP timer which is adjusted in the zero crossing interrupt routine. If the uP was carefully programmed the timer could generate very precise zero crossing events

 

[Bob Edge]

Hi guys,
Thankyou for your interest. I have found the attached paper. This describes exactly what you sugested Luben.
It is a little beyond my abilities to be honest. But I will give it a go.

One thought I had was to use the two opto isolators, and simply use this to detect the zero crossing because you would only need to detect one switching, then the other in a short time, and if this is not the case then any signal must be eroneous.

Anyway I thought I would just post the paper as it may be of interest to you guys.

Regards
Bob...

 

[Bob Edge]

Hi guys,
Thankyou for your interest. I have found the attached paper. This describes exactly what you sugested Luben.
It is a little beyond my abilities to be honest. But I will give it a go.

One thought I had was to use the two opto isolators, and simply use this to detect the zero crossing because you would only need to detect one switching, then the other in a short time, and if this is not the case then any signal must be eroneous.

Anyway I thought I would just post the paper as it may be of interest to you guys.

Regards
Bob...