Bad Zero Crossing

It has a distinct name; "Zero Crossing Point" they call it and, as it's name suggests, is where the voltage of an AC waveform becomes zero while transcending from the positive half cycle to the negative or vice versa. And for every cycle, this happens twice.

The whole zero crossing bit does not even take a moment in time! However, if we were to take the portion of the cycle where the voltage is below that at which a typical silicon junction is able to conduct, we could say the zero crossing is about 10µs. Now in anyones book that is a pretty tiny chunk of time, and to think that there is even a possiblility to disturb the waveform during this time!

But that is the problem! By causing a disturbance at this time a piece of circuitry whose job it is to detect this crossing may very well believe there has been more than one crossing! We show just a few examples of what occurs when the transition from one half-cycle to the next is not an orderly one.

Clocks Running Fast

Electronic clocks use portions of the waveform as a means of "keeping time". The usual point is where the wave crosses zero. By simply counting the number of transitions through zero, time can be accurately kept by the internal circuitry. Unfortunately when there are glitches on the "zero crossing" point, the clock tends to run fast.... sometimes drastically.

Although 'running fast' is indicative of bad design, this is a real issue and affects the network supplier, even just listening to the complainant costs money. Ask any power quality investigator and they'll tell you loads of stories of the "alarm-clock", "clock on the microwave" and "timer on the video" (to name but a few) not keeping time - and it's all blamed on the power being supplied.

Note that the term "not keeping time" was used. Not only has bad power been known make clocks run fast, but slow too. This is simply (as an example) when ripple on the zero crossing point confuses the digital electronics (usually a microprocessor in modern clocks) as it was not expecting to have to process anotheanother zero crossing for a few milliseconds to come.

Jittery controls

It's all very well when it is a clock that runs away with itself, but ripple at a zero crossing point carries an inherent danger too. Most solid-state switches use a technique called "zero cross switching" meaning they turn on and off at the zero cross. The turning off is not usually a problem, but it's the turning on that can cause danger.

If the zero crossing is 'indistinct' then the switch could turn on early, late, or even not (microelectronics get confused), but the most dangerous of all, false triggering. Imagine the switch controls a motor that opens a cement bucket sluice and the crane is directly above a pit with a few workers in it.......

OK, so that was a bit dramatic (although it has happened), but such switches are also used in mixing plants to control the flow of various ingredients. Incorrect switching will result in bad mixes - and if it's concrete and the building later collapses....

Some of the less dramatic effects of bad switching is motors controlled by solid-state switches. If the switch keeps turning on early or late the pulse on the windings will capacitively couple to the armature and this energy has a deep desire to find its way to earth. The only path it has is the bearings which then prematurely fail.

Phase Unlocked Loops

There are many modern devices that require that they operate in phase with the incoming supply. Large switched-input UPS or motor speed controllers are just some of the examples of such devices. The technique by which this is usually done is is to have a circuit whose sole purpose in life is to supply a signal that is synchronised to the incoming supply, and goes by the name of "phase locked loop".

In the good ol' days PLLs (for this purpose) were usually created around very narrow frequency range oscillators that deliberately had sluggish feedback loops. This was done to minimise the effects of disturbances that may have been present on the incoming supply.

Then along came a whole herd of smarty-pants software chaps who claimed "we can do that with just a little bit of coding" - but that was the problem, they only used a little bit of code. So out the window flew yesteryear's wisdom and experience, only to be replaced with a power quality investigators nightmare, spurred on by a host of designers claiming the power is the reason their products don't work.

Good software based PLLs are not the easiest to generate, especially when trying to mimic the narrow frequency ranges and sluggish feedback loops (everyone wants to go for 'snappy' lock times and general purpose frequency compatibility). Designers are prone to use available 'building blocks' and dispense with 'purpose built software' which is tested to ensure the device works under all conditions.

But how does this affect the PQ? Simple. Imagine for a moment a large rectifier system made from software controlled switches as opposed to diodes. The incoming supply suffers a glitch and the software gets itself in a knot, and turns all the wrong switches on. The currents that flow are enormous. If the switches are followed by a battery bank, the system could even start supplying power back onto the mains.

A few faults are shown under the section "Interpreting the Readings".

Brown-Out Protection  >>

| | Ask a Question |

© 06.09.01