INTERPRETING THE READINGS:
Crest Factor
 

There are two aspects to Crest Factor. The first is source and load, the second is an instrument's ability to measure it.


Source and Load

We introduced the concept of Crest Factor in the page on "Understanding Running Voltage", but I would like to recap a few important points before moving on.

When asking of a group of power engineers "what voltage is 230VAC?", I am usually met with a host of seriously blank, puzzled looks! When it is defined that a pure sinewave voltage of 230VACrms actually swings from zero to ±325V at the crests, and that it is these peaks that are captured by the likes of rectifiers and smoothing capacitors, the penny suddenly seems to fall.

When it comes to running voltage of hi-tech equipment the RMS value of the incoming supply is of absolutely no use to man or beast. It is extremely important when calculating the amount of power drawn by a resistive load, as proven in the previous page, but this is as far as it goes.

Alongside is shown two waveforms. The first has an RMS of 230V, the second an RMS of 240V. According to the user of a "true RMS" multimeter it is the first that is well within limits and has quite some way to go before damage to hi-tech devices becomes a concern. In actual fact the first that has a hidden danger while the second is in fact being kind to hi-tech gear.

The reason for this phenomena is the first has a triangular waveshape with a peak voltage of 353V, the second is more rounded and only a peak of 339V. This difference is indicted in "crest factor" - the ratio of Peak/RMS. A pure sinewave has a crest factor of 1.414.

What was not introduced at the time was the need for crest factor on current measurements. UPS systems are the most vulnerable here. The electronics used to generate the output waveform (commonly known as the "inverter section"), has a limited ability to generate output current.

To make things simple, we'll assume a small 1kVA UPS at 230VAC. Our example UPS will, under resistive loads, be able to supply 4.53A. More importantly, the peaks will be 6.15A. This UPS, however, is known to not survive for long before switching itself to 'bypass' with an overload indication.

Hi-tech loads are nothing like resistive loads and generally have peaks far in excess of a sinewave's 1.414. The best experienced to date is about 1.7, but it is typically around 2.5 and higher. Let's be kind and settle for an average and work on a crest factor of 2.

Along comes an IT specialist with a "true RMS meter" (because that is what he was told he needs) and measures the load to be about 4.0A. Doing a small calculation he believes his UPS will be able to handle it.

But, using the crest factor we have just settled on means his peak currents are 8A. This is a far cry from the UPS's ability of 6.15A and the UPS is in excess of 30% overload. In actual fact, the best his UPS is able to deliver (with a crest factor of 2) is about 3.1A or 0.7kVA i.e. 70% of the rated output.

It does make one wonder why facilities managers of hi-tech data centres insist on "True RMS" multimeters! Er, yes, well,.....


Instruments

Now for the important part, the ability of instruments to operate correctly under extreme crest factor conditions and how to tell the instrument is letting you down.

I have seen some incredibly interesting results from instruments not capable of measuring true RMS under extreme crest factor conditions. Hi-tech loads have been known to have current crest factors well in excess of 6. This means the RMS current is only one sixth of the peak value! An example would be a low voltage lighting circuit drawing 1A RMS will in fact be asking for 6A on the peaks.

I saw one individual attempting to market an active harmonic filter using an interesting 'side effect' the filter had of lowering the RMS power consumed. What was unknown to him was the instrument used during the "before & after" comparisons had a problem with the crest factor on the "before" portion. I bet if he used the trick about to be explained, he would find the current goes up a little on installation as the filter draws a little energy.

The input amplifier on an instrument used for measurement needs to cope with waveform peaks. Should these peaks be 'clipped', the circuits tend to try and compensate and land up increasing the output value. Such clipping therefore destroys any RMS accuracy. Obviously the overload can be so intense the RMS value is reduced, but this only occurs when the overload is near a square wave.

One trick I have learned is to drop the instrument sensitivity by one decade (increase the voltage/current range by 10) and ensure the indicated value is very close to that which was indicated on the previous range (accepting the higher range may round either up or down, but should be no more than ±1 on the least significant digit).

Minimum, Maximum, and Average  >>


| | Ask a Question |

© 29.02.04