UPS power factor considerations

Sinusoidal or not?

The traditional approach to power factor leaves some gaps when reviewing modern power systems.

Most mature engineers think of power factor in terms of phase shift. This is because before modern switched mode power supplies became commonplace in electronic systems most non-unity power factor loads were inductive. Typically these loads might be a transformer or electric motor.In a simple electric motor being run in a normal fashion at full speed one expects to see a lagging power factor of around 0.8. The current is distinctly out of phase with the supplied voltage but it is sinusoidal

In the case of a modern computer file-server or telephone switch the power factor is often quoted around 0.75 suggesting that the power consideration is identical. However it is NOT. In the case of a SMPS based load the watts:VA ratio is caused by harmonics in the current NOT phase shift. The load current will be `peaky' (have a high crest factor) and anything but sinusoidal.

In choosing a UPS and planning a power distribution system the power factor consideration is an essential term in the overall equation.

Rule 1: PF = watts/VA

Using old UPS systems when the load was sinusoidal and had a lagging power factor allowed some economies in the design. A 10kVA motor rated at 0.8PF would only take 8kW. This means that theUPS battery and charge requires only to supply 8kW + re-charging current for the autonomy time. This led to UPS products being supplied with apparently quite small chargers to save cost.

In old UPS designs where the output stage was low frequency an output capacitor was sometimes added to effectively `power factor correct' the load. This permitted the designer to allow for only 8kW in the inverter again saving cost.

If we move onto SMPS loads this type of UPS is at a significant disadvantage because the capacitor now makes the situation even worse. The correcting capacitor adds to the apparent load seen by the UPS. The use of a SMPS causes a much larger UPS to be required. However it is true that the SMPS exhibits a non-unity PF (unless it is internally corrected) in that the VA is greater than the watts consumed. The battery can still be sized for the watts.


In modern UPS boxes the designer is aware of a lot of these problems and the UPS is usually designed to accommodate large peak currents which occur at every cycle in the waveform. Current limiting to protect the UPS is usually vested in switching to bypass earlier than is necessary to maintain some thermal margins. However with microprocessor controlled inverters the output stage can have several different overload checks - any one of which will activate protection. For example output switch / heatsink temperature / actual peak current / integrated current over time and many other solutions.

The method

When it comes to planning an installation there is a simple approach which although not entirely exhaustive provides a sensible way to size the UPS.

The user should add up all the actual RMS currents being used in the system together with any expected expansion. (We propose 30% is a good starting point for an expansion term). Do NOT use fuse ratings of equipment which bear little relation to the actual current required. Assume all the load is unity power factor

. This is nearly a worst case. The UPS is usually quoted with a power factor rating use this to identify its wattage rating and select a unit where the wattage exceeds the estimate above. Get the supplier to calculate the battery run time based on the actual watts of load.

If you would like any more help on this subject please ask!