True RMS

When measuring the value of an alternating current signal it is often necessary to convert the signal into a direct current signal of equivalent value (known as the RMS, root mean square, value). Most instrumentation and signal converters (for example handheld multimeters of the sort used by maintenance engineers) carry out this conversion by filtering the signal into an average value and applying a correction factor.

The value of the correction factor applied is only correct if the input signal is sinusoidal. Basically, this means that when measuring a source with such a signal, for example a light bulb, the current measurement will be correct. While working with switched devices like servers, this will never provide you a correct view on the real power usage.

The true RMS value is actually proportional to the square-root of the average of the square of the curve, and not to the average of the absolute value of the curve. For any given waveform the ratio of these two averages will be constant and, as most measurements are carried out on what are (nominally) sine waves, the correction factor assumes this waveform; but any distortion or offsets will lead to errors. Although in most cases this produces adequate results, a correct conversion or the measurement of non sine wave values, requires a more complex and costly converter, known as a True RMS converter.

Example

Putting this in a formula:

U = Voltage expressed in Volt
P = Power expressed in Watt
PF = Power Factor in %
I = Current expressed in ampère

U x I = VA (method used by most PDU devices)
U x I x PF = W (method used by the DCMU, 320.000 measurements per second)

With perfect sinusoidal signal, i.e. light bulb, the Power Factor will be 1, so 230V x 1 ampère X 1 = 230W

When no sinusoidal signal, i.e. the server, higher power peaks are needed, take a PF of 50%, so 230V x 2 ampère X 0,5 = 230W

So, this examples show that with a lower Power Factor (PF) there is a need for more ampère. More ampère means more costs.

Conclusion: calculating the Power without taking into account the Power Factor, will not provide the True Power usage! The DCMU measures the True RMS collecting 320,000 samples per second!