You are on page 1of 4

Watt do you Mean, or RMS?

by Leslie Green CEng MIEE Even practising engineers evidently have trouble deciding whether current should be measured as a mean or RMS value. Some may feel that RMS, being more technical, is always the correct one to go for, but this is a false idea. In practice, using a true RMS current measurement inappropriately can give a result in error by more than a factor of ten! It has been suggested to me that since I designed 6 digit true RMS DMMs for several years, I have an unfair advantage over other engineers in this area. In fact this subject matter is suitable for any under-graduate course and should be known about by all practising engineers. My intention is to present this article at such a simple level that even the most mathematically challenged enthusiast can understand the true basics. the switch on for 1 second and off for 1 second, repeating this cycle indefinitely. I think everyone would be happy to say that the average (mean) current was now 1 amp. What about the mean power? Well again most people would be happy to say that the mean power was 12 watts. Apparently the mean current multiplied by the fixed supply voltage gives the mean power in the load. This statement is in fact more generally true than the example suggests. Now I am going to increase the power, primarily for dramatic effect. Suppose I use a 48V ideal battery and a 4 ohm load, switched on for 1 second in every 10 seconds. The current when the switch is on is 12A and the power when the switch is on is 576W. The mean current is 1.2A and the mean power is 57.6W. This idea of turning on for some fraction of the whole cycle time is formally called a duty cycle, usually represented by the letter D. In the example above D = 1/10 = 0.1, and this factor is used to multiply both the peak current and the peak power to give the mean values. Now you may be feeling a bit uneasy at the moment because you were told at school that you are supposed to use RMS values to give the equivalent heating effect of a current. I could multiply the mean current by the steady battery voltage and it would give the correct mean power in the load. However, you may have felt that you should square the mean current and multiply by the resistance to get the power. Try it.

Quiz
Here is a quick test to see if you have really grasped the problems related to the title. The answer to each situation is either mean or RMS. a) Current measurement in a forward biased good silicon diode used to establish the power dissipated in the diode. b) Current measurement in the alternator output on a car to establish the power delivered by the alternator. c) Comparison of car battery charger output currents. d) For calculating the power dissipated in the battery charger leads of c (above). If all of your answers above were the same you are wrong. If you felt hesitant with your answer to any of the above questions then further reading is also definitely indicated.

Power
Electrical power in steady direct current applications is easy to calculate, you just multiply the potential difference (voltage) by the current. In alternating current systems there is more to worry about, so to maintain simplicity I am going to neglect AC systems entirely. Instead I am going to concentrate on direct current applications, although the current may be switched on and off at a regular rate. Suppose I have 12 volts from an ideal battery and I connect it to a resistive load of 6 ohms via a switch. When the switch is closed I get 2 amps of current and 24 watts of power. Now suppose I turn

1.2 2 4 = 5.76 W
Oops! Wrong by a factor of 10. It is a safe bet that if you are going to square a current to get a power, you should be using an

Leslie Green

logbook@lineone.net

RMS value. What current when squared and multiplied by the resistance gives the correct power? If I RMS R = P
2

Waveforms
The simplest load-current waveform to consider is a repeating rectangular pattern going from 0 to some fixed maximum level M. See Figure 1. FIGURE 1:
M

then I RMS =

P 57.6 = = 3.79A R 4

Perhaps you might now care to re-consider your answers to the quiz above and see if you have changed your mind on any of them before continuing to the next section.

Quiz Answers
a) It is always difficult to set simple questions to an experienced audience, The experienced reader will appreciate that there are a lot more problems associated with measuring the power in a diode than I am wanting to consider for the purpose of this quiz. The answer I was looking for is that the voltage across a good silicon diode is fairly constant at say 0.7V, regardless of the forward biased current flow. Using this model, the mean current multiplied by the fixed volt drop in the diode would give the mean power. A slightly more sophisticated linear model might consider the diode as a fixed volt drop, but with a definite added series loss resistance as well. In that case you would use the mean current multiplied by the fixed voltage then add on the RMS current squared multiplied by the loss resistance. Anyone familiar with diodes used in switched-mode power supplies, however, would also know that there are other switching losses when the diode is slow to turn on and slow to turn off, but those losses are well outside the scope of this article. b) The battery voltage on charge is fairly fixed, despite rapid fluctuations in the alternator current. It is therefore the mean current that defines the power delivered. c) A battery is charged by the mean current. The RMS current is always greater than or equal to the mean current. The manufacturer quoting RMS current can then quote a higher current and attract a semi-sophisticated public who believe that RMS is the more correct measure. (I must admit to being fooled by this one when I bought my 10A RMS battery charger 20+ years ago!) d) The power dissipated in the battery charger leads is the only case given here where RMS current is desirable/appropriate. The heating effect in the leads is found by squaring the current and multiplying by the lead resistance.

t Duty cycle, D = . T Mean current= M D . RMS current= M D .


Suppose I now connect two of these pulsating loads to the same battery, but without any regard for synchronisation between the two currents. The power drawn is clearly independent of any synchronisation between the two loads. Likewise the mean current is independent of any pulse synchronisation. The RMS current is another matter, however. Suppose that the two waveforms are in exact synchronism. All that happened was that the maximum value doubled. The RMS current therefore doubled, according to the previous equation but wait a minute, you vaguely remember from school that RMS currents are summed by squaring, adding, and then square rooting the result. This is called root of the sum of the squares (RSS) summation. Ok, lets suppose D < 0.5 and the high periods of the waveforms do not overlap (in time). The high period has doubled, so the value of D has doubled, increasing the RMS current by a factor of the square root of 2. We therefore have a bit of uncertainty as to whether the RMS current in the common feed wire to these two loads is 2.000 or 1.414 its previous value. Possibly those two values are sufficiently close that you would just ignore the problem, so let me make the problem worse for you. I am now going to take 16 nominally identical loads and consider the RMS current in the common feed wire. (This is actually a real problem from the world of industrial ink jet printing.) The RMS current is now either 16 or 4 larger than the single load current. Now you should see that if I had 100 driver stages on an integrated circuit, the RMS current could be either 100 or 10 larger than a single driver, depending on the relative positions in time of the pulses. This ambiguity is resolved in first year electronics text books, if mentioned at all, by declaring that the currents should be uncorrelated when using RSS

Leslie Green

logbook@lineone.net

summation. This concept is evidently very important if it means that the RMS current in the feed-wire can change by a factor of 10 or more. Correlation is usually expressed mathematically using calculus, at which point 95% of the readers roll their eyes skywards, the magazine slipping quietly out between limp hands. An explanation without the maths is always less ideal, technically, but suits a broader readership. I have stated earlier that RMS currents are always squared before use, and this is the key to the problem. Consider two current waveforms, A and B. Mathematicians would write these as A(t) and B(t) to show that A and B vary with time, that they are functions of time. To minimise the symbols I will just state that A and B vary with time and leave it at that. I am going to add A and B, then square the result.

software may well be using the whole screen full of data, the default setting. You may have to dig into the scope manual to find out if it is using the full screen of data, and how to limit the bounds of the measurement to limits set by vertical cursors. Some modern scopes now also give cycle-mean and cycle-RMS as distinct measurements; in other words the measurement is automatically done over an integer number of cycles. It is now easy to see how large the error could be. If the scope is giving the RMS value of the whole screen, and you happen to have the start pulse of one cycle at the left edge of the screen with the start pulse of the next cycle at the right edge of the screen, the measured on period is twice what it should be. If the pulses are narrow, the duty cycle will be almost doubled, increasing the measured RMS value by almost root 2 (1.414). If your scope cannot measure RMS values except over the whole screen, or if you need to consider different repetition rates of a given pulsed waveform, you have to understand your basics every well. Suppose the pulsed waveform is mostly zero but has a small burst of activity. Hand calculation of the RMS value is impractical so an automatic measurement is required. You want to zoom the timebase in to get the best possible resolution on the waveform, as in Figure 2. But now the RMS reading is wrong. How can you get to the correct RMS value?

( A + B )2 = A 2 + B 2 + 2 AB

A2 + B 2

In other words, squaring the summation of A and B is not the same as squaring the individual terms, unless it so happens that the effective value of the 2AB term becomes zero. In actual use, each side of the equation is summed over one or more complete cycles of the waveforms. If the summed value of the 2AB term is zero over this interval, the waveforms are said to be uncorrelated. For the pulsed waveforms used in the previous examples, the waveforms must not overlap in time if they are to be uncorrelated. In AC circuit theory a fundamental sinewave and any of its harmonics are always uncorrelated, allowing the RMS value to be found by RSS summation of the harmonics.

FIGURE 2:

Measurement Pitfalls
Even an experienced engineer can get into trouble when measuring mean and RMS values. Typically a waveform, such as that shown in Figure 1, would be displayed on a digital storage oscilloscope screen, and the automatic measurements would display the mean and RMS values. As a check, you may get out your calculator and try doing the maths independently to check up on the scope. You will be glad you did because the discrepancy between your quick sum and the apparently precise scope readout could be huge (40%). For a simple rectangular waveform it is hard to understand where your maths could have gone wrong and yet it is almost inconceivable that your trusty respectable scope would give the wrong answer. The problem is one of complete cycles. Your calculation is based on one complete cycle of the waveform. The measurements in the scope 0 To solve this problem you need to think in terms of energy. This current waveform when squared and multiplied by the resistance of the feed-wire will give the power dissipated in the feed-wire. The mean power multiplied by the total time shown on the scope screen will give the energy dissipated in the feed-wire by this one pulse.
2 dissipation = I RMS Rwire t

I RMS =

1 dissipation t R wire

The dissipation and wire resistance were just tools used to give a physical reality to the formula, and can now be neglected as we will be comparing one RMS current with another. The scope will display a particular RMS current value when the timebase is set to give say 10s across the whole screen (10

Leslie Green

logbook@lineone.net

divisions at 1s/division). If you change the timebase to 20s across the whole screen the RMS value will drop by a factor of root 2, as seen by the dependence on the inverse square root of time in the formula above.

Actual RMS Current = Measured RMS Current scope timebase span actual repetition period

Armed with this formula you can accurately calculate the actual RMS value of a repetitive pulse from a measurement of just the active part of the pulse waveform.

Summary
The mean current multiplied by a steady (DC) voltage gives the power into or out of a system. A current which is squared and multiplied by a resistance must be an RMS current. The RMS current is always larger than the mean current, except for a steady direct current where they are equal. Narrow peaky waveforms have disproportionately large RMS values compared to their mean value. Automated measurements of waveforms using oscilloscopes must be done over an integer number of cycles, either using cyclemeasurements or manually, by using vertical cursors as the measurement bounds/limits/gating.

Leslie Green

logbook@lineone.net

You might also like