logbook@lineone.

net

calculus

is for life, not just for Christmas

Changing Conditions

In recent interviews of graduate applicants for the

position of design engineer, I have been appalled by
their lack of understanding of elementary calculus 1.
If you think your calculus is up to scratch, then
there are two sets of simple test questions in this
article. Do them. If you get any question wrong then
your maths is not even close to being acceptable. You
need to read the whole article carefully and try again.
Calculus is not something you should learn at
school and then promptly forget; it is a general way of
thinking that will outlast any technological changes
that can happen in your career.
This article is to give adequate stress to the subject
for those who didnt realise its importance and to act
as a primer to those who never studied it in the first
place.

Given the nice simple equation F m A , the

student is never asked to think about what would
happen if the force were to change with time, or if the
mass were to change with time. What would happen to
the acceleration in these cases?
As soon as changes occur in these simple
situations, more advanced mathematics is called for.
This step is not difficult to grasp if you understood the
steps leading up to it. If you failed to get the earlier
material, however, you get lost and confused.
I was introduced to calculus in Advanced level
maths classes by the simple statement that if y x n
the differential coefficient is given by

dy
nx n 1 . It
dx

was not even mentioned that we were now dealing

with a subject called calculus. With instruction like
this, it is hardly surprising that engineers are being
trained who dont know what the subject is all about.
We need to go right back to basics and start again.

Calculus and Engineering

You will find all electrical engineering textbooks write
equations using calculus notation or simplified forms
derived using calculus. It is therefore in your best
interests to know the basics of the subject, so that you
do not feel and act like a novice.
As a teenager at school you would have done
simple problems such as calculating the power in a
DC circuit or investigating the acceleration of a mass
when a constant force was applied. The equations
given were:

Calculus for Engineers

Once upon a time, charge was moved from one
conductor to a nearby conductor. This caused a
difference of potential between the two conductors. It
was observed that moving twice as much charge
created twice the potential difference.
Using mathematical terminology it was said that
the potential difference, V, was proportional to the
charge, Q, which had been moved. This was written as
V Q . Now all we have done here is to write an
experimental observation in a concise mathematical
form.
The next step was to put in a constant to replace
the proportionality sign. We now say that Q C V ,
where we have a constant, C, for any particular
system of two conductors. We call this the
capacitance between the two conductors. There is now
an equation, and a constant which we can measure.
If charge moves with time we say that an electric
current is flowing. More specifically we say that the
rate of flow of charge past any fixed point is the
definition of current.
Now saying that the rate of change of charge with
respect to time is the current is a very longwinded

P I V ; Power is the product of the current and

the voltage.
F m A ; Newtons second law of motion.
These equations are for steady conditions, with no
change occurring in the driving forces. This is a nice
simple introduction to science and lets the student feel
that it is possible to learn about science and make
calculations.

The Latin word calculus means a small pebble used for

counting, as well as the subject of reckoning. The
original presentation of this branch of mathematics was
written in Latin.

November 2000

by Leslie Green CEng MIEE

Leslie Green

logbook@lineone.net

calculus

way of expressing the idea. So we use a mathematical

notation instead and say :

a chart recorder or a scope; it is anything but constant.

If we want to measure the slope at any point, we
construct a right-angled triangle, where the long side
(red = hypotenuse) is in the same direction as the
waveform (the tangent to the curve). The slope is then
defined as the length of the vertical line in the triangle
(purple) divided by the length of the horizontal line in
the triangle (green).
This is how one could go about measuring a slope
from a graph. The method of calculus gets the slope
from the equation of the curve. Rather than using x
and y, which seems like a maths lesson, lets take a
simple equation for a voltage, V A t K B .
A, B and K are just constant values. We want to
know what the slope of this curve is at some point in
time, t1 . We know the value of V at this point; it is

dQ
. This is the notation presented in 1684 by
dt

Leibniz, when he published the first paper on

calculus2.
We can apply this notation to the equation
Q C V and we get

dQ d C V

. Now we
dt
dt

can look at the special case where the conductors are

not moving relative to each other. In this case the
capacitance is constant and the C can be moved
outside of the rate of change expression. This now
gives us a very important equation: i C

dV
. In
dt

words this says that the instantaneous current in a

capacitor is equal to the product of the capacitance
and the rate of change of voltage across the capacitor.
If you have ever looked carefully at capacitor data
sheets, you may have seen limiting values of

simply V1 A t1K B . But that hasnt yet told us the

slope of the curve.
Now comes the clever part. We can look at the
value of V at some very short time later. We know that
from the graph drawn previously, the triangle
measuring the slope was very large compared to the
bend in the waveform. However, if we were to get out
a magnifying glass and zoom in on a small part of the
waveform, it would look almost straight. The closer in
that we looked, the straighter it would appear.
Hence we are going to see what happens to V1
when time moves on by a very small amount t .

dV
.
dt

This is nothing mysterious; if the rate of change of

voltage across a capacitor is too high, the current
through it will be too high and the capacitor will burn
out. If you see a calculus expression like this and
neglect it because it is too hard, then you will be a
liability to everyone around you.

Origins
The rate of change of a quantity with respect to time is
not a complicated issue. If you draw a graph of the
quantity, with time along the bottom axis, then the rate
of change is simply the slope of the line, also known
as its gradient.
If you wanted to know what the slope of any
particular function was, then you could just draw a
graph and measure the slope with a ruler. Now you
may have thought that you should measure a slope
with a protractor, an angle measuring device. Well,
you can also measure a slope in terms of how far up
you go compared to how far along you go; hence the
use of the ruler.

V1 V

V1
t1

t 1 t

The length of the purple vertical line is V and the

length of the green horizontal line is t . These sizes

voltage

are related by two equations: V1 A t1K B , and

the one obtained by moving forward in time by a small
amount,

K
V1 V A t1 t B . All we do now

is subtract the first equation from the second.

V A t1 t t1K

Now comes the difficult part. Expanding the terms

inside the square bracket uses another mathematical
technique, the binomial expansion (remember
Pascals triangle).

time

t1 t K t1K K t t1K 1 t K

The blue waveform is a plot of the voltage across a

capacitor against time, as you might expect to see on

There are K+1 individual terms. The first term, t1K , is

equal but opposite to a similar term inside the square
brackets above; it therefore cancels. We are then left
with K individual terms, where the power of the term
t is steadily increasing. We now make use of a

Nova Methodus Pro Maximis et Minimis, , Acta

Eruditorum 1684 by G.Leibniz; pp466-473 of the 16821687 volume (1966 reprint). Unfortunately this is all in
Latin.

November 2000

Leslie Green

logbook@lineone.net

calculus

simple fact. If a number is small, then higher powers

of that number are even smaller still. For example,
take the number 0.01: when squared it gives 0.0001
and when cubed it gives 0.000001. Thus if we make
t small enough, we can neglect all the higher
powers of it. We can then simplify our equation to
V A t K t 1K 1 . At this point we say that we

current

make t
so small that is almost zero;
mathematicians would say it is the limit as t
approaches zero. We then make the approximately
equal sign () into an equals sign, because the error is
arbitrarily close to zero. The curly is replaced by a
straight version, a d, to show that this limiting value
has been used. This gives dV dt KA t 1K 1

The area under the curve above is the product of

current and time, which is actually the total amount of
charge that has moved during the interval in question.
We could evaluate this area by drawing the graph on
paper, cutting it up into little strips and then measuring
the area of these strips. This is actually done in some
elementary maths classes.
The next step is to move on to finding the area by
using the trapezium rule; mathematically splitting the
graph up into strips and evaluating the area by means
of numbers. The next logical step in this sequence is
to make the strips really small.
Consider a more general curve described as f t .
This is read f of t, and simply means that the function
f varies in some manner according to the value of t.
The beginning of the area of interest has been
arbitrarily set to 0. The end of the area has been
marked as t1 . If the area is divided into n equal-width

We now have the slope of the line at the point t1 . It

is simply

dV
KA t 1K 1 .
dt

This is differential calculus, because it deals in

small differences. There is considerable complexity
involved in applying these techniques to particular
mathematical functions, but then mathematicians
have been working on these problems for over 300
years. All we have to do is look in a book to find out
how to get the slope of any particular function.
There is of course a mathematical name for getting
the slope of a function; it is called differentiating the
function. The result is called the derivative, or more
specifically the first derivative. You can also find the
slope of the slope, the second derivative, and so on.
For our function V A t K B we differentiate with
respect to t to get the derivative

time

t1

vertical strips, then each of these strips is

dV
KA t 1K 1 .
dt

t1

wide.

For convenience we can call the width of these strips

t . We zoom in on one of these narrow strips so that
the curve seems to be straight over the small
distance.

TEST YOURSELF:
What do these fundamental equations mean?

d
dt
di
2) E L
dt
dV
3) i C
dt
dV
4) r
di

f t

1) E n

f 0

Integral Calculus

We have a rule for getting the derivative of a

function;` for differentiating it in other words. This
gives the slope of the curve. Another useful operation
would be to find the area under the curve.

time

f t f 0

This strip has an area of t

The

next

strip

f 2 t f t
.
2

has

an

area

of

The total area is given by a summation of all of these

little strips. The capital sigma, , means sum the
expression which follows, starting from the limit under
the sigma symbol (m=0) and ending with the limit
above the sigma symbol (m=n-1) in steps of 1.

November 2000

Leslie Green

logbook@lineone.net

m n 1

Area

calculus

m 1 t f m t
2

m0

This is then very similar to the simple DC case. You

should be able to see that:

Inspection of this summation shows that there are

always two halves of each term like f m t ,
except at the end points. We can therefore write:

2
V EQUIV

m n

Area

2
V EQUIV

f m t t

We now let n head off to infinity, which makes t

tend towards zero. This reduces the error at the end
points to an arbitrarily small value. The approximate
equality is replaced by an equality and the summation
sign becomes an integral sign.

t1

V EQUIV

t dt

i dt Q

dQ
dt

Using this notation we can now formulate precise

statements about power, energy, equivalent heating
effects etc.
power = rate of change of energy
energy

v i dt
0

v
0

v
1
dt

R
R

v 2 dt
0

Often the waveform is cyclic, with a period of say T. In

these cases we are interested in the average (or more
specifically the mean) power. We then write:
energy per cycle
mean power, P
cycle time
1 1
P

R T

dt

and

K V dt K

V dt

Calculus operations are able to be split up between

simple additions and subtractions. Thus:

d V1 V 2 dV1 dV 2

dt
dt
dt

It would be nice to use a simplified form of this

equation, where there was a voltage which when
squared gave the same heating effect as the true
calculus expression. It would then be:

November 2000

d K i
di
K
dt
dt

T
2

v 2 dt V RMS

Integration is somewhat harder than differentiation in

terms of the mathematical manipulations. Often one
has to see what function when differentiated gives the
function we are trying to integrate. You almost guess
the answer, differentiate it and see if you were correct!
In any case, most of the hard work has been done
by our predecessors compiling tables of the difficult
functions. What remains is for us to break any
particular problem down into a form such that we can
look the answer up in such a table.
For both differentiation and integration, constant
multipliers can be freely moved outside of the calculus
part. For example, with a constant K;

Measuring current is always more difficult than

measuring voltage, so it is more convenient to say
that the current is the voltage divided by the
resistance.
energy

1
T

RULES OF DIFFERENTIATION AND

INTEGRATION

power dt

In electrical terms the energy could be the heat

dissipated in a resistor, for example.
energy

The equivalent voltage is the square Root of the Mean

value of the sum of the Squared voltage, or RMS for
short.
This is the most elementary of electrical
engineering topics and yet you still get engineers who
are unable to recognise that RMS voltage and RMS
current are valid quantities, but that use of the term
RMS power is technical illiteracy, as there is no such
valid, useful quantity.
This lack of understanding is not assisted by
manufacturers adverts which talk of true RMS power
measuring devices. This is tricky. Power measuring
devices can measure peak power, assume the signal
is sinusoidal, and give a calibration of the actual mean
(average) power.
A better technique is to measure the RMS voltage
across a resistance, which gives the true power. In this
case the manufacturer is trying to say that their
measurement method uses a true RMS voltage
measuring technique, which is more difficult to do but
gives a more accurate answer. The adverts truncate
this to a true RMS power measuring device, which
confuses those with inadequate understanding of their
basics.

Closer examination of the integral sign shows it to be

a large S, suggesting summation. Notice that the limits
of integration given at the bottom and top of the
symbol are in the same units as the variable of
integration; in this case t. We are summing the
function f over a range of time values.
We can now bring on the fundamental theorem of
calculus which says that differentiation and integration
are an inverse pair of operations. In other words, if you
integrate then differentiate, you end up where you
started. Let me make this more real with an example.
Using time as the reference, the integral of current is
charge, and the derivative is charge is current. In
symbolic notation:

v 2 dt

By taking the square root of both sides we get

m0

Area

1
T

i1 i 2

dt

i i dt

and

i 2 dt

Leslie Green

logbook@lineone.net

calculus

A constant has no rate of change with respect to

anything. Thus the derivative of a constant is always
zero, regardless of the variable we are differentiating
with respect to.

e K t

This table uses time as the variable of integration. The

derivatives are also done with respect to time.

dK
0.
dt

Using the constant K again:

Now you are ready for the final test These are very
basic questions actually; if you cant do them or get
them wrong, after having read this article carefully,
then you definitely need to dust off those text books
and put in some extra study.

The integral of a constant gives the product of the

constant and the variable of integration.

K e K t

K dt K T
0

This is obvious if you think about the integral as being

the area under the curve. If the curve has a constant
value, the area is rectangular and is simply the
product of the height of the curve above the axis and
the distance along the axis that we integrate over.

FINAL TEST
1) What is the maximum slew rate of a 3V RMS
1MHz sinusoidal signal?
2) What is the current flowing in a 0.1F capacitor
when a voltage is applied to it which increases
linearly from 0V to 100V in the space of 1.5s?
3) What is the RMS value of a repetitive 25kHz
rectangular voltage waveform which is at 10V
for 10s, then at 1V for the rest of the cycle.
4) What is the RMS power dissipated in a 1 resistor
when the voltage of #3 above is applied across it?

LIMITS OF INTEGRATION
So far I have skilfully avoided the subject of the limits
of the integration. If I were to differentiate a function
such as V K 1 t n K 2 the additive constant would
disappear.

dV
nK 1 t n 1
dt

This means that when I integrate some function, I

always have to add on an arbitrary (unknown)
constant. For example:

t n 1
t n dt
K
n 1

The rule for integration is seen as being the rule for

differentiation done backwards. In differentiating I
multiplied by the index (n), then reduced the index by
one. For integration I increase the index by one (n+1)
then divide by this new index.
What I have given here is an integral without limits.
This is called an indefinite integral. I have asked the
question, What is the area under the curve up to a
specified point? but I have not given the starting point
of the summation! This is a graphic explanation of why
an indeterminate constant is required.
The last rule required for integration is that the
limits can be split into parts. Suppose we have a
waveform which is doing one thing up to a time t1, but
then does something else up to the time T. We can
split the integration up over these two intervals to
make the integration easier.

i dt
0

t1

i dt

i dt
t1

By splitting the integral up into these intervals, it is

possible to look up the individual function pieces in
tables.
Differentiate this to
get that
K

K t n
K t n 1
n 1
sin t
cos t
November 2000

Integrate this to get

that
0

nK t n 1
K t n
cos t
sin t
5

Leslie Green

logbook@lineone.net

calculus

ANSWERS TO BE IN A REMOTE
PART OF THE MAGAZINE

Answers to Final Test

1) A sinusoidal voltage is mathematically represented
by the equation V V MAX sin t . Now is
the angular frequency in radians/second. It has to
be in radians/second to make the calculus work
correctly. The slew rate is the derivative of the
voltage with respect to time.

Answers to Test Yourself

1) E n

d
dt

This

is

Faradays

law

of

induction. For a coil of n turns, the EMF

generated is equal to the rate of change of
the flux, , multiplied by the number of turns.
The law can also be stated as the rate of
change of flux linkages, but this form is more
understandable for stationary devices such
as transformers.
2) E L

di
dt

dV
V MAX cos t
dt

The maximum value of a cosine function is one,

so:
dV
V MAX
dt MAX
Remember that 2 f and that the peak
value of a sinusoid is 2 times the RMS value.
dV
2 10 6 2 3 26.66 V/s
dt MAX

This is a key equation for an

inductor, stating that the EMF across an

inductor is equal to the rate of change of
current through it, multiplied by its
inductance. The equation can be re-arranged
to work out the rate of change of current
from an applied EMF. This is essential for an
understanding of switched mode power
supplies.
3) i C

dV
dt

This

is

the

current

in

2) i C

3) 25kHz is a period of 40s. The waveform is

therefore at 10V for 10s and 1V for 30s.

a
2
V RMS

capacitor caused by a changing potential

difference across its terminals. This was
explained in the text.

dV
4) r
di

dV
100
0.1 10 6
6.667 A
dt
1.5 10 6

40

10

10 2 dt
0

10

1
2
100 t 100 1 t 1040
V RMS

40
1
2
1000 30 25.75
V RMS

40
V RMS 5.074V

The small signal resistance of

a component is the slope of its voltage/current

curve. It is the small signal form of Ohms law.

40

12 dt

As a quick check, we ensure that the RMS value is

smaller than the maximum voltage (10V) and
larger than the minimum voltage (1V).
4) If you answered with just a number then you failed.
If you were unsure, but thought it seemed wrong
then you failed; you need a much better
understanding of the basics than that. The answer I
am looking for is that the question is faulty. The
mean power in the resistor is of course 25.75W.
The peak power is 100W. The crest factor is 1.97.
The question asks for a stupid quantity which, if
calculated, would give a meaningless result. If
anyone asks you for RMS power you have to
interpret the request as being for mean power and
then point out the error.

November 2000

Leslie Green