Using DC motors in fighting robots

1. Introduction

This page describes how DC motors work, and how we can

use them to build the traction system of a robot. It covers
both permanent magnet motors, and series wound motors
(such as car starter motors). If you are interested in
converting a starter motor for use in a robot, see the
separate page Converting starter motors.

2 Motor principles

All motors require two magnetic fields, one produced by the

stationary part of the motor (the stator, or field), and one by
the rotating part (the rotor, or armature). These are
produced either by a winding of coils carrying a current, or
by permanent magnets. If the field is a coil of wire, this may
be connected in a variety of ways, which produces different
motor characteristics.

The basic law of a motor, the reason why they rotate, is

governed by Fleming’s left hand rule (see figure below). This
tells you the direction of the force on a wire that is carrying
current when it is in a magnetic field.
The next diagram shows the force acting on a wire carrying
current, obeying the left hand rule:

If we now bend the wire round in a loop, and place it in a

magnetic field caused by two permanent magnets, we have
the situation shown in the diagram below. Here, both sides
of the wire loop will have a force on them, trying to make the
wire loop rotate. The current is applied to the loop through
the commutator, which is shown as two pieces of metal
formed into a ring in the figure. Current is applied to the
commutator by stationary graphite blocks, called brushes,
which rub against the commutator ring.

The loop will continue to rotate anticlockwise (as we see it in

the figure) until it is vertical. At this point, the stationary
brushes won't be applying current around the loop any more
because they will be contacting the gap between the
commutator segments, but the inertia of the loop keeps it
going a little more, until the DC supply reconnects to the
commutator segments, and the current then goes around
the loop in the opposite direction. The force though is still in
the same direction, and the loop continues to rotate.

This is how DC motors work. In a real motor, there are many

wire loops (windings) all at varying angles around a solid iron
core. Each loop has its own pair of commutator
segments.This block of core and wire loops is called the
rotor because it rotates, or the armature.

The fixed magnets in the diagram above generating the field

may be replaced by electromagnets which are generally
more powerful. The elctromagnets are supplied by the same
power supply as the armature winding, either in series
(series connected) or in parallel (shunt connected) as shown
in the diagram below.

If permanent magnets are used, the motor is said to be a

permanent magnet (PM) motor. DC motors can also have
permanent magnets in the armature, and electromagnets for
the stator coils. In this case, the stator windings must be
switched in some way to make the permanent magnets in
the rotor follow them to cause rotation. This connection is
less common for small motors.
There may be more than just the two fixed magnets, called
poles. In some motors there may be four poles (imagine one
more coming from above and one more from below in the
figure). There will always be an even number of poles, since
there is an N pole for every S pole, and in the equations
governing the motor, the number of poles is often quoted as
p, the number of pole pairs. These magnets are called the
stator because they are stationary, and the electromagnet
coils are called the field windings because they generate the
magnetic field.

The supply is connected to the commutator segments

through graphite brushes. These are held in little sockets
with a spring behind them, so the brush is pushed onto the
segments. This guarantees a good electrical connection
(although there will be a fraction of an Ohm resistance
across them). Eventually, these brushes wear down
completely. If you get the motors from a scrap yard, the
brushes may need replacing. New brushes should be
available from automotive spares shops.
A four pole motor is shown in the diagram below. This shows
how the magnetic field is generated by the poles and flows
through the rotor:

If you can imagine getting the cylindrical rotor of a starter

motor, and unrolling it, the wires would look something like
this:
In this diagram, the stator magnetic poles are shown
underneath the rotor wires. You can trace the electrical
current from the supply, through a commutator segment,
through all the rotor wire loops, and back through a
commutator segment. For a four pole motor (two pairs of
poles), two pairs of commutator segments will be supplying
current to the rotor wire loops

We can now draw a circuit diagram of the motor. The coils of

wire will have a small resistance, and a quite large
inductance. Rb represents the resistance through the
brushes to the commutator. Rf is the resistance of the field
windings, and Ra is the resistance of the armature (rotor)
windings.
If the motor is has permanent magnets for the field, then of
course Rb and Rf don't exist. Most starter motors use a serial
connection. (For example values of these resistances and
inductance, the starter motors I am using (from a Ford
Fiesta) have an inductance, La, of 71μH, and resistances of
Rb=0.008 Ohms, Rf=0.016 Ohms, and Ra=0.017 Ohms,
giving a total, Rt, of 0.041 Ohms.

2.1. Back EMF

In the same way that a wire carrying a current in a magnetic

field has a force acted upon it (left hand rule), conversely a
moving wire in a magnetic field gets a voltage induced
across it. This is called Lenz's law. In a motor, the
consequence of this is that the supply voltage makes the
rotor rotate, and the rotor generates a ‘back emf’ or reverse
voltage, which nearly matches the supply voltage when the
motor is rotating but not driving anything.

The resistance of the coils of wire that make up the field and
armature are normally very low. So you would think that the
motor would take a very large current (current = voltage ÷
resistance). However this back emf means that there is
actually only a small voltage across the resistance of the
coils, because the back emf, ea, is nearly as large as the
supply voltage, Vt. Therefore , the actual current taken is

Some typical values for a 900Watt starter motor are: Rt =

0.041 Ohms, Vt = 12V, ea = 11.9v, so the motor current is
2.4 Amps.
The back emf is dependant on the speed of the motor, and is
given by the equation where k is a constant for the motor,Ω m
is the speed of the motor, and Φ is the strength of the
magnetic field generated by the field coils or magnets. From
this equation you can see that when the motor starts, and its
speed is zero, then the back emf will be zero, and so the
current taken by the motor will be very large. This is also
true when the motor is stalled.

When a load is put on the motor, the speed drops, and so

the back emf drops, and so the current drawn from the
battery will increase.

2.2. Torque

The torque produced by a permanent magnet or shunt

wound motor is given by the equation:
T = kfΦia

The torque produced by a series motor is given by the

equation

T = kfia2

where kf is a constant for the motor, and ia is the current,

and Φ is the field strength. A series wound motor generates
its field from the same armature current, so Φ increases as
the motor draws current. Therefore the torque is
proportional to the square of the current. This means that
doubling the current that we put through a series wound
motor quadruples the torque. This is a distinct advantage of
series wound motors.

2.3 Current

When you read a DC motor datasheet, it will sometimes (if

you are lucky!) have some characteristic graphs that tell you
how the motor is likely to respond. These mostly have Load
Torque as the bottom x-axis, and plot motor current, speed,
power, and sometimes efficiency as they relate to the load
torque. Sometimes they may have motor current on the
bottom axis instead of load toruqe. In this case the graphs
are still almost the same shape because the motor current is
almost directly proportional to the load torque in all DC
motors. Different types of field windings produce different
shapes for these graphs.

For a permanent magnet or shunt wound DC motor, the

current drawn by the motor increases linearly with load
torque:

Notice that the line does not intersect the (0Nm / 0A) origin
point. The zero torque point intersects at 25 Amps for this
motor (an Iskra 900W starter motor). This means that the
motor will not even start to turn until it is pulling 25 Amps!

2.4. Speed

For all DC motors, the speed is nearly proportional to the

supply voltage, except for the effect of the armature
resistance:
The values Ra and kf are constant for any particular motor. It
can be seen that the speed will drop off as the load torque
increases, but differently for each type of motor:

From this graph, and the equation, it can be seen that the
speed of a series motor with no load torque will rise to
infinity! This doesn't happen in practice because there is
always a little load torque because due to friction in the
motor. It can also be seen that as a little load torque is
added to a series wound motor, the speed drops off very
quickly to start with, but then drops off less quickly.

With both motor field types, eventually, when the load

torque becomes even larger, say when a stronger robot is
pushing us backwards, the motor speed becomes negative
(the motor starts rotating backwards).

In practice, the speed line for a PM or shunt wound motor

may not be a straight due to frictional loads which are not
linearly dependant on speed.

The torque speed graph of a DC motor when it is being

driven by a speed controller with current limiting can be
found in the speed controllers section.

2.5. Power and efficiency

The power a motor is taking from the battery is

P = [V - Ea]ia or P = [V - Ea]2/Ra

The power the motor is putting into the load is

P = TΩ

When the motor is running with no extra load other than its
friction, then the load torque is zero, and so the output
power is zero. The input power is small also because the
back emf is almost equal to the supply voltage. The only
power drawn from the battery is driving the loss torque:

Ploss = Tloss Ω

When the motor is driving such a large torque that it has

stalled, then the back emf is zero, and the input power is
very large:

P = V2/Ra

and the output power is zero because the speed is zero.

At some point in between, the output power is at a

maximum. If we plot the product of the torque and speed,
we will have the graph for power:
The power versus torque graph is a quadratic curve, with a
maximum which the manufacturer specifies as the power
rating of the motor. For a starter motor on a small car, this is
typically around a kilowatt.

The efficiency of the motor is defined as the amount of

mechanical power you are getting out of the motor divided
by the amount of electrical input power you are putting in,
so:

The efficiency can also be plotted against load torque. Here

is the efficiency graph for an Iskra shunt wound starter
motor:
Just because the maximum efficiency is at around 1Nm
doesn't mean that you have to run the motor at that torque
though!

There is a section about the power rating of DC motors in the

spinning disks chapter here.

2.6. Characteristic graphs for some motors

Most of the graphs presented above were based on example

values. Let's look at the characteristic graphs for some real
motors:

Iskra 12 Volt 900 Watt shunt wound starter motor

Bosch GPA750
Lynch 12V
Lynch 24V
Scott 4BB-02488 motor (datasheet)
Scott 4BD-1460 motor datasheet.

The Bosch GPA750 has the following nameplate values:

Parameter Value
Nominal voltage 24 V
 Nominal power 750 W
Nominal current 40 A
3300
Nominal speed
rpm
Continuous
2.2Nm
torque
Stall torque 11 Nm

Graphs for other motors, and sometimes the same motors at

different voltages can be found at the following sites:

Lynch (LEMCO) motors

Iskra motors
Scott motors
Bosch motors

3. Testing

When designing your robot, it is very useful to know how fast

the motors will be going under various load conditions.
Basically, this means knowing the exact shape of the speed-
torque graph described in section 2.4. This will tell us how
fast the motors will go on no-load, and how much torque
they will develop when they are starting or stalled. With the
gearing information and wheel sizes, this will tell us how
much force your robot will be able to push, and give us an
idea of the maximum speed of the robot.

To fully test the motor, you will need the following

equipment:

• Speed measuring device

• Spring balance (cheaply available from angling shops)
• load band
• digital voltmeter (with 0-2V range) and ammeter (0-
10A)
The speed measuring device must be able to tell you the
RPM of the motor. This may be the hardest piece of
equipment to get hold of. Stroboscopes used for adjusting
the timing of car engines are generally not of any use, since
they require a reference pulse from the spark plugs. Here
are some methods that you could use to perform this:

• Independant stroboscope with dial reporting flash

frequency. A chalk mark is made on the shaft of the
motor. When the motor is rotating, adjust the frequency
of the stroboscope upwards from its lowest setting until
the chalk mark appears to be stationary. Then read off
the frequency from the stroboscope dial. The rpm is the
frequency x 60.
• Home made LED stroboscope.
• Frequency counter. Attach a piece of card to the end of
the motor shaft with a single slot (or two diametrically
opposite if just one slot causes the card to become
unbalanced and fly off) cut in it, similar to the speed
measuring device described in the speed controllers
page here. The rpm is then given by the frequency read
from the frequency counter, multiplied by 60 and
divided by the number of slots cut in the card.

Spring balances are available from general hardware stores,

and angling shops. They are normally used for weighing. The
load band is a tough piece of material that wraps around the
wheel (which must be attached to the motor), that won’t
disintigrate too quickly. A piece of old seat belt from a car in
the scrap yard is ideal. Here is the test setup:
Fit the motor into the apparatus as shown. There must be a
means of adjusting the height of the motor so that band can
be tightened. Start it off with the band stretched slightly
tightened (but not too tight). Run the motor, and record the
speed it is going at (call this w1). The spring balance should
change reading a little (if it doesn’t then the band is not tight
enough, so move the motor and wheel downwards a little),
record the spring balance reading (call this F1).

Now move the motor and wheel further down (tightening the
band further) and repeat. make sure that the speed and
spring balance reading have changed appreciably - enough
to see the difference, but not so much that it stalls the
motor. Record the speed (w2) and spring balance readings
(F2) again.

The spring balance reading must be converted to Newtons

before we can use it. The speed must be converted to
radians per second also. The following table will help you do
these conversions.

To to multiply by
convert
 kg N 9.81

g N 0.00981

lb N 4.45

oz N 0.278

rpm rad/s 0.105

revs/sec rad/sec 6.28

The load torque on the motor in the tests is F × r where F is

the force in Newtons read by the spring balance, and r is the
radius of the wheel (in metres).

By gradually tightening the band, the shape of the torque-

speed graph can be established. For a PM or shunt motor, we
already know the graph is almost a straight line, so if we
knew the stall torque, and the no-load speed, we can just
draw a straight line between them. For series motors, the
band must be tightened gradually to obtain several readings
so the curve can be plotted.

This apparatus is useful if you need to match more than one

motor for your robot. If you have two motors and are running
them with open-loop speed control, then if they are not
sufficiently matched for torque-speed characteristics, then
your robot will tend to swerve when you are trying to drive it
in a straight line.

3.1. Stall torque

The band appartus described above may not be suitable for

measuring the stall torque, since this can be quite large. An
alternative is to fix (weld / bolt or whatever) a bar to the end
of the motor shaft, perpendicular to the shaft. Mount the
motor in a vice horizontally. The bar should preferably be at
least half a meter long. Rotate the bar so it is horizontal, and
place bathroom scales underneath the end of it. You may
have to place a wedge between the bar and the scales so
you you know the exact length of the bar between the motor
shaft and the scales.

Now power up the motor so it will be forcing the bar

downwards onto the scales (make sure you have the
direction correct and it won't flip upwards!), and read the
scales. Do not keep the motor powered up for long as it will
be taking a very large current.

The stall torque can now be calculated:

Tstall = 9.81 x W x L

where W is the weight in kg and L is the length of the bar.

3.2. Maximum speed

We can get some idea of the maximum speed of the robot

now. Since we don’t know how much friction there will be in
the gearing, and in the robots transmission from wheels to
ground, we'll have to guess that! If, for example, the gearing
is set to reduce the speed by a ratio 4:1, and the wheels are
30cm in diameter, then the road speed of the robot will be:

no load speed (rads/sec) ÷ gear reducing ratio × wheel

radius - speed loss due to friction
If the no load speed is 200 rads/sec, say, and the speed loss
due to friction is 2 metres per second, then the road speed
will be 200 ÷ 4 × 0.15 - 2 = 5.5 metres per second, which is
about 12 miles per hour.

To measure the speed of the motor, you will need a

tachograph or Stroboscope . Projects to build both of these
are given in the circuits section of this web site. Click on the
names above to go to them.

3.3. Motor resistance

The motor resistance tells us how much current the motor

will take when it is stalled or when it is just starting up. For
PM and shunt motors, this is the sum of the brush resistance
and the armature resistance. For series motors, this will also
include the series field resistance. The resistance is normally
very small (generally less than 0.1 Ohms), and so cannot be
measured with a simple digital or analogue multimeter.
There are several ways that it can be done.

3.3.1. Low resistance meter

This is a measuring instrument specially designed for

measuring very low resistances. It may also be an LCR
bridge which can measure the inductance also. Fix the rotor
tight so it cannot rotate (so no back emf is induced), then
use the instrument to measure the terminal resistance. I
used one of these to get the following results for a Ford
Fiesta starter motor:

Rtotal = 0.041Ohms

Ltotal = 200μH

You can build your own low resistance meter. Some circuits
are presented here.

3.3.2. Use another load in series

This is probably the easiest way if you can’t do (1). There are
several easy methods. What we are trying to do is pass an
appreciable current through the motor so we can measure
the voltage dropped across it. The rotor of the motor under
test must be fixed.

This requires two motors. One is the motor under test with
its rotor fixed, and one is allowed to run freely. To pass
current through the motor under test, simply apply a load to
the free running motor (trying to slow it down). As its load
increaes, it will take more current, and the voltage drop
across our motor under test will increaese. You want as large
a current as you can measure with your ammeter, so that
the voltage is large enough to measure.

3.3.3. Use a clamp meter.

These are meters that can be clamped onto or around wire,

and measure the current flowing in the wire by the magnetic
effect. AC current is measured using a single coil in the
meter as the secondary of a current transformer. To
measure DC current, the meter measures the voltage of a
Hall effect device in the meters clamp.

Maplin sell one for £100 (part number LD17T), which also
includes all the normal ranges of a DMM. If you do not
already have a DMM, it may be worth investing in this one. It
also has a frequency measurement range up to 20kHz which
would be useful when measuring motor rpm. Note that the
other similar cheaper meters to this one in Maplins range do
not measure DC current, only AC and so are not suitable.

If you already have a DMM, you can buy a clamp attachment

which just plugs into your DMM. These can be quite
expensive (well over £100 just for the probe, but CPC supply
one (order code IN00710) made by Tenma (model number
72-6174) for £30.

4. How much current will my motor take?

I'm often asked this question, so here's a whole section just

to answer it!

It's impossible to say how much current they will take on

average, because it depends completely on what the motor
is doing at the time.

An ideal electric motor (this doesn't exist of course but it is

what motor manufacturers aim for), will convert 100% of the
input electrical power into output mechanical power. This
means that if the motor is not required to deliver any
mechanical power - for example the vehicle is rolling at
constant speed and there's no friction (I did say it's the ideal
case!) - then it will draw no electrical power.

The mechanical power output is the load torque, 'T' (how

much the vehicle is resisting the motor going round)
multiplied by the rotational speed of the motor, 'ω'. The
electrical input power is the voltage across the motor, 'V',
multiplied by the current drawn by the motor, 'I'. In this ideal
situation, the mechanical power equals the electrical power:

Pm = Pe

so

Tω = VI
Therefore, if the load torque is zero, Pm is zero. The voltage
will still be across the motor, so I will be zero - the motor
takes no current.

Of course the world is not ideal - there will always be friction

in real life - the wheel bearings, any gearing you have, the
wheels against the floor, the bearings in the motor itself etc,
and so T will never be zero. It follows therefore that 'I' will
never be zero either. There will always be a little bit of
current required to drive the vehicle even on a flat surface.
For starter motors, this can be reasonably large. The
problem with starter motors is that they are designed to do a
quick dirty job for a few seconds. Their main design
consideration is cost - they must be cheap for the car
manufacturers. Most do not have proper roller bearings, but
use a phosphor bronze sleeve bearing. The shaft of the
motor simply rubs against the very hard bronze. This
obviously causes a lot more friction that a roller bearing, but
the car manuafacturers don't really care about that - it
doesn't matter much for them. I have heard that it can take
up to 20 Amps just to overcome this friction, although I have
not experienced that myself. One alternative which is quite
easy and inexpensive if you have welding equipment is to
remove the bronze sleeves, and replace them with roller
bearings. A roller bearing shouldn't cost any more than £2
(US$3). This should reduce the frictional current to a couple
of amps at most.

As for the maximum current that the motor can take, this is
governed by the resistance of its coils and brushes inside.
Going back to an ideal motor, this would have zero
resistance. When this ideal motor is stalled, it will take an
infinite current, which will produce an infinite torque, and
push whatever blockage is out of the way. In a real motor,
the torque it can produce is not infinite, it is limited by the
maximum current that can flow through the motor. The
maximum current is simply the voltage applied to it divided
by the motor's internal resistance... Imax = V / R. This is
often called the stall current because it is the current the
motor will take when it is not rotating, but full voltage is
applied (i.e. it is stalled). A typical small car starter motor
may have an internal resistance of 0.04 Ohms, so at 12V the
stall current will be 12 / 0.04 = 300 Amps. Using our old
equation Tw = VI, this maximum current corresponds to a
maximum torque, which is often called the stall torque, and
is how hard the motor will twist if you hold the shaft
stationary. The motor is not rated to take this current for any
length of time. Typically it will be rated for continuous use at
about a quarter of this stall current.

In between these two currents, the motor may take any

current, which will depend on the mechanical load. When
accelerating the vehicle or pushing against an opposing
force, it will take more, when decelerating or going downhill,
it may take zero current.

5. Links

Theory of starter motors. Quite a good page.

http://www.apra.org/publications/electrical/fordther.htm

Controlling voltage spikes (motor suppression). SGS

Thomson Acrobat document.
http://us.st.com/stonline/books/pdf/docs/1703.pdf

Using car windscreen wiper motors

http://www.geocities.com/sprite-midget/wipers.htm

5.1. Motor manufacturer's sites

Lynch electric motor company