DC Motor Braking and Speed Reversal.

(Basic theory and Modified UHU board analysis)

DC Motor Electrical Model

A simple electrical model for a DC motor is shown in the following figure. The back
Electromotive Force (EMF) generator is a variable voltage source which is a linear
function of motor speed. The series resistance is usually very low and is the only factor
limiting the current under a locked rotor condition.

The inductance is very important to the operation of PWM motor controls. The higher the
inductance the lower the ripple current at a given frequency.

Fig 1. Electrical Model of a DC Motor

Reversing capability requires the use of an H-bridge in order to reverse the motor current.
There are many different ways to drive the H-bridge.

Chopping Techniques

Chopping capitalizes on the inductance of the load to maintain load current when the
driving voltage is removed. The driving voltage is normally supplied through power
switches, and diodes normally conduct across the load when the switches are opened.
Two different methods are common for chopping.

Preliminary release 02/17/2008 by Kreutz. 1

The more efficient method chops one low-side power switch while one high-side switch
is on. This is referred to as a two-quadrant PWM.

Two-quadrant PWM normally operates with a low duty cycle, as winding current is
charged principally by the supply voltage, yet winding inductance is discharged by the
voltage drop in the diode circuit (see figure below). Motor back EMF reduces the
effective supply voltage and increases the effective diode voltage drop, so
the duty cycle tends to increase with speed.

In a two-quadrant system, one transistor is turned on and the diagonally opposite

transistor is Pulse Width Modulated (PWM). Usually the bottom transistors are
modulated, however pulse width modulating the top transistors works equally well. If the
bottom transistors are used for PWM, the top transistors are used for steering. In order to
reverse the direction of the motor, the active “steering” transistor is turned off, the other
steering transistor is turned on. Again, the diagonally opposite transistor is PWM’ed.

Fig 2.Two-Quadrant Drive circuit with current limit

Preliminary release 02/17/2008 by Kreutz. 2

The main advantage of two-quadrant chopping is efficiency. Its main drawback is that it
can’t quickly decrease winding current. This can be very troublesome in position
feedback systems.
The reversing capability of this two-quadrant system is limited to static reversal. That is,
the direction cannot be changed while the motor is still moving. The following two
figures illustrate what occurs during dynamic reversal. Both of these figures use the
simple model for the DC motor. Take the case in the next figure where the FWD signal is
initially high. Transistor Q1 is on and transistor Q4 is PWM’ing. Current is flowing from
left to right in the motor, and the back EMF of the motor is positive on the left and
negative on the right. When Q4 is turned off, current will free-wheel through the diode of
Q3.

Fig 3. Forward Operation

Now suppose the back EMF is quite high and the FWD signal is toggled low as in the
next Figure. Q3 will turn on holding the negative side of the back EMF generator high.
The effective voltage on the positive side of the EMF voltage source is then the supply
voltage plus the back EMF voltage. The motor current will first decrease to zero and then
change directions and flow as shown in Figure 4. When Q2 is off, the back EMF of the
motor will cause current to flow downward through Q3, through the motor, up through
the diode of Q1 and back into the motor supply voltage. The current will increase in this
direction at the rate determined by the back EMF and will be limited only by the motor
resistance. If the back EMF energy is large enough it could easily destroy the P-channel
transistors.

Preliminary release 02/17/2008 by Kreutz. 3

 Fig 4. Motor Current after Reversal

The simple current limit circuit shown in Figure 2 will not limit the current during
reversal because (in a Two–Quadrant PWM circuit) the current is not flowing through the
sense resistor. Even if the over current condition could be detected; the upper P-channel
devices must be PWM’ed off in order to reduce the current. If the P-channels must be
PWM’ed, a four-quadrant control should be considered. Since the two-quadrant system
does not normally generate any braking forces, slewing the pulse period to zero and
reversing directions does not ensure the motor has stopped, and may again result in a
large current peak.

On the original UHU board the PWM method employed is the Four-Quadrant PWM,
four-quadrant PWM systems chop both switches, and circulate load current through two
diodes backwards into the supply. Again ignoring back EMF, four-quadrant chopping
produces a nearly symmetrical current waveform, as current rises due to the supply
voltage impressed on the load inductance, and decays due to reverse supply and load
inductance. With four-quadrant chopping, a motor can decelerate as quickly as it can
accelerate.

In addition to switching both, upper and lower Mosfets, a few additional functions are
required for reliable Four-Quadrant operation. In Two-Quadrant PWM there is an
inherent dead-time between conduction of opposite lower and upper switches so cross-
conduction is avoided. Four-Quadrant control immediately reverses the state of the
switches at torque reversal, thus requiring a delay between turning the conductive device
off and the opposing device On to avoid simultaneous conduction, high power losses
and/or power stage damage.

On torque reversal, energy stored on the rotating load is transferred to the power supply,
effectively charging the DC bus capacitor, a clamping circuit is required to dissipate that
energy and limit the maximum bus voltage. Sometimes an overvoltage protection is
triggered in order to disable the output when the bus voltage exceeds certain level.

Preliminary release 02/17/2008 by Kreutz. 4

Normal PWM configurations do not allow braking current to be modulated because the
braking current does not normally flow through the sense resistor. On our case (the UHU
drive) brake current flows through the Sense resistor, but creates a negative voltage at the
input of the peak current detector, so, it is not controlled.

Current control

Because the power devices have finite current carrying capability, one must consider the
effects of motor operation on the power devices. Motor current usually depends on load
torque and the commanded acceleration. The most demanding conditions on the power
electronics are usually a stalled or locked motor, or maximum acceleration with an inertia
load. If the pulse width is increased abruptly to quickly accelerate the motor, very high
currents may flow. This will cause an undesirable jerk on the motor and the mechanical
system. Worse, it could exceed the current rating of the power devices. A simple rate
limit does not protect the Mosfets under a locked motor or shorted condition. A simple
cycle-by-cycle current limit will limit the current and indirectly limit acceleration. This
allows full utilization of the power devices. In two-quadrant systems a single-sense
resistor may be used. Pulse width modulating the lower transistors allows the current
sense resistor to be conveniently located at the bottom. A simple circuit is shown in Fig 2.
That is also the method employed on the UHU drive, with some differences that will be
explained later on.

Referring to Figure 2, when the voltage across the sense resistor exceeds the reference
voltage, the comparator output will switch high. This will set the Reset-Set latch,
disabling the PWM signal and turning the lower transistor off. When the PWM input is
switched low the latch will be reset and the PWM signal will be enabled to turn on the
transistor on the next cycle. The effect is to limit the peak current on a cycle-by-cycle
basis. This current limit will permit sub-harmonic currents to flow, which can create
some audible noise.

The maximum average motor current will depend on the motor inductance and back
EMF. When the motor is under a stalled condition, the back EMF will be zero and a large
voltage will appear across the internal inductance and resistance of the motor. A low
inductance motor will have higher peak-to-peak ripple current, resulting in lower current
during the valleys between peaks and consequently a lower average current. This may
cause a small motor with high inductance to actually generate more stall torque than a
larger motor with low inductance. The average current through the power Mosfets and
diodes will also be a function of the motor parameters.

The worst case average diode current is for a low inductance motor since the current will
free-wheel most of the time. The N-channel transistor current is the opposite of the diode
current and will also be worst for a high inductance motor. The P-channel steering
transistor current is the same as the motor current and will be worst for a high inductance
motor. If the motor drive circuit will be used with a variety of motors, both transistors
and the free-wheel diode should be chosen to handle the cycle-by-cycle current limit
value on a continuous basis.

Preliminary release 02/17/2008 by Kreutz. 5

The UHU drive uses a Four–Quadrant PWM method, but also adds a 50% duty cycle for
zero average current through the motor (no movement). So maximum PMW duty cycle
will make the motor move CW while minimum PWM duty cycle makes it move CCW.
Maximum and minimum PWM duty cycles are limited to values >0 and < 100% due to
the need to recharge the bootstrap capacitors on the high side Mosfet drivers. This
method although very efficient in removing the energy accumulated on the armature’s
inductance, is demanding on the circuit losses, requiring bigger heat-inks and faster turn-
on and turn-offs in order to minimize losses.

The original circuit made use of the 500 nSec dead-time inherent for the IR2184 in order
to avoid cross-conduction losses. To help controlling the switching times and avoiding
cross-conduction, the designer used asymmetrical turn-on and turn off Mosfet gate drive
circuits. In order for the original drive circuit to function properly it is necessary to use
the original Mosfets or Similar Mosfets with comparable total Gate charge.

On the modified UHU board we changed the circuit to use the IR21844, basically making
the dead-time adjustable. We also made the switching times shorter and introduced an
RLC output filter in order to control dV/dt at the motor terminals. That will make the
motors more reliable because their internal insulation will be less stressed, they will work
a lot cooler too.

The modified circuit, nevertheless, kept the original cycle–by–cycle peak current limiting
circuit, with only a small functional change, the circuit will allow to work on the
intermittent torque rate of the motor for short time before limiting the current at the
maximum continuous torque rating adjusted by the user.

In order to protect the output transistors during sudden movement reversal at high speed,
it would be necessary to modify the torque control circuit from the simplest form in Fig 2
to a more complex circuit which should be able to make the current to flow through the
sense resistor under all circumstances (including during the free wheel interval), by using
the absolute value of the voltage measured on the sense resistor to control the current
during braking as well as during normal operation. The use of average current feedback
will also reduce the audible noise and instability due discontinuous inductor current at
light loads

On the UHU controller, the output from the SR flip-flop controls shuts-down inputs to
both IR2184/IR21844, turning off all the Mosfets on the H-bridge, the free-wheel current
will circulate through the opposite leg’s diagonal diodes back to the power supply
(regenerative energy current). The worst case is when reversing direction at maximum
speed and maximum power supply voltage. In that case, as discussed earlier, the Motor
BEMF will be in phase with the motor current, the limiting factors for the diode current
will be the armature’s resistance and the two diodes’ forward voltage drop (negligible at
high voltage and high rpms). Under those circumstances, when the diode rates are
exceeded, the Mosfets will be damaged.

Preliminary release 02/17/2008 by Kreutz. 6

Due to the nature of the PWM mode used on the UHU board, there is never 100% or 0%
duty cycle, so the intermittent rates for the diode current applies, and in most cases no
series resistor is needed or only a small resistor would be required.

A mathematical analysis of the circuits is beyond the intent of this paper, but will be
provided later for the right calculation of the series resistor needed to limit the reverse-
braking current. Normal braking should not exceed Mosfet ratings unless the motor load
is driven by an external power source that makes its rpm to increase above the maximum
working rpm at maximum power supply voltage.

Bibliography:
Unitrode U-115
Motorola AN1317

Preliminary release 02/17/2008 by Kreutz. 7