Dc Motor Fyp Edittd (1)

7/30/2019 Dc Motor Fyp Edittd (1)

1/20

http://hyperphysics.phy-astr.gsu.edu/hbase/magnetic/motdc.html

DC Motor Operation

ByStevenon 3/28/2012 7:00 AM

Filed Under [Power Systems] |0 Comments

Popular tags:Motors,Theory,DC Systems

Coils of wire on the rotor carry a d.c. current which generates a magnetic field. A stator magnetic field is createdusing either permanent magnets or electrical field windings.Any current-carrying conductor placed within an external magnetic field experiences a force. Torque is produced inthe motor by this mechanism.As a d.c current provides a constant magnetic field, this needs to change each half revolution and is calledcommutation. Commutation is typically accomplished by metal brushes against the contacts. Where a field winding isused to generate the stator magnetic field, different motor characteristics can be obtained depending on how thiswinding is connected.

Operation Principals

For the equivalent circuit theparameters are given by:

Where:

- lap winding

- wave windinggiven that

n= number of conductor

p= number of pole pairs

= motor speed, rad s-1

= flux per pole, Wb

T= torque Nm

and

The power input to armature is given by:
http://hyperphysics.phy-astr.gsu.edu/hbase/magnetic/motdc.htmlhttp://hyperphysics.phy-astr.gsu.edu/hbase/magnetic/motdc.htmlhttp://myelectrical.com/stevenhttp://myelectrical.com/stevenhttp://myelectrical.com/stevenhttp://myelectrical.com/notes/categoryid/20/power-systemshttp://myelectrical.com/notes/categoryid/20/power-systemshttp://myelectrical.com/notes/categoryid/20/power-systemshttp://myelectrical.com/notes/entryid/153/dc-motor-operation#Commentshttp://myelectrical.com/notes/entryid/153/dc-motor-operation#Commentshttp://myelectrical.com/notes/entryid/153/dc-motor-operation#Commentshttp://myelectrical.com/notes/tag/motorshttp://myelectrical.com/notes/tag/motorshttp://myelectrical.com/notes/tag/motorshttp://myelectrical.com/notes/tag/theoryhttp://myelectrical.com/notes/tag/theoryhttp://myelectrical.com/notes/tag/theoryhttp://myelectrical.com/notes?tag=dc+systemshttp://myelectrical.com/notes?tag=dc+systemshttp://myelectrical.com/notes?tag=dc+systemshttp://myelectrical.com/filestorage/legacyImages/7230.DcMotorPrincipals.jpghttp://myelectrical.com/filestorage/legacyImages/7230.DcMotorPrincipals.jpghttp://myelectrical.com/filestorage/legacyImages/7230.DcMotorPrincipals.jpghttp://myelectrical.com/filestorage/legacyImages/7230.DcMotorPrincipals.jpghttp://myelectrical.com/filestorage/legacyImages/7230.DcMotorPrincipals.jpghttp://myelectrical.com/filestorage/legacyImages/7230.DcMotorPrincipals.jpghttp://myelectrical.com/notes?tag=dc+systemshttp://myelectrical.com/notes/tag/theoryhttp://myelectrical.com/notes/tag/motorshttp://myelectrical.com/notes/entryid/153/dc-motor-operation#Commentshttp://myelectrical.com/notes/categoryid/20/power-systemshttp://myelectrical.com/stevenhttp://hyperphysics.phy-astr.gsu.edu/hbase/magnetic/motdc.html


2/20

The armature power loss is given by:

The output power is given by:

The torque is given by:

Typically, IaRa is 3% of Vaand the power dissipated in field winding 3.5% of the motor rating.

Field Windings

Permanent magnets or field windings of a dc motor provide the necessary magnetic fields for the motor tooperate(often call excitation):

permanent magnets - provides a continuous and constant magnetic field by the use of permanently magnetizedmaterials (typically on only the smallest motors)

separately excited - separate windings and power supply provide the magnetic field for the motor

shunt winding - this is connected in parallel with the motor armature to provide the excitation

series winding - this is connected in series with the motor armature to provide the excitation

compound winding - is a hybrid between shunt and series where to field windings are connected both inparallel and in series with the armature

Characteristics of DC Motors


3/20

Speed Control of DC Motors

Separately excited, shunt

Constant power drive - Ia,Va is held constant and Ifvaried

Constant torque Ifis held constant andIa, Vavaried

DC Converter output Voltages

Semi converter, [DC converter equations do not lookcorrect. Please can someone check.]

Full converter, - with infinite inductive filtering
http://myelectrical.com/filestorage/legacyImages/1447.DcMotorCharacteristics.jpghttp://myelectrical.com/filestorage/legacyImages/1447.DcMotorCharacteristics.jpghttp://myelectrical.com/filestorage/legacyImages/1447.DcMotorCharacteristics.jpg


4/20

Standard Features:

500V Armature, 300V Field

Drip Proof, Fully Guarded Blower Ventilated, Thermostat

(DPFG-BV), Protection IP-22 US frame sizes

Power ratings are continuous load in40C ambient

Service factor 1.0, Class F insulation

Laminated Frame for greaterefficiency

DC Tachometer Generator 50V /1000RPM

Suitable for contant torque to 5% ofbase speed

Terminal Board in Junction Box foreasy connections

Pulse Width Modulator for 12 and 24 Volt applications

This circuit was featured in an article in Home Power Magazine #75

(C) G. Forrest Cook 1999

Introduction

A pulse width modulator (PWM) is a device that may be used as an efficient light

dimmer or DC motor speed controller. The circuit described here is for a general

purpose device that can control DC devices which draw up to a few amps of current.

The circuit may be used in either 12 or 24 Volt systems with only a few minor wiring

changes. This device has been used to control the brightness of an automotive tail

lamp and as a motor speed control for small DC fans of the type used in computer

power supplies.

A PWM circuit works by making a square wave with a variable on-to-off ratio, the

average on time may be varied from 0 to 100 percent. In this manner, a variable

amount of power is transferred to the load. The main advantage of a PWM circuit over

a resistive power controller is the efficiency, at a 50% level, the PWM will use about

50% of full power, almost all of which is transferred to the load, a resistive controller


5/20

at 50% load power would consume about 71% of full power, 50% of the power goes

to the load and the other 21% is wasted heating the series resistor. Load efficiency is

almost always a critical factor in solar powered and other alternative energy systems.

One additional advantage of pulse width modulation is that the pulses reach the full

supply voltage and will produce more torque in a motor by being able to overcome theinternal motor resistances more easily. Finally, in a PWM circuit, common small

potentiometers may be used to control a wide variety of loads whereas large and

expensive high power variable resistors are needed for resistive controllers.

The main Disadvantages of PWM circuits are the added complexity and the

possibility of generating radio frequency interference (RFI). RFI may be minimized

by locating the controller near the load, using short leads, and in some cases, using

additional filtering on the power supply leads. This circuit has some RFI bypassing

and produced minimal interference with an AM radio that was located under a foot

away. If additional filtering is needed, a car radio line choke may be placed in serieswith the DC power input, be sure not to exceed the current rating of the choke. The

majority of the RFI will come from the high current path involving the power source,

the load, and the switching FET, Q1.

SpecificationsPWM Frequency: 400 Hz

Current Rating: 3 A with an IRF521 FET, >10A with an IRFZ34N FET and heat

sink

PWM circuit current: 1.5 ma @ 12V with no LED and no load

Operating Voltage: 12V or 24V depending on the configuration

Theory

The PWM circuit requires a steadily running oscillator to operate. U1a and U1d form

a square/triangle waveform generator with a frequency of around 400 Hz.

U1c is used to generate a 6 Volt reference current which is used as a virtual ground

for the oscillator, this is necessary to allow the oscillator to run off of a single supply

instead of a +/- voltage dual supply.

U1b is wired in a comparator configuration and is the part of the circuit that generates

the variable pulse width. U1 pin 6 receives a variable voltage from the R6, VR1, R7voltage ladder. This is compared to the triangle waveform from U1-14. When the

waveform is above the pin 6 voltage, U1 produces a high output. Conversely, when

the waveform is below the pin 6 voltage, U1 produces a low output. By varying the

pin 6 voltage, the on/off points are moved up and down the triangle wave, producing a

variable pulse width. Resistors R6 and R7 are used to set the end points of the VR1

control, the values shown allow the control to have a full on and a full off setting


6/20

within the travel of the potentiometer. These part values may be varied to change the

behavior of the potentiometer.

Finally, Q1 is the power switch, it receives the modulated pulse width voltage on the

gate terminal and switches the load current on and off through the Source-Drain

current path. When Q1 is on, it provides a ground path for the load, when Q1 is off,the load's ground is floating. Care should be taken to insure that the load terminals are

not grounded or a short will occur.

The load will have the supply voltage on the positive side at all times. LED1 gives a

variable brightness response to the pulse width. Capacitor C3 smooths out the

switching waveform and removes some RFI, Diode D1 is a flywheel diode that shorts

out the reverse voltage kick from inductive motor loads.

In the 24 Volt mode, regulator U2 converts the 24 Volt supply to 12 Volts for running

the pwm circuit, Q1 switches the 24 Volt load to ground just like it does for the 12Volt load. See the schematic for instructions on wiring the circuit for 12 Volts or 24

Volts.

When running loads of 1 amp or less, no heat sink is needed on Q1, if you plan to

switch more current, a heat sink with thermal greas is necessary. Q1 may be replaced

with a higher current device, suitable upgrades include the IRFZ34N, IRFZ44N, or

IRFZ48N. All of the current handling devices, switch S1, fuse F1, and the wiring

between the FET, power supply, and load should be rated to handle the maximum

load current.

Construction

The prototype for this circuit was constructed on an IC proto board with parts and

wires stuck into the holes of the proto board. One version of the finished circuit was

used to make a variable speed DC fan, the fan was mounted on top of a small metal

box and the PWM circuit was contained inside of the box.

A simple circuit board (see picture) was built using a free circuit board CAD program,

PCB (1) that runs on the Linux operating system. The circuit board image was printed

on a PostScript laser printer onto a mask transfer product called Techniks Press-n-Peelblue film (2). The printed on film is then ironed on to a cleaned piece of single sided

copper clad board. The board is etched with Ferric Chloride solution. Holes are drilled

with a fine gauge drill bit, parts are soldered in, and the board is wired to the power

and load. This technique is great for producing working boards in a short time but is

not suitable for making large numbers of boards. A board pattern is shown in Fig 3,


7/20

this may be photo-copied onto a piece of press-n-peel blue film, or used in a

photographic etching process.

Alternately, the dead-bug construction method may be used. This involves taking a

piece of blank copper PC board, glueing a wire-wrap IC socket sideways to the board

with 5 minute epoxy, then soldering all of the parts to the wire wrap pins. Groundedpins can be soldered directly to the copper board.

It is advisable to check your wiring very closely before applying power the first time.

Wiring mistakes can cause instant destruction of sensitive semiconductor components.

Alignment

No alignment is required with this circuit.

Use

This circuit will work as a DC lamp dimmer, small motor controller, and even as asmall heater controller. It would make a great speed control for a solar powered

electric train. The circuit has been tried with a 5 Amp electric motor using and

IRFZ34N FET and worked ok, D1 may need to be replaced with a faster and higher

current diode with some motors. The circuit should work in applications such as a

bicycle motor drive system, if you experiment with this, be sure to include an easily

accessible emergency power disconnect switch in case the FET shorts out and leaves

the circuit full-on.

Wire the circuit for 12 Volts or 24 Volts as per the schematic, connect the battery to

the input terminals, and connect the load to the output terminals, be sure not to ground

either output terminal or anything connected to the output terminals such as a motor

case. Turn the potentiometer knob back and forth, the load should show variable

speed or light.

PartsU1: LM324N quad op-amp

U2: 78L12 12 volt regulator

Q1: IRF521 N channel MOSFET

D1: 1N4004 silicon diodeLED1 Red LED

C1: 0.01uF ceramic disc capacitor, 25V

C2-C5: 0.1uF ceramic disk capacitor, 50V

R1-R4: 100K 1/4W resistor

R5: 47K 1/4W resistor

R6-R7: 3.3K 1/4W resistor

R8: 2.7K 1/4W resistor

R9: 470 ohm 1/4W resistor

VR1: 10K linear potentiometer


8/20

F1: 3 Amp, 28V DC fast blow fuse

S1: toggle switch, 5 Amps

Parts Sources

Jameco 1-800-831-4242http://www.jameco.com/

Digi-Key 1-800-DIGIKEYhttp://www.digikey.com/

(1) FreePCB Software for Linux

(2) Techniks, Inc P.O. Box 463 Ringoes, NJ 08551 908-788-8249Press-N-

Peel

Printed Circuit Artwork

PWM PCB artwork inGIF format and inPostScript format

PWM parts silk screen inGIF formatand inPostScript formatSilkscreen clarifications: the unlabeled circular pattern near LED1 is a wire jumper

and the part labeled FB1 is really R9.

The connections for the 8 pin header on the printed circuit board are as follows:

Header layout as oriented

in PCB artwork:

7 5 3 1

8 6 4 2

1: VR1 clockwise

2: VR1 counter clockwise3: 12V Supply + and Load +

4: VR1 wiper

5: 12V Supply + and Load +

6: 12V Supply -

7: Load -

8: 12V Supply -

See thevariations on this circuitfor more ideas.

Backto FC's PWM Circuits page.

http://www.hitex.com/fileadmin/img/download/Basic_DC_Motor_Speed_PID_Control_With_The_Infin

eon_C167_Family.pdf

Basic DC Motor Speed PID Control With The Infineon C167 FamilyBasic DC Motor Speed PID Control With The Infineon C167 Family
http://www.jameco.com/http://www.jameco.com/http://www.digikey.com/http://www.digikey.com/http://sourceforge.net/projects/pcb/http://sourceforge.net/projects/pcb/http://www.techniks.com/http://www.techniks.com/http://www.techniks.com/http://www.solorb.com/elect/solarcirc/pwm1/pwmpcb.gifhttp://www.solorb.com/elect/solarcirc/pwm1/pwmpcb.gifhttp://www.solorb.com/elect/solarcirc/pwm1/pwmpcb.pshttp://www.solorb.com/elect/solarcirc/pwm1/pwmpcb.pshttp://www.solorb.com/elect/solarcirc/pwm1/pwmpcb.pshttp://www.solorb.com/elect/solarcirc/pwm1/pwmsilk.gifhttp://www.solorb.com/elect/solarcirc/pwm1/pwmsilk.gifhttp://www.solorb.com/elect/solarcirc/pwm1/pwmsilk.gifhttp://www.solorb.com/elect/solarcirc/pwm1/pwmsilk.pshttp://www.solorb.com/elect/solarcirc/pwm1/pwmsilk.pshttp://www.solorb.com/elect/solarcirc/pwm2/http://www.solorb.com/elect/solarcirc/pwm2/http://www.solorb.com/elect/solarcirc/pwm2/http://www.solorb.com/elect/pwm/http://www.solorb.com/elect/pwm/http://www.hitex.com/fileadmin/img/download/Basic_DC_Motor_Speed_PID_Control_With_The_Infineon_C167_Family.pdfhttp://www.hitex.com/fileadmin/img/download/Basic_DC_Motor_Speed_PID_Control_With_The_Infineon_C167_Family.pdfhttp://www.hitex.com/fileadmin/img/download/Basic_DC_Motor_Speed_PID_Control_With_The_Infineon_C167_Family.pdfhttp://www.hitex.com/fileadmin/img/download/Basic_DC_Motor_Speed_PID_Control_With_The_Infineon_C167_Family.pdfhttp://www.hitex.com/fileadmin/img/download/Basic_DC_Motor_Speed_PID_Control_With_The_Infineon_C167_Family.pdfhttp://www.solorb.com/elect/pwm/http://www.solorb.com/elect/solarcirc/pwm2/http://www.solorb.com/elect/solarcirc/pwm1/pwmsilk.pshttp://www.solorb.com/elect/solarcirc/pwm1/pwmsilk.gifhttp://www.solorb.com/elect/solarcirc/pwm1/pwmpcb.pshttp://www.solorb.com/elect/solarcirc/pwm1/pwmpcb.gifhttp://www.techniks.com/http://www.techniks.com/http://sourceforge.net/projects/pcb/http://www.digikey.com/http://www.jameco.com/


9/20

We do get asked for some strange things sometimes: Vauxhall Vectra fans may recall the original launch

TV advert which for a few seconds featured on-screen, a large multi-dialled clock, which was supposed

to show time speeding forward to catch up with the leap into the future made by the new Vectra.

Others might have suggested it was simply counting down the hours towards a cambelt failure.... click

here to view the advert on youtube

The clock was a stage prop designed and built for the advert by a London model-making company. It

now resides in a North London flat as a rather unusual coffee table. Unfortunately the expensive

variable speed-regulated motors used to drive the three dials on the clock face were reclaimed by the

production company and so despite the complex gearing system installed, it did not run.

A rather odd telephone call from the owner revealed that he was looking for some way to get it going

again at minimum cost. He had seen an advert in a hobby magazine for C167 starter kits and wondered

if there were some examples around of how to control the speed of a DC motor. Being helpful types and

major fans of the Vectra (no chance), we came up with a solution based on a recycled Phytec

miniMODULE167, a 12v DC motor,a fan, an infra-red LED, a photodiode and some simple C code

The objective was to make the clock run in "real time", accurate enough to keep good time for the

duration of the average dinner party but be able to run at high speed (as in the advert) on demand to

impress the guests, just after the traditional serving of pepp


10/20

The result was a simple Proportional-Integral-Derivative (PID) controller for a 12v permanent magnet DC

motor. PID is very widely used in industrial control systems and something we get asked for examples of

very frequeStrangely, a trawl of the Web revealed no C-coded

scratch. To make the clock run at a constant speed, here 600rpm, some form of accurate speed

regulator mechanism wasrequired. This would ensure that over time, the average motor speed wouldbe constant. The nature of the clock mechanism was that the load on the motor varies. For example, as

the various hands move, small load peaks occur which tend to disturb the running speed. The drag and

motor efficiency were also subject to change, particularly as a result of temperature. A more

appropriate motor drive mechanism to have used in this type of application would have been a stepper

motor but most requests we get ar

A means of altering the voltage applied to the motor to control

A means of measuring the absolute speed of the motor shaft o Some software running on a C167 to

keep the speed constant, despite

An input, here a potentiometer, to set the a

An interrupt to measure the motor speed

A periodic interrupt to works out the difference between the actual measured speed and the required

speed o Some way of converting the speed error int

If the motor runs at 600rpm when the PWM drive to the motor is at a 50% duty ratio, increasing this to

60% will make the motor try to run faster. Reducing the duty ratio to 40% will slow the motor down. In a

very simple "open loop" speed controller, the program, the potentiometer on analog channel 0 is read

to yield a value between 0-1023 (10-bits). This value is then fed into the PWM unit to allow the motorspeed to be varied.

In the real world, this type of controller is not very useful. While it allows the motor speed to be set, it

does not allow for changes in load and a basic flaw is that the absolute speed is not known unless an

external ta


11/20

Closing The Loop The closed-loop controller is a very common means of keeping motor speed at the

required "setpoint" under varying load conditions. It is also able to keep the speed at the setpoint value

where for example, the setpoint is ramping up or down at a defined rate. The essential addition to the

previous system is a means for the current speed to be measured. In the example, a three bladed vane

was attached to the motor shaft. An infra-red LED was obscured from the view of a photodiode by the

vane blades so that a series of pulses with a frequency proportional to motor speed is now available.

In this "closed loop" speed controller, a signal proportional to the motor speed is fed back into the input

where it is subtracted from the setpoint to produce an error signal. This error signal is then used to work

out what the magnitude of controller output should be to make the motor run at the required setpoint

speed. For example, if the error speed is positive, the motor is running too fast so that the controller

output should be reduced and vice-versa. The clever part is how the

drive is worked out.

At firssight it might be imagined thatsomething simple like "if the errspeed is negative, multiply it by

some scale factor (usually known as "gain") and set the output drive to this lethe voltage applied to the

motor is proportional to the error speed. In practice, this approach is only partially successfor the

following reason: if the motor is at the setpoint speed under no load there is no error speed so the

motor freeruns. If a load is applied, the motor slows down so that a positive error speed is produced.

The output increases by a proportional amount to try and restore the speed. However, as the motor

speed recovers, the error reduces and so therefore does the drive level. The result is that the motor

speed will stabilise at some speed below the setpoint at which the load is balanced by the error speed x

the gain. If the gain is very high so that even the smallest change in motor speed causes a significant

change in drive level, the motor speed may oscillate or "hunt" slightly . This basic strategy is known as

"proportional control" and on its own has only limited use as it can never force the motor to

The next improvement is to introduce a correction to the output which will keep adding or subtracting a

small amount to the output until the motor reaches the setpoint, at which point no further changes are

made. In fact a similar effect

this by another gain before adding the result the proportional correction found above. This new term is

based on what is effectively the integral of the error speed.


12/20

Thus far we have a scheme where there are two mechanisms trying to correct the motor speed which

constitutes a PI (proportional-integral) controller. The proportional term is a fast-acting correction which

will make a change in the output as quickly as the error arises. The integral takes a finite time to act but

has the ability to remove all the steady-state speed error.

A further refinement uses the rate of change of error speed to apply an additional correction to theoutput drive. This means that a rapid motor deceleration would be counteracted by an increase in drive

level for as long as the fall in speed continues. This final component is the "derivative" term and it is a

useful means of increasing the short-term stability of the motor speed. A controller incorporating all

three strategies is the well-known Proportional-Integral-Derivative, or "PID" controller.

For best performance, the proportional and integral gains need careful tuning. For example, too much

integral gain and the control will tend to over-correct for any speed error resulting in oscillation about

the setpoint speed. Several well-known mathematical techniques are available to calculate optimal gain

values, given knowledge of the combined characteristics of the motor and load, i.e. the "transfer

function". However, some simple rules of thumb and a little experimentation can yield satisfactory

results in practical applications. What The Gains Do Integral Gain: Ensures that under steady state

conditions that the motor speed (almost) exactly matches the setpoint speed. A low gain can make the

controller slow to push the speed to the setpoint but excessive gain can cause hunting around the

setpoint speed. In less extreme cases, it can cause overshoot whereby the speed passes through the

setpoint and then approaches the required speed from the opposite direction. Unfortunately, sufficient

gain to quickly achieve the setpoint speed can cause overshoot and even oscillation but the other terms

can be used to damp this out.

Proportional Gain: Gives fast response to sudden load changes and can reduce instability caused by high

integral gain. This gain is typically many times higher than the integral gain so that relatively small

deviations in speed are corrected while the integral gain slowly moves the speed to the sepoint. Like

integral gain, when set too high, proportional gain can cause a "hard" oscillation of a few Hertz in motor

speed.

Derivative Gain: Can be used to give a very fast response to sudden changes in motor speed. Within

simple PID controllers it can be difficult to generate a derivative term in the output that has any

significant effect on motor speed. It can be deployed to reduce the rapid speed oscillation caused by

high proportional gain. However, in many controllers, it is not used.

Practical Implementation Of A Simple PID DC Motor Speed Controller In the original example, a SK167

C167 starter kit was chosen to implement the PID controller in software. The C167 is way too powerfulfor such a task but this excess of number crunching power makes the software easier to write and

debug!

Basic Considerations The main PID controller routine was designed to be fairly general purpose and

hence modular. Whilst here it is used to control a DC motor, it could be re-deployed to other situations

where some parameter has to be controlled to a set value under varying conditions. The actual control

software is located in a single function and its major inputs and output are held in a structure. Although


13/20

it was designed originally for a specific job it is really only intended as an example of the basic

techniques involved and to allow those with no control system knowledge to experiment with a simple

PID system.

The routines that gather the inputs and process the output are kept in separate functions in another

module. The Keil C166 v4.01 compiler was used but it should not prove too difficult to port it to the

rather less common Tasking compiler, or indeed the 8-bit C505 or C515.

For simplicity and to allow the easy modification of the important PID gain parameters, the C167's 10-bit

analog to digital converter was used to derive 10-bit resolution inputs from simple trimmer

potentiometers. The PWM used to drive the motor was chosen as 10-bits so that motor speed can be

defined to approximately 0.1%, sufficient for most practical application. At this resolution, the C167

CAPCOM unit generates a 2.4kHz carrier which does produce some audiable motor winding ringing. The

C167's high resolution dedicated PWM unit could have been used to increase this to a supersonic

19.5kHz but to allow easy porting to other C167 variants this route was not taken. Fortunately the use of

a 10 bit resolution on the inputs and output makes some of the arithmetic easier!

The C167 IO pins are allocated as per: P2.0 CAPCOM channel 0 - optical chopper encoder input P2.1

CAPCOM channel 2 - PWM drive output P5.0 analog channel 0 - setpoint input P5.1 analog channel 1 -


14/20

derivative gain input P5.2 analog channel 1 - proportional gain input P5.3 analog channel 1 - integral

gain input

Designing The PID Controller Routine The PID control problem has to be converted from a theoretical

continuous process into a real "discrete" system running on a microcontroller. What this means in

practice is that the measuring of the setpoint and motor speed and the calculation of the output is onlyperformed a regular interval. In the context of a microcontroller, this might correspond to some code

run from a timer interrupt.

The PID controller can thus be expressed as: Output = Proportional Gain * (error_speed) + Integral_gain

* S (previous_error_speeds) + Derivative_gain * (error_speed - last_error_speed)

This is quite easy to implement as a C interrupt function. The timer 3 interrupt is easy to configure to

produce a regular interrupt event, with timer 2 acting as a reload register. The period of the interrupt

determines the effective sampling rate of the PID controller.

The service routine introduces the small refinement to the basic control strategy in the form of a"deadband". This stops the controller trying to constantly correct very small speed errors which in fact

can trigger instability.

/*** Work out current error speed ***/ this_error = PID.setpoint - PID.feedback_input ; /*** Check for

error speed being less than deadband width ***/ temp = this_error ; // Is error speed negative if(N) {

temp = -temp ; // Get absolute value } /* Is error speed within deadband? */ if(temp


15/20

derivative term is simply calculated thus: /*** Calculate Derivative Term ***/ derivative_term =

((long)(this_error - PID.last_error) * derivative_gain)/(long)Inputs_Scale_Factor ;

Given that the maximum error difference is 1023 and the derivative gain can never exceed 32767, there

is no possibility of an overflow from 16 bits in the result and so no saturation checks are made.

The integral term is based on the sum of all previous observed error speeds. The complication is that not

only must the accumulated error be checked for overflow but so must the resulting integral term. This

highlights one of the most tricky aspects of this type of software - overflows must not be left unchecked

as unlike in for example PC programs, an unexpected overflow can cause serious mechanical damage to

a real motor!

/*** Find Accumulated Error ***/ acc_error_temp = ((long)PID.accumulated_error) + (long)this_error ;

/*** Check For Accumulated Error Out Of Range ***/ if(acc_error_temp > (long)32767) { // Is error >

maximum value? acc_error_temp = 32767 ; // Limit to max +ve value } if(acc_error_temp < (long)-

32768) { // Is error < minimum value? acc_error_temp = -32768 ; // Limit to max -ve value

} PID.accumulated_error = (short) acc_error_temp ; /*** Calculate Integral Term ***/

integral_term_temp = ((long)PID.accumulated_error * (long)integral_gain) ; /*** Check For Integral

Term Out Of Range & Apply Saturation ***/ if(integral_term_temp > (long)((long)32767 *

Inputs_Scale_Factor)) { integral_term = 32767 ; } else { if(integral_term_temp 32767)

{

control_output_temp = 32767 ;

}


16/20

else

{

if(control_output_temp < 0)

{

control_output_temp = 0 ;

}

}

Generating the analog voltage to control the motor The C167 CAPCOM unit can be configured to

generate pulsewidth modulation with virtually no software overhead. To allow a very precise control of

the motor speed, a 10-bit PWM was chosen, yielding a theoretical


17/20

if(first_edge_mode) { first_edge_mode = Off ; // Period calculation next time... } else {

motor_period_temp = CC0 - last_CC0 ; // Calculate time since last edge if(motor_period_temp >

Speed_Jitter_Threshold) {

A very awkward real-world complication is that on the first edge as the motor starts to move, there is a

random high frequency switching with pulses of around 30us width. This is due to the poor quality of the

signal conditioning hardware in the photodiode circuit. In addition, it also produces an unwanted extra

30us pulse after the real falling edge. Fortunately the frequency of the unwanted "switching bounce" is

so high that if any such pulse is seen, it is ignored by checking that the motor_period is greater than athreshold corresponding to around 100us. Some might suggest that the effort required to solve this

problem might have been better spent designing proper hardware!

Stall detection In any speed measurement like this, the problem always arises as to what to do when the

motor stops. The longest period (i.e. lowest speed) that can be measured by the above means

corresponds to 0xFFFF counts. Fortunately, the DC motor used was not able to run stably below about 2

revs/sec so this limitation was not a problem. However, some means of actually forcing the period to

0xFFFF is required as the last measured period is not defined as there is no "last edge" in cases where

the motor stops dead. Therefore a special "stall detect" mechanism is required to cope with this.

Another CAPCOM channel (CC2) is used in a special mode where it simply generates an interrupt. On

each optical chopper edge, CC2 is set equal to the current timer 1 count + 0xFFF0. If no further edges

occur (i.e. the motor has stopped) the CC2 interrupt occurs and the motor period is forced to 0xFFFF.

The first edge flag is also set so that the speed measurement will restart correctly.

Getting The Motor Speed From The Period Measurement The PID controller expects a speed feedback

value in the range 0-1023. The motor frequency is likely to be in a completely different range so some


18/20

scaling is required. The scale factor is the motor period at the maximum possible motor speed divided

by the 1023 scale factor.

#define Max_Speed_Count 670 /* Timer 1 counts between edges at max speed */ #define

Motor_Speed_Scale_Factor (Max_Speed_Count*1023L)

The motor speed is found by dividing the scale factor by the measured motor period.

motor_speed = (unsigned long)Motor_Speed_Scale_Factor/motor_period_temp ; PID.feedback_input =

motor_speed ;

This results in the speed feedback being in the same units and dynamic range as the setpoint speed

simplifying the PID routine's error _speed calculation.

Setting Up The Controller This requires a bit of experimentation but a few simple rules can be applied.

PID interrupt period This perhaps the most basic setting. The optimum period is a trade-off between

speed of response and CPU loading. Bearing in mind that this job of this routine is to decide what

corrective action to take and then apply it, there is no point running the interrupt too fast: if the PID

controller makes corrections faster than the motor can react to them, instability is bound to result. In

the example, it was found by experimentation that it took around 40ms for any change in PWM duty

ratio to be reflected in a change in speed. Therefore the interrupt was set to run at this rate.

The PID_Sample_Period constant is set according to (required interrupt period)/(timer 3 count rate). In

the example this is 0.025ms/0.8us = 31250.

Speed Constants Max_Speed_Count is the number of timer 1 counts between optical chopper edges at

the maximum possible motor speed. In the example this was 670 counts and timer 1 runs at 3.2us per

bit, giving 670 * 3.2us = 2.144ms per edge.

Speed_Jitter_Threshold is the number of timer 1 counts which the erroneous jitter pulses correspond to.

In practice this is simply Max_Speed_Count less a 20% safety margin that prevents motor overspeed

being confused with jitter.

Output_Offset - unfortunately the motor used in the example did not start to move until at least 30% of

the 100% PWM duty ratio is applied. To stop the integral term having to wind up from zero to 30%,

which takes a significant time, a fixed amount is simply added to the output. The value of the offset was

chosen as being around 20% of the maximum controller output figure.

Setting Up The Gains These two controls need to set in tandem as one very much influences the other.

There are many ways to arrive at an initial setting of the gains but here is one we found worked

reasonably well; Ideally there needs to be some means of easily applying a zero to full speed step

change to the setpoint input or at least a single transition from the minimum to maximum expected

setpoint. This is so that the response of the controller to step changes can be observed. Set the setpoint

to maximum speed and with the integral and derivative gains at zero, increase the proportional gain so


19/20

that the speed reaches the maximum possible before a speed oscillation sets in. Reduce the setpoint to

zero. Repeatedly apply a step change in setpoint to 75% of full speed and increase the integral gain

gradually until the speed starts to overshoot.

The speed should now rise quickly with the step change and settle at the setpoint without significant

overshoot. The integral gain setting will be particularly influenced by the moment of inertia of the loadand some experimentation will be required. The controller is now configure as a proportional-integral

controller which should quickly correct speed errors without oscillation.

The derivative term can be introduced but in many applications it is not necessary. However some of the

fun of PID controllers is finding the right combination of values to give fast response to setpoint and/or

load changes. The derivative term is good at removing sudden speed changes or the rapid "hunting" of

speed that can result from the proportional gain being too high. It can also be useful in eliminating

overshoot when the integral gain is very high. The deadband constant can also be useful in preventing

the controller from continually tweaking the speed around the exact setpoint when the error is

insignificant.

Software Performance In such a system, the runtime of the main PID interrupt is critical. In the example,

at 20MHz with zero waitstates, 16-bit non-mux bus, the interrupt ran in a worse case time of 6us.

Mission Accomplished Going back to the original mission of getting the clock to run and keep reasonable

time over a few hours was achieved. The output speed needed to be 600 rpm so that the clock's 100:1

gearing would yield 1 rpm for the second hand drive. The controller's speed scaling constants were set

accordingly. Thus a setpoint of 256 bits produced a

motor speed of 600rpm. After tweaking the deadband width and integral gain, the speed had a steady-

state error of +/-0.8% with a peak transient error of +1.5%, -1%. This translates into the clock gaining orloosing 4 minutes over an evening, sufficient to give the appearance that it was a real timepiece,

especially if the guests were the worse for drink.

Future Improvements In many systems, the gains required for good performance at high speeds are

different from what is best at lower speeds. A mechanism for coping with this will be added.

A fundamental problem of simple PID controllers is "wind-up". This occurs when a heavy load is

suddenly removed. The accumulated error will be very high and the output at maximum. The speed will

increase rapidly until this error can be "burnt" off. The derivative gain can be useful in suppressing this

effect but a proper mechanism for handling it is really needed.

Finally, the controller can easily be adapted to receive its parameters via the CAN module and by the

time this Newsletter hits the streets, this should have been done.

Getting Hold Of The PID Controller Example The source code for the PID controller is available on an

"Open Source" basis for you to modify for your own purposes. It was written using the Keil C166 v4

compiler and uVISION2 workbench. It is included in the Infineon (Siemens) C166 starter kits, although

the motor and optical chopper will have to be added by the user! You can download the program from


20/20

here for just 1 (worth it if it means your project can at least get started!). A suitable C167 starter kit on

which to run the code can be found here or you can buy one on-line here.

The program has been written in such a way that it could be ported to an 8-bit CPU quite easily. The

obvious choice might be the C505C as this has both a CAPCOM unit and 10-bit AD convertor. We hope

to have such a version soon.

Further Material On PID Controllers There is some good and more academic material on PID control

theory and practice at several university sites on the web. At

http://www.eng.uwyo.edu/tutorials/matlab/ctm/ examples/motor/PID2.html you will find some

specific examples of DC motor controller tuning.

Dc Motor Fyp Edittd (1)

Documents

Transcript of Dc Motor Fyp Edittd (1)