6.11s Notes for Lecture 3 June 14, 2006 J.L. Kirtley Jr.

Download Report

Transcript 6.11s Notes for Lecture 3 June 14, 2006 J.L. Kirtley Jr.

6.11s Notes for Lecture 3
PM ‘Brushless DC’ Machines: Elements of Design
June 14, 2006
J.L. Kirtley Jr.
6.11s June 2006 L3
1
Cross Section View: Surface Magnet Machine: Note windings
6.11s June 2006 L3
2
Alternate: Surface Mount (‘Iron Free’) Armature Winding
6.11s June 2006 L3
3
Magnets Inside the Rotor
6.11s June 2006 L3
4
Machine Design for Very High (negative) Saliency
6.11s June 2006 L3
5
Focus on Rating:
q
q Ea
V
P  jQ  VI 
NI
2
2 N
Ea
Ea  

 
Rating is number of
phases times voltage
times current
Internal voltage is
frequency times flux
N
2R B1

kw
p
And flux is the integral of
Flux density
We will consider winding
factor below

6.11s June 2006 L3
6
Internal Voltage Construction: Here is flux Density from Magnets
This is an approximation to
the shape of the field in the
air-gap (only an
approximation)
B( )   Bn sin np 
Radial field
n
np m 
Bn  sin 
k gn Br
  2 
hm
kgn 
hm g
4
But see the notes for this
done right
6.11s June 2006 L3
7
Magnetic field can be found through a little field analysis
The result below is
good for magnets
inside and p not
equal to one. See
the notes for other
expressions
Stator winding outside:

Rsp1  p
p
p 1
p 1
2p
1 p
1 p
kg  2 p
R2  R1  Ri
R2  R1 


2 p 
Rs  Ri p  1
p 1

Stator winding inside:


Rip1  p
p
p 1
p 1
2p
1 p
1 p
kg  2 p
R2  R1  Rs
R2  R1 


2 p 
Rs  Ri p  1
p 1

6.11s June 2006 L3
8
Current Capacity
N slots
N a Ia  J slot Aslot
2q
N slotsAslot  2Rshs
Better :


N slotsAslot   R  hs   R 2  N shsw t
2
N sw t  2R1 s 
This begs two questions:

How to establish current density?
How to establish slot fraction?
6.11s June 2006 L3
9
Voltage Ratio
E  V  X d Ia   2VX d Ia cos
2
a
2
2
E  V  X d Ia   2VX d Ia sin 
2
a
2
2
 V 2 X d Ia 2
V X d Ia
1     
sin 
  2
Ea Ea
E a   E a 
2


V
X I
X I
 1  d a cos   d a sin 
Ea
Ea
 E a

6.11s June 2006 L3
10
Calculation of Inductance: Start with a Full-Pitch Coil Set
This current
distribution makes
the flux distribution
below
6.11s June 2006 L3
11
4 N a Ia
Br  0
sin p 
 pgm
Fundamental Flux Density

p
  N a  Br  Rd 
0
La   0
Idealized inductance of a fullpitch coil
4 R Na
 p 2 gm
3 4 R N a kw2
Ld   0
2  p 2 gm
3 4 R N a k w2
Ld   0
2  p 2 (g  hm )

Flux Linkage
Taking into account phase-phase
coupling (for 3 phase machine)
and winding factor
And for the PM machine the
magneti is part of the magnetic
gap
6.11s June 2006 L3
12
kw  k b k p

 fp 
This is what we mean by
short pitch: see the
original drawing
p
 B1 sin  Rd 
0

sp 

2R B1
p

2p 2p

 B sin  Rd


1
2p 2p

2R B1


sin
p
2
sp

kp 
 sin
 fp
2

6.11s June 2006 L3
13
Breadth Factor: Coils link flux slightly out of phase
Here is a construction of the
flux addition. It takes a bit of
high-school like geometry to
show that:
The breadth factor is just the
length of the addition of the
vectors divided by the length of
one times the number of vectors
m
sin
2
kb 

msin
2

6.11s June 2006 L3
14
Slot Leakage:
Suppose the slot were to look
like this: It actually has two
coils that have Nc half turns
each.
Flux linked by one coil from
one driven coil is:
  sN c2 Ia
hd 1 w s 
s  0 

w
3
h
 d
s 
Llc  N c2s

Use top of slot dimensions for
tapered slots: very small error
6.11s June 2006 L3
15
There are 2p(m-Nsp) slots with both coils in the same phase
And 2p Nsp slots with coils ineach of the different phases (in
each phase)
So slot leakage is
Lsl  4Llc  2 p(m  N sp )  Llc  2 pN sp
 Llc 8 pm  6 pN sp 

6.11s June 2006 L3
16
Winding resistance is important
Ra  
Aw
As
Aw  a
2N c
So there are various ways of
estimating winding length and area:
Area is easier:

Winding length must account
for end turns and that is a
geometric problem
w
Na
Nc 
2mp
w
 N a 2  2
e



6.11s June 2006 L3
17
We have power conversion figure out
Losses are:
Armature conduction loss: I2 Ra
Core Loss
Friction, windage, etc
To get core loss we use the
model developed earlier,
depending on the species of
iron and fields calculated
thus:
R
Bc  B1
pd c
Bt 
B1
1 s 

6.11s June 2006 L3
18
The Process of design is a loop
6.11s June 2006 L3
19
There are (at least) three types of performance specifications:
Requirements are specifications that must be met
a. Rotational Speed or frequency
b. Rating
Limits are specifications that must not be exceeded
a. Tip Speed
b. Maximum operating temperature
Attributes are specifications that, all other things being
equal, should be maximized or minimized
So the design process consists of meeting the
requirements, observing the limits and maximizing the
attributes
6.11s June 2006 L3
20
Multiple attributes make maximization iffy
No simple way of telling if A is
better than D (or C)
But B is clearly
superior to
(dominates) E
6.11s June 2006 L3
21
Novice Design Assistant:
Is deliberately not an expert system
Uses Monte Carlo to generate randomized designs
Each variable in the design space is characterized by:
Mean Value
Standard Deviation
Maximum value (limit)
Minimum value (limit)
Setup file (msetup.m) specifies
Number of design variables
For each: the above data
Number of attributes to be returned
function file called by nda.m: called attribut.m
returns attributes and a go-no-go (limits not violated)
6.11s June 2006 L3
22
Operation
For the PM machine fluxes are
given by simple expressions
So torque is:
Now normalize the machine
in the following way: probably
use field flux for
normalization
6.11s June 2006 L3
23
Then per-unit torque
is:
Per-Unit Currents to achieve the maximum torque per unit current are:
6.11s June 2006 L3
24
Note that per-unit flux achievable for a
given terminal voltage is:
And this is related to current by:
6.11s June 2006 L3
25
6.11s June 2006 L3
26
6.11s June 2006 L3
27
Base speed
6.11s June 2006 L3
28
Here is your basic three phase bridge
6.11s June 2006 L3
29
Suppose we have this situation:
6.11s June 2006 L3
30
Here is one way of switching that circuit:
The arrows designate when a switch is ON
6.11s June 2006 L3
31
Here is what is on in State 0:
Va = V, Vb = V, Vc = 0 Vn = 2V/3
6.11s June 2006 L3
32
Here is what is on in State 1:
Va = 0, Vb = V, Vc = 0 Vn = V/3
6.11s June 2006 L3
33
Here is what is on in State 2:
Va = 0, Vb = V, Vc = V Vn = 2V/3
6.11s June 2006 L3
34
Here is what is on in State 3:
Va = 0, Vb = 0, Vc = V Vn = V/3
6.11s June 2006 L3
35
Here is what is on in State 4:
Va = V, Vb = 0, Vc = V Vn = 2V/3
6.11s June 2006 L3
36
Here is what is on in State 5:
Va = V, Vb = 0, Vc = 0 Vn = V/3
6.11s June 2006 L3
37
Voltages: Line-Line Voltages are well defined
6.11s June 2006 L3
38
To generate switching signals:
•Totem Pole A is High in states 0, 4 and 5
•Totem Pole B is High in states 0, 1 and 2
•Totem Pole C is High in states 2, 3 and 4
This allows us to use very simple logic:
A = S0 + S4 + S5
B = S0 + S1 + S2
C = S2 + S3 + S4
6.11s June 2006 L3
39
To generate switch signals
Note that either top or bottom switch is on in each phase
Generation of states: we will do this a bit later (see below)
6.11s June 2006 L3
40
This ‘six pulse’ switching strategy:
•Makes good use of the switching devices
•Also requires ‘shoot-through’ delays
•Has very simple logic
We propose an alternative switching strategy
•Makes minimally less effective use of switches
•Uses a little more logic
•But does not risk shoot through
6.11s June 2006 L3
41
Here is a comparison of switching strategies
180 degree sixpulse
120 degree six
pulse
Give up a little
timing between
switch closings
6.11s June 2006 L3
42
Switches Q_1 and Q_5 are on: State0
Va = V, Vb = 0, Vc = V/2
6.11s June 2006 L3
43
Switches Q_1 and Q_6 are on: State1
Va = V, Vc = 0, Vb = V/2
6.11s June 2006 L3
44
Switches Q_2 and Q_6 are on: State2
Vb = V, Vc = 0, Va = V/2
6.11s June 2006 L3
45
Switches Q_2 and Q_4 are on: State3
Va = 0, Vb = V, Vc = V/2
6.11s June 2006 L3
46
Switches Q_3 and Q_4 are on: State4
Va = 0, Vc = V, Vb = V/2
6.11s June 2006 L3
47
Switches Q_3 and Q_5 are on: State5
Vc = V, Vb = 0, Va = V/2
6.11s June 2006 L3
48
This switching pattern results in these voltages
6.11s June 2006 L3
49
Switches turn on:
Q1
State_0 OR State_1
Q2
State_2 OR State_3
Q3
State_4 OR State_5
Q4
State_3 OR State_4
Q5
State_1 OR State_5
Q6
State_1 OR State_2
Each switch is on for two states
6.11s June 2006 L3
50
So here is how to do it
3 bit input to ‘138 selects one of 8 outputs
Active low output!
‘138 has 3 enable inputs: two low, one high
6.11s June 2006 L3
51
NAND (Not AND)
Is the same as Negative Input OR
The ‘138 output is
‘active low’:
Matching bubbles
makes an OR function
6.11s June 2006 L3
52
Now we must generate six states in sequence
If we have a ‘clock’ with rising edges at the right
time interval we can use a very simple finite state
machine
This could be a counter, reset when it sees ‘5’
6.11s June 2006 L3
53
Here is a good counter to use: 74LS163
This is a loadable counter: don’t need that feature
Clear function is synchronous: so it clears only ON a
clock edge
Part is ‘edge triggered’: changes state on a positive
clock edge
P and T are enables: must pull them high
6.11s June 2006 L3
54
And here are the counter states: note how CL works
6.11s June 2006 L3
55
We already detect state 5 with the ‘138
6.11s June 2006 L3
56
The ‘138 is a simple selector: use like this:
And here are the pinouts of the ‘163 and ‘138
6.11s June 2006 L3
57
Variable Voltage: do the Pulse Width Modulation thing
6.11s June 2006 L3
58
Nomenclature: Two more views of the machine
6.11s June 2006 L3
59
This is a ‘cut’ from the radial direction (section BB)
6.11s June 2006 L3
60
Here is a cut through the machine (section AA)
Winding goes around the core: looking at 1 turn
6.11s June 2006 L3
61
Voltage is induced by motion and magnetic field
Induction is:
E' E  v  B
Voltage induction rule:
V  u B  R B  C
Note magnets must agree!


6.11s June 2006 L3
62
Single Phase Equivalent Circuit of the PM machine
Ea is induced (‘speed’) voltage
Inductance and resistance are as expected
This is just one phase of three
Voltage relates to flux:
Ea  0
6.11s June 2006 L3
63
PM Brushless DC Motor is a synchronous PM
machine with an inverter:
6.11s June 2006 L3
64
Induced voltages
are:
Assume we drive with
balanced currents:
Ia  I1 cos t

2 
Ib  I1 cost  

3 

2 
Ic  I1 cost 


3 
E a  E cost   

2 
E b  E cost    

3 

2 
E c  E cost   


3 
Then converted power is:


Torque must be:

1
1
P  EI1 cos  0I1 cos
2
2
p
p
T  P  0I1 cos


2
6.11s June 2006 L3
65
Now look at it from the torque point of view:
T  I BR  CI

6.11s June 2006 L3
66
Terminal Currents look like this:
I1 
4

sin
120
4 3
I0 
I0
2
 2
So torque is, in terms of DC side current:

3
3 3
T  p0 I1  p
0 I 0
2


6.11s June 2006 L3
67
Rectified back voltage is max of all six line-line voltages
Va
Vc
Vb
0
12
Vab
<Eb>
6.11s June 2006 L3
68
Average Rectified Back Voltage is:

 E b 

3 3

6

3



30 cos tdt
6
0
Power is simply:
Pem 
3 3

0 I0  KI0


6.11s June 2006 L3
69
So from the DC terminals this thing looks like the DC
machine:
T  KI
E a  K
Kp
3 3

0

6.11s June 2006 L3
70
Magnets must match (north-north, south-south) for the
two rotor disks.
Looking at them they should look like this:
End A
End B
Keyway
6.11s June 2006 L3
71
Need to sense position: Use a disk that looks like this
6.11s June 2006 L3
72
Position sensor looks at the disk: 1=‘white, 0=‘black’
6.11s June 2006 L3
73
Some care is required in connecting to the position sensor
Vcc
GND
Channel 1
Channel2
(you need to figure
out which of these
is ‘count’ and which
is ‘zero’)
6.11s June 2006 L3
74
Control Logic:
Replace open loop with position measurement
6.11s June 2006 L3
75
Why do we need to PWM only the top switches?
What happens with you turn OFF switch Q1?
6.11s June 2006 L3
76