r> Yahoo! is not affiliated with the authors of this page or responsible for its content.
AN42001 Low Cost Non-Dimming 220V Ballast Design 1
March 1996
Application Note 42001
Low Cost Non-Dimming 220V Ballast Design
GENERAL DESCRIPTION
This application note describes a high performance low
cost ballast design using the ML4831 electronic ballast
controller IC. The design can be evaluated by assembling
the parts listed in this document.
Operating over the range of 198 to 242V
RMS
this power
factor corrected 60W electronic ballast is designed to
power two series-connected F32T8 fluorescent lamps and
displays all the features of Micro Linears ML4831 ballast
controller IC. The mode of operation used for preheating
and striking of the lamps is the widely accepted variable
frequency, non-overlapping inverter topology.
Figure 1 displays the block diagram of this 220 volt ballast
design.
198-242
VAC
+
+
EMI
FILTER
BRIDGE
RECTIFIER
PFC
INVERTER
LAMP
NETWORK
ML4831
INTERRUPT
Figure 1
Applying AC line voltage to the ballast will supply start-up
power to the ML4831 enabling gate drive for the PFC
boost MOSFET Q1 and the inverter FETs Q2 & Q3. PFC
action generates a well regulated 380VDC supply for the
lamp inverter circuit and a low DC supply voltage for the
ML4831. The inverter stage consists of 2 totem pole
configured N-channel power MOSFETs with their
common node supplying the lamp network. The pair of
MOSFETs are driven out of phase by the ML4831 at a 50%
duty cycle.
The lamp network is single low pass LC section which,
when controlled from the ML4831, provides:
Adjustable Lamp Power
Required Lamp Starting Voltage
Controllable Preheating Filament and Lamp Voltages
Near Unity Power Factor
High Input Impedance During Starting or
Lamp-Out-of-Socket Conditions
Less than 5% Change in Lamp Current Over Line
Voltage Operating Range
The series connected lamps are across the output of the
network.
LAMP NETWORK DESIGN
The ML4831 allows the designer to select the filament
preheating frequency and the lamp starting/minimum
operating frequency.
The operation of the lamp network can be described by
these equations:
Open Circuit (lamp starting)
e
ein xc
xl xc
o
=
+
(1)
During Lamp Operation
Q
rl
rin
= 1
(2)
Where:
e
IN
=
Equivalent RMS Network Input Voltage
e
O
=
Open Circuit Network Output Voltage
VB
=
PFC Output Voltage
x
C
=
Reactance of Shunt Capacitor
x
L
=
Reactance of Series Inductor
Q
=
Transformation Q of network
r
L
=
Equivalent Lamp Resistance at Po
r
IN
=
Transformed Value of r
L
Needed to Produce P
O
P
O
=
Desired Lamp Power (Arc and Filament)
High frequency measurements using reference lamps and
ballast, as described in ANSI Standards C82.3-1983 and
C78.375-1991, must be performed to determine lamp
current and voltage at the desired ballast factor. These
values are used to determine P
O
and r
L
.
Since the PFC uses a boost type converter:
VB > 2
×
V
RMS
line (max);
VB > (1.414) (1.1) (220), and thus
VB > 342VDC (380VDC is used).
The RMS amplitude of a square waves fundamental is 2/ times its peak-to-peak value.
So,
e
IN
= 0.45VB = 171V
RMS
And by assuming negligible losses in the reactances:
r
e
P
IN
IN
O
=
=
= 2
2
171
55
533
(3)
by John Sampson
REV. 1.0 10/25/2000 Application Note 52
2
From high frequency reference ballast measurements,
F32T8 lamps operating at an 0.88 ballast factor,
Lamp Current = 0.18A
RMS
Lamp Voltage = 136V
RMS
Lamp Arc Power = 49 watts (total)
P
O
(total) = 55 watts (allocating 6 watts to filaments)
r
L
= 1500 (total),
and from equation 2, Q = 1.35
Values for x
C
and x
L
can be found from:
Q
x
r
r
x
L
IN
L
C
=
=
(4)
Thus x
C
= 1113 and x
L
= 718 , and, from equation 1,
e
O
= 481 V
RMS
.
CHOOSING THE STARTING/OPERATING FREQUENCY
The operating frequency, f
MIN
, was found by selecting a
frequency that makes the shunt network capacitor a
standard value using the values for x
C
and x
L
found in
equation 4. A 4.7nF capacitor makes f
MIN
30.5kHz and
the inductor 3.75mH.
For lamp rectification protection, line isolation, and lamp
out detection 33nF capacitors, C9 and C22, (see figure 5)
were added in series with the inductor (T3 primary) on
both the high and low sides of the line. The size of the
inductor was increased to 5.4mH to compensate for the
added capacitance.
Choosing the Preheating Frequency
The lamp starting scenario is the ballast feature that has
the greatest impact on lamp life. The ML4831, when used
with a properly designed lamp network, allows a designer
to select:
a) The preheating frequency which sets voltage across the
lamps and cathodes during the preheating interval
b) The length of the filament preheating time interval
ANSI C82.11-1993 sets the minimum preheating time at
0.5 seconds. A time of 0.7 seconds was used in this design
with 265V
RMS
across the lamps. 260 volts was selected to
ensure low glow current.
First select the voltage across the lamps during the
preheating interval. Then use equation 1 to find the
preheating frequency of 44kHz. This should be above
f
MIN
to make the network inductive.
Selecting Oscillator Components
Inverter frequencies f
MIN
and f
PREHT
, that were chosen to
be 30.5kHz and 44kHz, respectively, are 1/2 of the
corresponding oscillator frequency. Refer to the ML4831
Data Sheet for the equations and device parameters to
calculate r
T
and r
S
. To get a discharge time near 1
µ
s, the
value for c
T
was chosen at 2.2nF.
NETWORK INDUCTOR DESIGN
Since maximum stress on the inductor occurs during
preheating, these conditions were used for its design.
At f
PREHT
(44kHz):
x
L
=
1492 x
C
=
772 e
O
=
265V
RMS
V
V
IND
RMS
=
=
265
772 1492 513
(5)
The E 25/7 (EF 25) core was selected because:
Low Cost and Availability
High Ae (core) X Aw (bobbin)
Efficient size for 90VA at 30kHz and 2000 gauss
For 2 turn filament windings at 4.5V
RMS
during
preheating:
v
t
=
=
4 5
2
2 25
.
.
(6)
So the inductor turns are:
N
turns
=
=
513
2 25
228
.
(7)
Operating at an induction level of 2194 gauss during the
lamp preheating interval.
Voltage across the inductor during normal lamp operation:
V
x
v
r
v
x
V
IND
L
L
L
L
C
RMS
=


+
=





+





=
( )
2
2
2
2
1034
272
1500
272
1113
315
(8)
Where:
x
L
=
Reactance of T3 (primary) at f
MIN
v
L
=
Voltage Across Lamps
r
L
=
Total Equivalent Resistance of Lamps
x
C
=
Reactance of C8 at f
MIN
REV. 1.0 10/25/2000 Application Note 52
3
The filament voltages during normal lamp operation:
V
V
FIL
RMS
=
=
(
)( )
.
315 2
228
2 76
(9)
LAMP OPERATION
In equation 4 the entire load resistance is across the shunt
capacitor. By taking the filament load, r
F
, from a winding
on the inductor, a resistance larger than the r
IN
calculated
in equation 3 is transformed which reduces the power
available for the lamps. In this design, the filament power
is small compared to the lamp arc power, so the mismatch
is small. This effect can be mitigated by operating at a
frequency slightly higher than f
MIN
.
Network Z
IN
is found from the equation:
Z
IN
= (r
LS
+ r
FS
) + j (x
LS
+ x
CS
)
(10)
Where:
r
LS
=
Transformed Series Lamp Resistance
r
FS
=
Transformed Series Filament Resistance
x
LS
=
Transformed Series T3 (primary) reactance
x
CN
=
Total Series Reactance of C9 and C22
x
CS
=
Transformed Series Capacitive Reactance
of C8
Expanded:
Z
r
Q
r
n
Q
j x
Q
Q
x
x
Q
Q
IN
L
F
F
L
F
F
CN
C
=
+
+
×
+



+
×
+
+
+
×
+




1
1
1
1
2
2
2
2
2
2
2
(11)
Q
r
n
x
F
F
L
= ×
2
(12)
Q
r
x
and n
inductor turns
filament turns
j
j
j
L
C
=
=
=
+
+
+





+
+






+ +
=
+
+
= 1500
1 1 35
2 5 114
1 31
1034 31
1 31
316
1113 1 35
1 1 35
531 34
1033 316 719
565
2
2
2
2
2
2
2
2
.
( . )(
)
(
)(
)
(
)( .
)
.
(
)
(13)
As you can see from the results, the transformed filament
resistance slightly reduced the power to the lamps.
P
O
(new) = 51.8 watts
(14)
LAMP OUT PROTECTION
As with all resonant topology circuits, the highest
component stress occurs at open load. This can be
controlled by operating only close enough to resonance
to produce adequate starting voltage. When we chose the
starting voltage and derived f
MIN
, the open circuit input
impedance of the network was defined as:
Z
IN
= r
FS
+ j (x
LS
+ x
CN
+ x
C
)
= 34 + j(1033 316 1113) = 396 (15)
Where:
r
FS
=
Transformed Series Filament Resistance
x
LN
=
Transformed Series Inductive Reactance
x
CN
=
Reactance of Series Capacitors C9 and C22
x
C
=
Reactance of Shunt Capacitor, C8
When operating at the resonant frequency of 40.4kHz Z
IN
is 34 but by operating at f
MIN
, Z
IN
is 396 capacitive.
Although Z
IN
is relatively high, switching losses would
waste power and require additional heat sinking if not for
the ML4831s duty cycle interruption feature.
The ML4831 uses duty cycle interruption of the inverter
gate drive with the off time set by C13 and R15 (See figure
5). Refer to the ML4831 Data Sheet for information on
value selection. Unloaded conditions are detected by
sampling a voltage across C22, with a small capacitor.
This voltage is then DC restored, rectified, filtered and
applied to pin 9 of the ML4831. If a lamp is out of its
socket, or does not ignite for any reason, the voltage
across C22 will be more than 50% higher than when both
lamps are operating and is high enough to activate the
interrupt.
Ballast Performance
A typical ballast of this design will have the following
performanc