
AN-19
A
6/96
6
Calculation
In some cases, voltage and current measurements and calculation
will be the most straightforward method of determining the
power dissipation of a given component.  In the ST204A, this
method was used for C1 (input filter capacitor), and C2 (output
filter capacitor), T1, L1 and for the primary bias and secondary
feedback circuits.  Some digital oscilloscopes have sufficiently
advanced math functions to be able to directly calculate the
DC Thermal Equivalent Method
This is a very powerful method for obtaining estimates of power
dissipation, especially for components such as switching
transistors, MOSFETs, and power diodes, where there are
reverse recovery and switching losses as well as forward
conduction losses.  To use this method, the temperature rise in
a given component is measured.  The same temperature rise is
then induced in that component using a  DC current.  From DC
voltage and current measurements, the average power dissipation
in the component can then be determined.  This method was
used to determine the power loss in U1, D1, D2, VR1, L2, and
BR1 of the ST204A circuit.
Designing for Higher Efficiency
Almost every step in a flyback power supply design can affect
efficiency, from initial design considerations of maximum duty
cycle and transformer primary inductance to the choice of
components.  The following paragraphs examine the effect of
initial design parameters and component selection on efficiency.
Where possible, general design guidelines are given for selecting
components and initial design parameters.
Transformer Inductance
Higher efficiency can be achieved in a flyback design by
reducing the RMS operating current of the primary 
TOPSwitch
.
Lower RMS primary current can be achieved by providing
sufficient primary inductance in the transformer to allow the
supply to run in the continuous mode of operation.  This reduces
both  the peak and RMS currents in the primary, cutting
dissipation in 
TOPSwitch
, the transformer windings, the output
rectifier, and the output filter capacitor.   As an added benefit,
the amount of energy stored in the leakage inductance of the
transformer is reduced.  This stored energy scales as the square
of the peak current, and must be dissipated each switching  cycle
in the primary clamp circuit.  Reducing the primary peak current
by increasing the transformer primary inductance reduces
primary clamp losses. Design techniques for choosing the
optimum transformer inductance for a given application are
presented in AN-16 and AN-17.  Transformer construction
techniques are shown in AN-18.  The primary inductance of
transformer T1 in the ST204A circuit is 627 
μ
H.  At 30 W
output, this value of primary inductance results in continuous
mode operation over much of the input voltage range.  A
comparison between discontinuous and continuous mode
operation is shown in Appendix A.
If a wattmeter is not available, overall efficiency measurements
for a power supply can be made by applying high-voltage DC
current to the input, and making voltage and current readings
with conventional meters. Most switching power supplies run
equally well with DC or AC input. However, do not use this
method with a power supply having a fan or transformer
connected directly across the AC input, as these components
will present a short circuit to a DC input.  Efficiency
measurements will be 1-2 points higher with DC input than with
AC input, as the power supply will be running at unity power
factor, causing less stress on the input components.  Also, there
will be no line frequency ripple on the bulk filter capacitor,
allowing the supply to run at a higher average input voltage and
lower average current, dissipating less power.  However, DC
input measurements can be used to obtain approximate efficiency
measurements.  If a high-voltage DC supply is not available, a
rectifier bridge and filter capacitor can be used to generate a
high-voltage DC bus from the AC line.  Ripple on this supply
should be kept to less than 5% at the maximum power of the unit
under test to reduce  the  error  due to  line  frequency  ripple.
average power from voltage and current waveforms.  It should
be noted that some active current probes have delays on the
order of 50 nsec.  This can cause a significant amount of error
in measuring the instantaneous power dissipation of a switching
device.  A more reliable means of measuring power dissipation
in such cases is to use the DC  thermal equivalent method.
Figure 7. Wattmeter Connection Methods.
PI-965-032293
V
DUT
A
AC
INPUT
V
DUT
A
AC
INPUT
Wattmeter
Wattmeter
(a)
Incorrect Connection Method
Voltage drop in wires introduces measurement error
(b)
Correct Connection Method