Choosing the Inductor for a Buck Converter
In this article, Dr. Ridley examines how the value of a buck inductor should be
selected. A supposedly simple process can turn out to be much more
complicated than expected, and the range of allowable inductors is found to be
quite large.
Design rules for choosing the inductor
Five or six times a year, I teach a class in power supply design to 24 working
engineers. One of the design examples involves a buck converter, and the
design starts with the choice of the inductor value. I always ask the students
what value inductor should be used, or how much ripple current should be in
the inductor. (Ripple current ratio is usually defined as the peak-to-peak value
of the inductor current at high line, divided by the maximum load). Their
answers typically vary, from a value of 10% up to perhaps 30%.
A traditional value of inductor current ripple is 10%, and you will find this in
several books. Reference [1] uses this value as a starting point for design, but
does suggest at the end of the article that the value can be changed
depending on the desired output ripple. Further examination of literature
shows a huge range of recommendations. References [1] to [13] suggest
ranges from 2.5% to 50%.
References [14, 15] select the inductor according to the minimum load, with
the intent of keeping the converter always in continuous-conduction mode
(CCM). This can lead to a very large inductor if the light load is very small.
Which one of these is the correct value?
As is often the case in designing power supplies, there is no one correct
answer and it will always depend on the very specific converter that you are
working on at the moment.
Where do the rules come from?
We usually find in power electronics that design rules-of-thumb arise from
some practical basis in reality. There are several different factors that drive the
choice of inductor value.
1.
Output Ripple Voltage older output capacitor types tend to drive
inductor values to a higher number. Modern innovations in capacitor design
have driven the ESR down to very low values, and rarely is the output ripple
the driving factor in choosing the inductor.
2.
Inductor Loss Higher ripple current leads to higher RMS current in the
inductor, plus greater AC currents and higher proximity losses. Core losses will
also increase with larger ripple current. A higher value of inductor, with low
ripple, has higher dc conduction loss.
3.
Switch Conduction Loss the RMS current in the switch climbs with
inductor current ripple.
4.
Rectifier Conduction Loss the RMS current in the rectifier climbs with
inductor current ripple. This is an important factor when using a synchronous
rectifier, but less important when using a diode.
5.
CCM operation References [14, 15] choose the inductor to give a ripple
which is twice the minimum load. This is to avoid DCM operation at light load.
In the early days of power supply design, this was an important factor to keep
good transient performance under all conditions. With a modern power supply,
using current-mode control, there is no reason at all to keep the converter in
CCM operation.
Buck Converter Design Example
Searching deeper into the literature than the short list included with this article
will not lead you to any conclusions regarding the right value of inductance. For
every converter design that you do, the proper answer will depend upon your
very specific set of circumstances. And for each specific case, a detailed
design must be completed and tested before coming to any conclusions about
the right value.
For example, Figure 1 shows a specific design case. The switching frequency
of the buck converter is 200 kHz, and the output specification is 5A at 12 V
from a 18-36 V input. The output capacitor is preselected at 1000 F with a 10
m? ESR. The choice of the output capacitor can be as wide ranging as the
choice of the inductor value, and is not discussed in more detail in this article
due to space constraints.
Buck Converter with Parameter Values
The ripple current in the converter is maximum at high line, and the value of
the ripple is shown in figure 2 for an inductance of 160 H which gives a
current ripple ratio of 10%.
Inductor Current Waveforms 10% Ripple.
The peak-to-peak value of the current is 0.5 A, and this results in a 5 mV ripple
on the output of the converter. In most practical design cases, the output ripple
at the switching frequency is not the main driver for inductor choice, since it is
much lower than is needed. It is almost always far lower than the
high-frequency switching noise on the output.
There nothing special about choosing the 10% ripple value for Figure 1. It is
just one value in the range that could be used. For each value, we must
optimize the design of the inductor before we can really assess the
performance properly. This can be a time consuming process, but for this
study, it was automated using the design software POWER 4-5-6 [18].
The inductor designed for this case had the following practical design
constraints applied when assessing each value of inductor:
1.
Magnetics Inc. RM8 core with R material was used.
2.
Turns were set to the nearest integer value.
3.
Maximum flux level was designed for 0.3 Tesla, with 10% overcurrent
limiting.
4.
Maximum wire size was 20 awg (0.9mm diameter) to limit mechanical
stress on bobbin.
5.
Multiple strands of wire and multiple layers were used as appropriate.
6.
Winding loss includes proximity loss as predicted by Dowells equations
[16].
7.
Core losses include advanced modeling techniques [17].
There are hundreds of cores that could have been chosen, but the RM8 was
convenient and available. It will not necessarily be the optimal choice since the
definition of optimal varies for every user.
Table I shows the results for designs, with a range of inductance from 640 H
down to 8 H. This corresponds to a ripple ratio ranging from 2.5 % (as in the
example of [11]) up to 200%, the boundary of discontinuous conduction mode
at full load.
Ripple
(H)
Ratio
640
2.5
160
Turns
DCR
ACR
Winding
Core
Inductor
FET
Total
Loss (W)
Loss (W)
Loss (W)
Loss (W)
Loss (W)
221
1.7
53.8
43
43
0.41
43.41
10
58
0.112
5.5
2.94
2.94
0.26
3.20
80
20
31
0.30
1.5
0.88
0.003
0.883
0.25
1.113
40
40
17
0.10
0.4
0.41
0.015
0.425
0.25
0.645
20
80
10
0.006
0.14
0.23
0.067
0.297
0.26
0.557
10
160
0.003
0.03
0.24
0.179
0.419
0.31
0.729
200
0.002
0.02
0.25
0.264
0.514
0.34
0.844
Inductor and Switch Loss with Different Ripple Values Using RM8 Core
It is clear from the table that for this design case, with the chosen core size, the
largest inductor was a very poor choice, with a loss of 43 W in the inductor
windings. The 10% ripple case also has a higher dissipation than is desirable.
The designs from 40% ripple to 200% ripple were all excellent.
It is interesting to see that the big change in ripple from 10% to 200% (20 times)
only incurred a relatively small increase in conduction loss of the power FET.
The increase in the RMS current value does not change as dramatically as one
might expect.
Core losses were low for all designs, as would be expected when using a
ferrite material. This allows the inductor to operate at high ripple current
without a big penalty. (This would not be true for many of the lower-cost core
materials found in the standard component designs.) Overall, the overall
lowest loss value of inductor gave an 80% ripple ratio. The current waveform
for this case is shown in Figure 3.
Inductor Current Waveforms 80% Ripple.
Summary
Choosing the proper value of an inductance can have a tremendous impact on
the final size, cost, and efficiency of a buck converter. There is, however, no
single specific value that is correct for all converters. The value of the inductor
can be freely selected over a practical range from 10% to 200%. In every
power supply, the specific constraints impacting the design choice will be
unique, and the optimal value for the inductor must be found by trying multiple
values and, ultimately, testing the designs in the circuit.
The design example given in this article had an optimal ripple of 80%. If I had
to choose one specific number to get people started, I would agree with
Reference [2] and begin with a ripple factor of 40%, but with the clear
understanding that it is not a problem to move significantly away from this
number, and the number should not be blindly accepted.
Other Recommendations
1.
Jerry Foutz, Switching-Mode Power Supply Design Tutorial
Simple Switching Topologies, http://www.smpstech.com/tutorial/t03top.htm
start with 10% ripple.
2.
Sanjay Maniktala, Switching Power Supplies A to Z, 40% ripple.
Semiconductor Vendors
3.
Micrel, MIC4574 application note,
http://www.micrel.com/_PDF/mic4574.pdf, 10% ripple (Figure 3), 34% ripple
(Figure 1).
4.
GMOS Technology Corporation, GT1512 Datasheet, 15% ripple.
5.
Microchip, MCP1612 Datasheet, 16.5% ripple.
6.
Texas Instruments, TPS4000 controller reference design, 25% ripple.
7.
Analogic, 30-40% ripple.
8.
Sanjaya Maniktala, National Semiconductor, Current Ripple Ratio
Simplifies Selection of Off-the-Shelf Inductors for Buck Converters
http://powerelectronics.com/mag/power_current_ripple_ratio/, 25-50% ripple.
9.
10.
Maxim application note 30% ripple.
National Semiconductor, Selecting Inductors for Buck converters,
http://www.national.com/an/AN/AN-1197.pdf, 30-40% ripple.
Magnetic Component Vendors
11.
Douglas R. Kokesh, Tyco Electronics Power Components, CoEv
Magnetics Group, Sizing a Power Inductor - An aid to employment of switch
mode power supplies,
http://www.designfax.net/archives/0304/0304power_inductor.asp, 2.5% ripple.
12.
Coiltronics, 10-30% ripple.
13.
GB International Custom cores, Using GBIs 4400 Series to Design
Buck Regulators, http://www.gbint.com/Files/Apps/General
Apps/GB-4400-001.htm, 2 x minimum load.
14.
Wurth Electronics, Power Inductors 8 Design Tips, 30% ripple.
15.
Jim Holdahl and Terry VanConant, CoEv, Demystifying Buck
Inductors,
http://powerelectronics.com/mag/power_demystifying_buck_ind
uctors, 2 times minimum load.
Magentics Design Example References
16.
Ray Ridley, Proximity Loss in Magnetics
Windings, www.switchingpowermagazine.com/downloads/13 Proximity
Loss.pdf
17.
Ray Ridley and Art Nace Modeling Ferrite Core Losses,
www.switchingpowermagazine.com/downloads/7 Modeling Ferrite Core
Losses.pdf
18.
Ridley Engineering, Power 4-5-6 Simulation and Design Software,
http://www.ridleyengineering.com/software.htm
