Title: A Low-Cost Soft-Switched DC/DC Converter for Solid-

Oxide Fuel Cells

SECA Technical Program Final Report

Initial Date: August 15, 2002

End Date: June 11, 2009

Prepared by Dr. Jason Lai

Report Issued: June 11, 2009
DOE Award Number: DE-FC26-02NT41567

Submitting Organization:
Virginia Polytechnic Institute and State University
302 Whittemore Hall
Blacksburg, Virginia 24061-0111
Telephone: 540-231-4741
FAX: 540-231-3362

Subcontractors: Electric Power Research Institute

Southern California Edison
DISCLAIMER

This report was prepared as an account of work sponsored by an agency of the United States
Government. Neither the United States Government nor any agency thereof, nor any of their
employees, makes any warranty, express or implied, or assumes any legal liability or
responsibility for the accuracy, completeness, or usefulness of any information, apparatus,
product, or process disclosed, or represents that its use would not infringe privately owned rights.
Reference herein to any specific commercial product, process, or service by trade name,
trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement,
recommendation, or favoring by the United States Government or any agency thereof. The views
and opinions of authors expressed herein do not necessarily state or reflect those of the United
States Government or any agency thereof.
 ABSTRACT

A highly efficient DC to DC converter has been developed for low-voltage high-current solid
oxide fuel cells. The newly developed “V6” converter resembles what has been done in internal
combustion engine that split into multiple cylinders to increase the output capacity without
having to increase individual cell size and to smooth out the torque with interleaving operation.
The development was started with topology overview to ensure that all the DC to DC converter
circuits were included in the study. Efficiency models for different circuit topologies were
established, and computer simulations were performed to determine the best candidate converter
circuit. Through design optimization including topology selection, device selection, magnetic
component design, thermal design, and digital controller design, a bench prototype rated 5-kW,
with 20 to 50V input and 200/400V output was fabricated and tested. Efficiency goal of 97%
was proven achievable through hardware experiment. This DC to DC converter was then
modified in the later stage to converter 35 to 63 V input and 13.8 V output for automotive
charging applications. The complete prototype was tested at Delphi with their solid oxide fuel
cell test stand to verify the performance of the modified DC to DC converter. The output was
tested up to 3-kW level, and the efficiency exceeded 97.5%. Multiple-phase interleaving
operation design was proved to be reliable and ripple free at the output, which is desirable for the
battery charging. Overall this is a very successful collaboration project between the SECA Core
Technology Team and Industrial Team.
 TABLE OF CONTENTS

Executive Summary ........................................................................................................................ 7

1. Introduction ............................................................................................................................ 8
2. Experimental .......................................................................................................................... 9
2.1 Modify Existing V6 DC-DC Converter to Change from 400 V to 13 V Output ............ 9
2.2 Design and Build Interfaces .......................................................................................... 11
3. Results and discussion ......................................................................................................... 12
3.1 Efficiency Test .............................................................................................................. 12
3.2 Steady-State Test with Current Limit Checking ........................................................... 13
3.3 Dynamic Test ................................................................................................................ 15
3.4 Test the New Converter with Delphi Power Supply..................................................... 16
3.4.1 Efficiency Test ...................................................................................................... 16
3.4.2 Steady-State Test .................................................................................................. 17
3.4.3 Dynamic Test ........................................................................................................ 19
3.5 Test the New Converter with Delphi SOFC ................................................................. 20
3.5.1 Efficiency Test ...................................................................................................... 20
3.5.2 Current Ramp Test ................................................................................................ 22
3.5.3 Steady-State Test .................................................................................................. 22
3.5.4 Dynamic Test ........................................................................................................ 23
4. Conclusion ........................................................................................................................... 25
5. References .............................................................................Error! Bookmark not defined.
6. Patent Issued ........................................................................................................................ 26
7. LIST of Acronyms AND ABBREVIATIONS .................................................................... 26
Appendix: Key Schematic Circuit Diagrams ................................................................................ 28
 LIST OF FIGURES

Figure 1. Original Virginia Tech V6 DC-DC converter: (a) circuit diagram, and (b) photograph.
............................................................................................................................................... 10
Figure 2. Modified V3 DC-DC converter for step-down applications: (a) circuit diagram, (b)
photograph showing V3 converter under test, and (c) detailed view of V3 converter. ........ 11
Figure 3. Fixed input and output voltages test under different switching frequency conditions. . 12
Figure 4. Efficiency evaluation under variable fuel cell input voltage and fixed output voltage
condition. .............................................................................................................................. 13
Figure 5. Current limit unchanged 8 minutes after input source is removed when battery is
charged: (a) initial current limit of 48 A and (b) 8 minutes after the source is removed and
reapplied back. ...................................................................................................................... 14
Figure 6. Current limit unchanged 8 minutes after input source is removed when battery is
discharging: (a) initial current limit of 67 A and (b) 8 minutes after the source is removed
and reapplied back. ............................................................................................................... 14
Figure 7. Switch voltage and inductor current waveforms. .......................................................... 15
Figure 8. Dynamic voltage regulation under different voltage command conditions. ................. 15
Figure 9. Battery voltage after 2-kW load dump. ......................................................................... 16
Figure 10. Efficiency test results with SOFC simulators. ............................................................. 17
Figure 11. Steady-state voltage and current waveforms at 13.4 V, 28 A output test condition. .. 18
Figure 12. Steady-state voltage and current waveforms at 13.9 V, 185 A output test condition. 18
Figure 13. Voltage and current waveforms under load step test from 20-A to 60-A condition. .. 19
Figure 14. Voltage and current waveforms under load dump test from 85-A to 25-A condition. 20
Figure 15. Photograph showing setup of VT DC-DC converter tested with Delphi SOFC. ........ 21
Figure 16. Efficiency evaluation results with Delphi SOFC. ....................................................... 21
Figure 17. Voltage and current waveforms under current ramp control using Delphi SOFC
controller. .............................................................................................................................. 22
Figure 18. Steady-state voltage and current waveforms of DC-DC converter tested with SOFC
under light load condition. .................................................................................................... 23
Figure 19. Steady-state voltage and current waveforms of DC-DC converter tested with SOFC
under medium load condition. .............................................................................................. 23
Figure 20. Voltage and current waveforms under load step from 16-A to 28-A test with Delphi
SOFC..................................................................................................................................... 24
Figure 21. Voltage and current waveforms under load dump from 95-A to 60-A test with Delphi
SOFC..................................................................................................................................... 25
EXECUTIVE SUMMARY

This report is to describe the basic principle and summarize the test results of a high-efficiency
DC-DC converter. The DC-DC converter was developed under SECA core technology program
and tested with the SOFC developed by a SECA industrial team. The development was started
with topology overview to ensure that all the DC to DC converter circuits were included in the
study. Efficiency models for different circuit topologies were established, and computer
simulations were performed to determine the best candidate converter circuit. Through design
optimization including topology selection, device selection, magnetic component design, thermal
design, and digital controller design, a bench prototype was fabricated and tested. Efficiency goal
of 97% was proven achievable through hardware experiment. This DC to DC converter was then
modified to converter 35 to 63 V input and 13.8 V output with necessary interfaces for
automotive charging applications. The converter was tested at Delphi with their solid oxide fuel
cell test stand to verify the performance of the modified DC to DC converter. The output was
tested up to 3-kW level, and the efficiency exceeded 97.5%. Multiple-phase interleaving
operation design was proven to be reliable and ripple free at the output, which is desirable for the
battery charging. Overall this is a very successful collaboration project between the SECA Core
Technology and Industrial Teams.
1. INTRODUCTION

Virginia Tech has developed high efficiency DC-DC converters for low-voltage fuel cells
for the SECA core technology program. The original Virginia Tech 6-phase DC-DC converter
(V6 converter) was designed for 20 to 50 V input and 400 V output for a typical DC-AC inverter
that can output 220 V AC power. However, the SOFC developed by Delphi is aimed for
automotive applications, which has a similar fuel cell voltage range, from 35V to 63V, but the
output is for 12-V vehicle battery and associated DC loads. Therefore, some modifications are
needed for the V6 DC-DC converter to work with the Delphi SOFC.
The major modification is to change the 6-phase output to 3-phase output to reduce the
number of inductors, and then to replace the transformer with the inductor to serve as a step-
down converter. Furthermore, a new interface board needs to be designed and built so the new
converter can communicate with the Delphi SOFC controller. The interface board takes
frequency command signals from Delphi SOFC control signals and converts them to internal
voltage and current commands for the DC-DC converter control.
Major efforts include:
(1) Modify existing V6 DC-DC converter to change from nominal 400 V to 13.8 V
output
(2) Design and build an interface board for communication
(3) Test the new converter with VT SOFC simulator
(4) Test the new converter with Delphi SOFC simulator
(5) Test the new converter with Delphi SOFC
(6) Final reporting

Basic specifications of the converter are listed as follows:

(1) Input: 35V to 63V with 36V as nominal
(2) Output: 13V to 16V with 13.8V as nominal
(3) Power: 3 kW
(4) Efficiency: >96%
(5) Cooling: Forced air cooling
(6) Size: Standard rack mount unit 5.25”H×19”W×16”D
(7) Interface: 20 to 900-Hz frequency for 0 to 200-A current setting
 During the preparation of modifying DC-DC converter, Virginia Tech received
tremendous help from Delphi technical team, especially from Rajaey Kased, who visited
Virginia Tech on August 28, 2008 to check out the functionality of the modified DC-DC
converter and then provided necessary suggestions for further modification so the test can be
smoothly conducted during the visit.
The final test was performed on September 25, 2008. Again, the Virginia Tech received
strong support from the Delphi technical team during the visit. The first test was performed with
Delphi SOFC controller but running with a DC power supply as the source. The test essentially
mimics the actual SOFC test condition. After rigorous tests with high voltage level (60 V) and
high current level (200 A) using power supply as the source and final calibration of the controller
parameters, the DC-DC converter was proven working very well under both steady-state and
dynamic conditions. The current ramp can also be precisely controlled. From no load to full
load, test results showed a voltage regulation of ±0.01 V for the 13.8-V output range and a
current regulation of ±1 A for the 200-A output range. Efficiency was consistently higher than
97% for the load above 1-kW conditions. No significant temperature rise or hot spot was
observed after the test. The temperature of key power components remains lower than human
body temperature after full load test. With the confidence of the converter performance, the
converter was then tested with actual SOFC. Entire converter test with SOFC was very smooth,
and the performance agreed with the test results obtained from fuel cell simulators.

2. EXPERIMENTAL

2.1 MODIFY EXISTING V6 DC-DC CONVERTER TO CHANGE FROM 400 V TO 13 V OUTPUT

The current V6 DC-DC converter output is connected to transformers for isolation and
voltage boost. After transformer, there is a rectifier board that converts high frequency AC to DC
and filter to 400-V DC. The new design needs 13 V, so the major change is to connect the
original 6-phase (V6) converter output through inductors to the battery and reduce the number of
phases to three (V3) to reduce the amount of magnetic components. Figure 1(a) shows the circuit
diagram of the original V6 converter, which contains a 6-leg bridge converter, a high frequency
AC transformer, and a rectifying and filtering stage. The input voltage was originally designed
for 20 to 50-V SOFC, and the output voltage was regulated at 400 V. Figure 1(b) shows the
photograph of the entire V6 DC-DC converter. The voltage boost function is obtained through
the high turns-ratio of the high frequency transformer.
 6-phase bridge converter
Solid-Oxide Fuel Cell

20V
to 400V
HF AC
50V Xformer

HF AC
Xformer Rectifier+LC filter
Rectifier+LC filter
6-phase bridge converter

(a) (b)

Figure 1. Original Virginia Tech V6 DC-DC converter: (a) circuit diagram, and (b) photograph.

Figure 2(a) shows circuit diagram of the modified three-phase (V3) DC-DC converter
designed for Delphi SOFC. The power circuit is simplified from six phases to three phases, so
the number of output inductor remains three. The three phases are operated 120° apart;
therefore, the original programmed phase sequence in V6 converter does not need to be changed.
The input voltage is higher than the original design level, so the devices need to be changed to a
higher voltage rated MOSFET. The new device is rated 75 V, which should be sufficient to
handle 63-V input. Notice that the bottom side MOSFET switches of the 3-phase bridge are
paralleled with Schottky diodes to reduce the switching loss. The buck converter requires only
diode for the bottom side. However, a typical diode has a fixed voltage drop of 0.7 V at light
load conditions, which accounts for 2% conduction loss. Even with Schottky diode, the voltage
drop is about 0.4 V at light load conditions. Under the rated load condition, Schottky diode will
see at least 0.7-V drop. With additional upper MOSFET switch conduction voltage drop,
switching loss, inductor loss, capacitor loss, and parasitic losses, the theoretical maximum
efficiency will be less than 95%. Therefore, we propose to use power MOSFET as the bottom
switch to operate under synchronous rectification mode, while keeping Schottky diode to reduce
the switching loss under light load condition, so the efficiency maintains high over the entire
load range. The calculated efficiency is higher than 97% in most load conditions.
Figure 2(b) shows photograph of the V3 DC-DC converter and its test setup. The
physical size is significantly reduced from the original V6 converter because of the elimination
of transformers and the rectifier stage. Fig. 2(c) shows detailed view of the V3 DC-DC
converter. The overall circuit structure remains the same as the original V6 DC-DC converter,
but the output section is replaced with three inductors.

Precision current shunt resistors Battery

Load
12V battery/ bank
load
io
Vfc

35-63V +
SOFC Vo Ro
− r tes
t
rter u nd e
o nve
ase C
3-phase DC-DC buck converter 3- p h

(a) (b)
Output
bus bar

Inductor
Heat sink
Circuit
board
Input
bus bars

(c)

Figure 2. Modified V3 DC-DC converter for step-down applications: (a) circuit diagram, (b) photograph
showing V3 converter under test, and (c) detailed view of V3 converter.

2.2 DESIGN AND BUILD INTERFACES

The interface between SOFC and DC-DC converter involves power and control and
communication. The power connections include input and output terminals. We added two DC
contactors to help startup control for both input and output connections. The input contactor
cannot be turned on without output connected to a 12-V battery, while the output connector
cannot be turned on when the converter output and battery voltage difference is too high. A high
voltage difference tends to spark over the contacts of the DC contactor and damage the
mechanical contacts. If the converter output voltage is less than the battery voltage, turning on
the contactor can also damage power MOSFET because the circuit functions like a boost
converter that produces excessive voltage and can damage the power devices.
For the control and communication interface, the design options can be digital or analog.
Digital interface is less sensitive to noise interference but requires additional conversion. Analog
interface is simple, but the signal can be easily corrupted by the noise. Therefore, our choice is a
more reliable digital interface, which should be compatible with the Delphi fuel cell controller’s
20 Hz to 900 Hz. For the current control or current limit, this frequency range represents 0 to
200 A. For the voltage control, the frequency range can be specified to cover the battery voltage
state of charge range, typically from 11 V to 16 V.

3. RESULTS AND DISCUSSION

3.1 EFFICIENCY TEST

The new DC-DC converter has been packaged in a standard rack-mount case and tested
with the SOFC simulator at Virginia Tech. The first test was to test efficiency at different
switching frequency to see what frequency yields the best efficiency. Figure 3 shows the tested
efficiency profiles results under fixed input of 36 V and fixed output of 13 V but different
switching frequencies, i.e., 33-, 40-, and 50-kHz conditions.

98.5%
Vin = 36V
98.0% Vo = 13 V
33-kHz switching
97.5%
Efficiency

97.0% 40-kHz switching

96.5% 50-kHz switching

96.0%

95.5%
0 500 1000 1500 2000 2500 3000
Output Power (W)

Figure 3. Fixed input and output voltages test under different switching frequency conditions.
 Lower switching frequency tends to yield higher efficiency over the entire load range.
However, the efficiency gain with further reduction of switching frequency seems to be
diminished, especially at the heavy load condition, which can be attributed to the increase of the
inductor loss. Therefore, we decided to use 33 kHz as the switching frequency for the
subsequent tests.
Figure 4 shows the efficiency evaluation results under variable fuel cell input voltage and
fixed output voltage condition. The fuel cell voltage is reduced from 50 V to 35 V from light
load to heavy load. The efficiency is going upward as the load increases. The efficiency is
above 97% in most load conditions.

100% 60
Fuel Cell Voltage

Fuel Cell Voltage (V)

99% 50

98% 40
Efficiency

97% 30
Efficiency
96% 20

95% 10
Output voltage = 13.8 V regulated
94% 0
1000 1250 1500 1750 2000
Power (W)

Figure 4. Efficiency evaluation under variable fuel cell input voltage and fixed output voltage condition.

3.2 STEADY-STATE TEST WITH CURRENT LIMIT CHECKING

The purpose of test is to make sure that the current limit command does not change after
removing the input source. Figure 5(a) indicates that the initial test has a current limit of 48 A
and the battery is charged with 8-A input. The fuel cell input source is removed for 8 minutes.
Figure 5(b) shows that after 8 minutes, the current limit remains unchanged. This proves that the
current loop control is stable and is not affected by the fuel cell voltage condition.
 Converter output current (48 A) Current limit unchanged

Battery input current (8 A)

Converter output voltage (13.8 V)

Figure 5. Current limit unchanged 8 minutes after input source is removed when battery is charged: (a)
initial current limit of 48 A and (b) 8 minutes after the source is removed and reapplied back.

The same test was conducted at a higher current level with battery discharging the current
to the load. Figure 6(a) indicates that the initial test has a total load current of 67 A, and the
battery is outputting 7 A. The input source is removed to allow the battery sourcing the load.
Figure 6(b) shows that after 8 minutes, the current limit remains unchanged. Therefore, no
matter the battery status is in charging or discharging mode, the current loop control is stable and
is not affected by the fuel cell voltage condition.

Converter output current (67 A) Current limit unchanged

Battery output current (7 A)

Converter output voltage (13.8 V)

(a) Initial condition (b) After 8 min. holding time

Figure 6. Current limit unchanged 8 minutes after input source is removed when battery is discharging: (a)
initial current limit of 67 A and (b) 8 minutes after the source is removed and reapplied back.

Figure 7 shows the voltage and current waveforms of three-phase mode under 60-V input
voltage and 100 kHz switching frequency. Voltage overshoot and undershoot at turn-off and
turn-on are observed, but since the peak overshoot voltage is less than the device rating of 75 V,
the device should operate safely. Note that the converter is capable of operating at 100 kHz, but
the actual operating frequency was determined to be 33 kHz to ensure high efficiency operation.
 Inductor current (5 A/div)

Vds votlage (20 V/div)

(10 V/div)
gate signal

2 µs/div
Turn-off Turn-on

Figure 7. Switch voltage and inductor current waveforms.

3.3 DYNAMIC TEST

Under the current limit condition, increasing the voltage command will not increase the
voltage because the duty cycle is saturated. However, for the same current limit condition,
decreasing the voltage command will reduce the voltage and current, as shown in Figure 8. The
initial output current limit was set at 56 A, and the voltage was 11V. By reducing the voltage
command to 7 V, the current limit is reduced to 38A.

Converter output current

Converter output voltage

Voltage step down command

Figure 8. Dynamic voltage regulation under different voltage command conditions.

 Figure 9 shows the battery voltage under severe load dump condition. The initial load
current is 143 A, and the battery is 13.8 V, or 2-kW condition. The test is to dump the 2-kW
completely to see how battery voltage will react. Without current loop control, this normally
implies an excessive amount of energy will charge to the battery, which will increase the battery
voltage to a dangerous level. However, our design adopts a fast current loop control with 1-kHz
control loop bandwidth. The output voltage is also well regulated with 90-Hz control loop
bandwidth. Therefore, the battery voltage spike is very well contained during such a server load
dump. The measured voltage spike is no more than 3.5 V, which is mainly due to parasitic
ringing, and the actual battery cell voltage level should be much less. Therefore, the designed
fast control loop allows output voltage well regulated, and thus protecting the battery from over-
voltage damage even under the most severe load transient condition.

Converter output current

Battery charging current

Voltage spike due to
parasitic inductance

Battery terminal voltage

Figure 9. Battery voltage after 2-kW load dump.

3.4 TEST THE NEW CONVERTER WITH DELPHI POWER SUPPLY

3.4.1 Efficiency Test

With well proven steady-state and dynamic performance, the newly modified converter
was first tested with the Delphi power supply using actual SOFC controller. Therefore, the test
is considered as under SOFC simulator condition. To test efficiency under actual fuel cell
voltage condition, we selected some high voltage low current points and moved down the
voltage but increased the current command until full load condition. Figure 10 shows efficiency
evaluation results for the voltage range of 60 V down to 35 V and output current from 5 A up to
200 A. The efficiency is generally higher under heavier load conditions. For the first point with
load less than 250 W, the efficiency is only 70%. At 500 W, the efficiency quickly moves to
90%, and at 1 kW, the efficiency reaches 97%. There are two unusual efficiency points that
indicate an efficiency of >99% between 1 kW and 1.5 kW range. We believe that these two
points are invalid due to potential measurement error. As a comparison, we added three points
measured with VT fuel cell simulator, the efficiency should be around 97%. As the load
increases to full load (3 KW) range, the efficiency stays around 97.5%, which is consistent with
what was measured in VT lab. This result confirms that the proposed approach with
synchronous rectification operated power MOSFET in parallel with Schottky diode can break the
theoretical efficiency limit of 95% when only Schottky diode is used.

100%
95%
Efficiency

90%
With VT fuel cell simulator
85%
80%
75%
With Delphi fuel cell simulator
70%
60 0 500 1000 1500 2000 2500 3000
Fuel cell voltage (V)

With VT fuel cell simulator

50
40
30
20
10
0
0 500 1000 1500 2000 2500 3000
Fuel Cell Output Power (W)

Figure 10. Efficiency test results with SOFC simulators.

3.4.2 Steady-State Test

Figure 11 shows the steady-state tested voltage and current waveforms at 13.4-V, 28-A
output condition. With a high initial battery voltage, the load is mostly supplied by the battery,
not by the fuel cell because a small current limit is commanded by the converter, and thus the
output of fuel cell current is nearly zero, or 0.13 A as indicated in the chart. Under such a light
load condition, the switch duty cycle is quite small (23.7%), which is enough to maintain a stable
battery voltage of 13.4 V.
 Output voltage (13.4V)

Gating control

Load current (28A)

Fuel cell current (0.13A)

Figure 11. Steady-state voltage and current waveforms at 13.4 V, 28 A output test condition.

As fuel cell simulator continues charging the battery with a higher output current
command, the duty cycle and battery voltage also continue increasing. Figure 12 shows the test
condition at 13.8-V battery voltage and 186-A load current condition. The duty cycle is
increased to 40.8%, and the fuel cell voltage and current are 34.6 V and 76.2 A, respectively.

Load current (186A)

Gating control

Output voltage (13.8V)

Fuel cell current (76A)

Figure 12. Steady-state voltage and current waveforms at 13.9 V, 185 A output test condition.
3.4.3 Dynamic Test

Figure 13 shows voltage and current waveforms under load step test condition. The load
current is increased from 20 A to 60 A. The ramp is controlled to have a rise time of 60 ms.
With fast current loop control, the load current, and thus the fuel cell current rise smoothly
without any overshoot. The output voltage maintains well regulated at 13.8 V under such a
dynamic load step condition.

Output voltage (13.8V)

Rise time = 60 ms
Load current

Fuel cell current

Figure 13. Voltage and current waveforms under load step test from 20-A to 60-A condition.

Figure 14 shows voltage and current waveforms under load dump condition. The load
current is reduced from 85 A down to 25 A. Since the SOFC controller does not set the ramp
rate, the response under load dump is faster than the load step. As can be seen from Figure 13,
the current fall time is 50 ms for a 60-A load dump, as compared to 60 ms for a 40-A load step.
 Output voltage is well
regulated during
Load current heavy load transient

Output voltage (13.8V) Fall time = 50 ms

Fuel cell current

Figure 14. Voltage and current waveforms under load dump test from 85-A to 25-A condition.

3.5 TEST THE NEW CONVERTER WITH DELPHI SOFC

3.5.1 Efficiency Test

After the converter test with Delphi SOFC simulator, the converter was proven to have
the claimed efficiency and to be able to communicate with the Delphi SOFC controller. The
converter was then moved to the fuel cell test stand. Figure 15 shows photograph of the test
setup with VT DC-DC converter and the Delphi SOFC. The fuel cell output is monitored and
controlled by setting a set of fuel cell parameters. The load contains a 12-V battery pack and a
programmable electronic load. The entire test condition is the same as that under SOFC
simulator test, except that the source voltage and available current are dependent on the SOFC
condition. For every test point, the SOFC parameters need to be adjusted to match the output.
Three efficiency points were tested to compare with the results obtained from the SOFC
simulator test. Figure 16 shows the efficiency and fuel cell output voltage as a function of the
output power. The first point starts at about 600-W condition, and the efficiency is about 95%.
This is slightly better than the one obtained from fuel cell simulator test (94.5%) because its
voltage is lower. The second point runs at about 900-W condition, and the efficiency is about
97%, which agrees the test result obtained from the simulator test very well. The third point is
measured at about 1.3 kW, and the efficiency reaches 98.5%. This agrees with the results
obtained from Delphi fuel cell simulator, but is more than 1% higher than the one measured with
VT fuel cell simulator. It appears that the instrumentation tends to favor output power between 1
and 1.5 kW range. Nevertheless, the VT converter has demonstrated a superior efficiency under
both SOFC and SOFC simulator test conditions.

Delphi
VT SOFC
Converter

Figure 15. Photograph showing setup of VT DC-DC converter tested with Delphi SOFC.

100%
99%
98%
Efficiency

97%
96%
95% With VT fuel cell simulator
94%
93%
92% With Delphi SOFC

60 0 500 1000 1500 2000 2500

Fuel cell voltage (V)

50
40
30 With VT fuel cell simulator
20
10
0
0 500 1000 1500 2000 2500
Fuel Cell Output Power (W)

Figure 16. Efficiency evaluation results with Delphi SOFC.

3.5.2 Current Ramp Test

The fuel cell power availability depends on various factors. With temperature as the
dominant factor, the fuel cell output current ramp needs to be slowed down with proper control.
Although the previous fuel cell simulator tests indicated that the VT converter has a fast current
loop control, and the current ramp can be achieved within 10’s of milli-second, the actual fuel
cell output ramp needs to be controlled in 10’s of second. Figure 17 shows test result using
Delphi SOFC controller to obtain the current ramp control of 40A/40s rise rate.

Output current step from 20A to 60A

Output voltage with 40s ramp control

Fuel cell current (10.2A)

Figure 17. Voltage and current waveforms under current ramp control using Delphi SOFC controller.

3.5.3 Steady-State Test

Figure 18 shows steady-state voltage and current waveforms of the DC-DC converter
tested with Delphi SOFC under light load condition. The fuel cell voltage and current are 57.3 V
and 10.2 A, respectively. The output load has a battery voltage of 13.1 V and the total load
current of 43 A. The switch duty cycle is 25.4% in this case.
Figure 19 shows steady-state voltage and current waveforms of the DC-DC converter
tested with Delphi SOFC under medium load condition. The fuel cell voltage and current are
52.4 V and 24.3 A, respectively. The output voltage and current are 13.1 V and 95 A,
respectively. The duty cycle is increased to 36.4%.
 Gating control

Load current (43A)

Fuel cell current (10.2A)

Figure 18. Steady-state voltage and current waveforms of DC-DC converter tested with SOFC under light
load condition.

Gating control

Load current (95 A)

Fuel cell current (24 A)

Figure 19. Steady-state voltage and current waveforms of DC-DC converter tested with SOFC under medium
load condition.

3.5.4 Dynamic Test

With proper tuning of fuel cell parameters, the Delphi SOFC allows load step and load
dump tests with reasonable current rise or fall rate. Figure 20 shows voltage and current
waveforms under load step from 16 A to 28 A. The load and fuel cell currents rise smoothly
with a rate controlled by the SOFC controller. This test also proves that the communication
between the VT converter and Delphi SOFC is working properly.

Gating control

Load current

100ms/div Fuel cell current

Figure 20. Voltage and current waveforms under load step from 16-A to 28-A test with Delphi SOFC.

Figure 21 shows voltage and current waveforms under load dump from 95 A to 60 A.
The load dump process was performed continuously with several steps, and the scope time scale
was changed to 1 s/div to capture the current waveforms. Again, the fuel cell and load currents
smoothly reduce and follow the command precisely.
 Gating control

Load current

Fuel cell current

1 s/div

Figure 21. Voltage and current waveforms under load dump from 95-A to 60-A test with Delphi SOFC.

4. CONCLUSION
The entire test with VT DC-DC converter operating under Delphi SOFC controller was
very smooth and successful. The VT DC-DC converter communicates with the Delphi SOFC
controller very well with current control scale well calibrated. Major test items are summarized
as follows:
• Tested voltage range: 32 to 60 V
• Test current range: 0 to 203 A
• Voltage regulation from 0 to 3kW test condition is within 0.01V.
• Current regulation from 0 to 200A test condition is within 1A.
• Efficiency exceeds 97% at load higher than 1 kW and peaks at 97.5% at 3-kW full load
condition.
• Dynamic load step and load dump follow the command rate precisely and operate
smoothly.
Some of the test conditions are actually tougher than the original specification. For
example, the low-end tested voltage of 32 V is lower than the original specified 35 V. The
highest current tested was 203 A, which is higher than the specified 200 A, and actually caused
the damage on the mechanical contactor during the test. The DC-DC converter ran robustly even
after the failure of the mechanical contactor. The efficiency exceeds the SECA goal of 97%.
Multiple-phase interleaving operation design was proved to be reliable and ripple free at the
output, which is desirable for the battery charging. The numbers are consistent throughout all
the tests in VT-FEEC lab and Delphi test stand. Overall this is a very successful collaboration
project between the SECA Core Technology Team and Industrial Team.

5. BIBLIOGRAPHY

[1] C. Liu and J.S. Lai, “Low Frequency Current Ripple Reduction Technique
with Active Control in a Fuel Cell Power System with Inverter Load,” IEEE
Transactions on Power Electronics, Jul. 2007, pp.1429 – 1436.
[2] S.-R. Moon and J.-S. Lai, “Multiphase Isolated DC-DC Converter for Low-
Voltage High-Power Fuel Cell Applications,” in Proceedings of IEEE
APEC, Anaheim, CA, Feb. 2007, pp. 654 – 660.
[3] R.-Y. Kim and J-S. Lai, “A Seamless Mode Transfer Maximum Power Point
Tracking Controller for Thermoelectric Generator Applications,” IEEE
Trans. on Power Electronics, Sep. 2008, pp. 2310 – 2318.
[4] R. P. O’Hayre, S.W. Cha, W. G. Colella, and F.B. Prinz, Fuel Cell
Fundamentals, New Jersey: Wiley, 2009.
[5] M. A. Peavey, Fuel from Water, 10th ed., Louisville: Merit, Inc.
[6] R. McAlister, The Solar Hydrogen Civilization, 2nd ed., Mesa: American
Hydrogen Association, 2005.

6. PATENT ISSUED
Jih-Sheng Lai et. al., “Multiphase soft switched DC/DC converter and active control
technique for fuel cell ripple current elimination,” U.S. Patent #7,518,886, April 2009.

7. LIST OF ACRONYMS AND ABBREVIATIONS

AC – Alternate current
CB – Circuit breaker
DC – Direct current
DG – Distributed generation
DOE – Department of Energy
DSP – Digital signal processor
EPRI – Electric Power Research Institute
IEEE – Institute of Electrical and Electronics Engineering
kVAr – kilo Volt-Ampere reactive
kW – kilo watts
LC – inductor-capacitor
LCL – inductor-capacitor-inductor
PCS – Power conditioning system
PLL – Phase locked loop
QPR – Quasi-proportional-resonant
SECA – Solid-State Energy Conversion Alliance
SOFC – Solid oxide fuel cell
V – Volt
VAr – Volt-Ampere reactive
Virginia Tech – Virginia Polytechnic Institute and State University
W – Watt
APPENDIX: Key Schematic Circuit Diagrams

1 2 3 4

DC+
DC-

DC+

OutA A1H 11H 11H

OutA OutA A1H A1H 11Hs 11Hs

DC+
Dual Loop Controller (UC3824)
A Sync OutB A1L A
Sync OutB OutB A1L A1L 11L 11L
11Ls DC- 11Ls Jout1
SoftStart

Gate Drives (HIP4081)

FIFO Logic Controller
SoftStart 12H 12H Out11 1

Full Bridge Power Stage

+15 12Hs 12Hs Out12 Out12 2
AH A2H
AL AH A2H A2L A2H 12L 12L
AL A2L A2L 12Ls DC- 12Ls Header 2
21H 21H
21Hs 21Hs
21L 21L Jout2
A3H DC-
WRTEN A3H A3L A3H 21Ls 21Ls Out21 1
WRTEN A3L A3L 22H 22H Out22 Out22 2
RDEN1
RDEN1 22Hs 22Hs
22L 22L Header 2

Page: 3

Page: 4
+15 +15V DC-
22Ls 22Ls
OutEN Dis

Page: 5
GND OutEN Dis 31H 31H
RESET Jout3
Page: 2

GND RESET 31Hs 31Hs

RESET_H RESET_H 31L 31L Out31 Out31 1
+15_A DC- Out32
V_Sense

+15_1 31Ls 31Ls Out32 2

I_Sense

+15_B +15_2 32H 32H

+5V +15_C Header 2

DC-
VCC +15_3 32Hs 32Hs
GND GND
GND
GND 32L 32L
B DC- B
32Ls 32Ls

V_Sense ILout DC-

IL2

+5V DC+_filtered VCC_input +15_A +15_A DC-

VOUT

IL2
ILout

DC-_filtered GND_input +15_B +15_B

+15_C +15_C
R12
Sensor Conditioning =Value +15 +15
Page: 6 T_H -15 -15
Power Supply
+15 Page: 7 +5V +5V
Vpp R11 Vin Vout
GND GND TOUT DC+
-15 =Value Q1 3.3V 3.3V

GND
Vnn
=MFG_PN
GND 5V PID
GND Page: 8
Dt2 Dt1
Tranzorb Tranzorb
C inverter_aux GND C
DC+_filtered 1
2
DC-_filtered 3
DC-
CON3_N DC+_filtered
DC+ U110 CHOKE C111
C110 In Out =(Value+Voltage) C113 C114
DC- GND_in GND =(Value+Voltage)
=(Value+Voltage)
DC+_filtered shutdown C112
Vfc aux power control =(Value+Voltage)
DC-_filtered GND1 =(Value+Voltage) DC-_filtered

disable control
Page: 9 EARTH_GND
+5V 5V OutEN OutEN
GND2 DIS Dis
SoftStart Socket
Softstart
T_H T_H
D RESET_H RESET_H D

Title: =Title Cann

Virginia Tech
ot
Size: Letter Dwg: =DocumentNumber Future Energy Electronics Center
Rev: =Revision
Drawn By: =DrawnBy Date: =Date =SheetNumber
of =SheetTotal Director: Dr. Jih-Sheng (Jason) Lai
www.feec.ece.vt.edu (540) 231-4741
1 2 3 4
 1 2 3 4

GND
GND

DC+ DC+
Roff1a Dg1a Dg1c Roff1c
A =(Value+Tolerance+Power) A
1N4148 1N4148 =(Value+Tolerance+Power)
11H Ron1a =(Value+Tolerance+Power) S11 S12 Ron1c 12H 11H
1A 1B 1C 11H
=(Value+Tolerance+Power) 11Hs 11Hs
11Hs 2 1 1A 1C 1 2 12Hs 11L
Rg1a =(Value+Tolerance+Power) =(Value+Tolerance+Power) DC- 11L
11H Q511a 11L Q511b 12H Q512a Zg1a Zg1c Rg1c 11Ls
1As Out11 Out12 1Cs 12H
Cdc1 Cdc10 Dg1b Dg1d 12H
1D Roff1b OUT11 OUT12 Roff1d 12Hs 12Hs
Film 1B 12L
1As GND 1Cs =(Value+Tolerance+Power) 12L
=(Value+Voltage) =(Value+Voltage) 1N4148 1N4148 =(Value+Tolerance+Power) DC-
12Ls
12L Q512b 11L Ron1b =(Value+Tolerance+Power) Ron1d 12L
FMM150-0075P FMM150-0075P
=(Value+Tolerance+Power)
1D 2 1
GND 11Ls Rg1b =(Value+Tolerance+Power) Rg1d =(Value+Tolerance+Power)
12Ls
Zg1b 1Ds Zg1d
DC- DC-
1Bs

DC+
Roff2a Dg2a Dg2c Roff2c
B =(Value+Tolerance+Power)
1N4148 1N4148 =(Value+Tolerance+Power) B
2A 2B 2C
21H Ron2a =(Value+Tolerance+Power) S21 S22 Ron2c 22H 21H 21H
=(Value+Tolerance+Power) 21Hs 21Hs
21H Q521a 21L Q521b 22H Q522a 21Hs 2A 2C 1 2 22Hs 21L
Rg2a =(Value+Tolerance+Power) =(Value+Tolerance+Power) DC- 21L
Zg2a Zg2c Rg2c 21Ls
2D 2As Out21 Out22 2Cs 22H
Cdc2 Cdc20 2As GND 2Cs Dg2b Dg2d 22H
Roff2b OUT21 OUT22 Roff2d 22Hs 22Hs
=(Value+Voltage)
=(Value+Voltage) 2B 22L
Film 22L Q522b =(Value+Tolerance+Power)
1N4148 1N4148 =(Value+Tolerance+Power) DC- 22L
22Ls
21L Ron2b =(Value+Tolerance+Power) Ron2d 22L
FMM150-0075P FMM150-0075P
GND =(Value+Tolerance+Power)
2D 2 1
21Ls Rg2b =(Value+Tolerance+Power) Rg2d =(Value+Tolerance+Power)
22Ls
Zg2b Zg2d
DC-
2Bs 2Ds

DC+
Roff3a Dg3a Dg3c Roff3c
C =(Value+Tolerance+Power)
1N4148 1N4148 =(Value+Tolerance+Power) C
3A 3B 3C
31H Ron3a =(Value+Tolerance+Power) S31 S32 Ron3c 32H 31H 31H
=(Value+Tolerance+Power) 31Hs 31Hs
31H Q531a 31L Q531b 32H Q532a 31Hs 3A 3C 1 2 32Hs 31L
Rg3a =(Value+Tolerance+Power) =(Value+Tolerance+Power) DC- 31L
Zg3a Zg3c Rg3c 31Ls
3D 3As Out31 Out32 3Cs 32H
Cdc30 3As GND 3Cs Dg3b Dg3d 32H
Roff3b OUT31 OUT32 Roff3d 32Hs 32Hs
=(Value+Voltage) 3B 32L
Film 32L Q532b =(Value+Tolerance+Power)
1N4148 1N4148 =(Value+Tolerance+Power) DC- 32L
32Ls
31L Ron3b =(Value+Tolerance+Power) Ron3d 32L
FMM150-0075P FMM150-0075P
GND =(Value+Tolerance+Power)
3D 2 1
31Ls Rg3b =(Value+Tolerance+Power) Rg3d =(Value+Tolerance+Power)
32Ls
Zg3b Zg3d
DC-
3Bs 3Ds

D DC+ D
C500 C501 C502 C503 C504 C505
DC- =(Value+Voltage)
Title: =Title Cann
Virginia Tech
ot
=(Value+Voltage) =(Value+Voltage)
=(Value+Voltage) =(Value+Voltage)
=(Value+Voltage) Size: Letter Dwg: =DocumentNumber Future Energy Electronics Center
Rev: =Revision
Drawn By: =DrawnBy Date: =Date =SheetNumber
of =SheetTotal Director: Dr. Jih-Sheng (Jason) Lai
www.feec.ece.vt.edu (540) 231-4741
1 2 3 4
 1 2 3 4

A A

Cbs11
=(Value+Voltage) Cbs31
U41 Dbs11 =(Value+Voltage) Dbs31
1 20 12H =MFG_PN U43
BHB BHO 12Hs 11H 1 20 32H =MFG_PN
A1H 2 19 11Hs 11H BHB BHO 31H
Dis BHI BHS 12L 11Hs A3H 2 19 32Hs 31H
3 DIS 18 11L BHI BHS 32L 31Hs
4 BLO 11L Dis 3 18 31Hs
Vss BLS 12Ls17 GND DC- DC- 4 DIS BLO
17
32Ls GND 31L
A1L 5 16 +15_1
L41 11Ls 11Ls Vss BLS DC- 31L
A1L 6
BLI Vdd
15
12H
12H A3L 5 BLI Vdd 16 +15_3
L45 31Ls DC- 31Ls
ALI Vcc +15_11 12Hs A3L 6 15 32H
A1H 7 11Ls14 GND Dvcc1 12Hs ALI Vcc +15_31 32H
AHI ALS DC- 12L A3H 7 Dvcc3 32Hs
8 HDEL ALO 13 11L L42 =MFG_PN
12Ls 12L AHI ALS 31Ls14 GND DC-
31L L46 =MFG_PN 32L 32Hs
9 12 11Hs DC- 12Ls 8 HDEL ALO 13 32L
LDEL AHS 9 12 31Hs 32Ls DC-
10 11 11H LDEL AHS 32Ls
AHB AHO Dbs12 10 11 31H
RHd1 =Value

RLd1 =Value

Rdis3 =Value
Dbs32

RHd3 =Value
RLd3 =Value
AHB AHO
Rdis1=Value

Cbs12 =MFG_PN
HIP4081A =(Value+Voltage) Cbs32 =MFG_PN
HIP4081A =(Value+Voltage)
B B

A1H A1L A2H A2L A3H A3L

1
1H 1L 2H 2L 3H 3L
+15_1 +15_11
C410 C411 C412
=(Value+Voltage)
=(Value+Voltage) =(Value+Voltage)

A1H A1H
Cbs21 A1L A1L
=(Value+Voltage) Dbs21 +15_2
U42 +15_21
1 BHB 20 22H =MFG_PN 21H A2H
C 2 BHO A2H C420 C421 C422 C
A2H BHI BHS 19 22Hs 21H A2L
3 18 22L 21Hs A2L =(Value+Voltage)
=(Value+Voltage) =(Value+Voltage)
Dis DIS BLO 21L 21Hs
4 Vss 22Ls17 GND DC- 21L
5 BLS 21Ls DC-
A2L BLI Vdd 16 +15_2
L43 21Ls A3H
A2L 6 15 22H A3H
ALI Vcc +15_21 22Hs 22H A3L
A2H 7 A3L
8
AHI ALS 21Ls14 GND DC-Dvcc2 22L 22Hs
+15_3 +15_31
HDEL ALO 13 21L L44 =MFG_PN 22Ls DC- 22L
9 LDEL 12 21Hs 22Ls Dis
10 AHS Dis C430 C431 C432
11 21H
Rdis2 =Value

AHB AHO Dbs22

RHd2 =Value
RLd2 =Value

=(Value+Voltage)
=(Value+Voltage) =(Value+Voltage)
Cbs22 =MFG_PN
HIP4081A =(Value+Voltage) +15_3 +15_2 +15_1 +15_1
+15_1 +15_2
+15_2
1

+15_3 +15_3
GND

15V-3 15V-2 15V-1

DC- DC-
1

D D

GND-3 GND-2 GND-1 Title: =Title Cann

Virginia Tech
ot
Size: Letter Dwg: =DocumentNumber Future Energy Electronics Center
Rev: =Revision
Drawn By: =DrawnBy Date: =Date of =SheetTotal
=SheetNumber Director: Dr. Jih-Sheng (Jason) Lai
www.feec.ece.vt.edu (540) 231-4741
1 2 3 4