Beruflich Dokumente
Kultur Dokumente
APPENDICES:
A. SCHEMATICS................................................................................. 23
B. SIMULATION RESULTS.............................................................. 31
C. UC3825B DATASHEET ................................................................. 34
D. COST INFORMATION ON TEXAS INSTRUMENTS
TMS320C24X DSP .......................................................................... 39
LIST OF TABLES:
1. Bill Of Materials For DC-DC Converter And Bulk Capacitors. 17
2. Bill Of Materials For DC-AC Inverter And Output Filter ......... 18
3. Bill Of Materials For DSP Control Board .................................... 19
4. DC-DC Converter Subsystem Costs .............................................. 21
5. DC-AC Inverter Subsystem Costs ................................................. 22
Page 2 of 40
1.0 Introduction
The report outlines the technical approach and the cost analysis to achieve the objectives
proposed by the 2001 Future Energy Challenge organizing committee. The Texas A&M team
believes it has developed an efficient and cost-effective inverter system. The team has developed
a low cost analog control solution for the DC-DC converter, an efficient 3-terminal DC-DC push
pull topology, a unique DSP control for DC-AC inverter control, and a rigorous cost reduction
The Texas A&M team provides a rigorous cost savings approach by reducing the number
of power switches in the design. Incorporating fewer power resistors enhances cost savings and
efficiency. The 3-terminal push-pull DC-DC converter topology provides isolation for safety,
suitable boosting of the fuel cell voltage to 400 volts, reduced cost and reduced size of the energy
a DC-DC boost circuit, a DC-AC inverter circuit and an output filter besides battery banks
The DC input from the fuel cell (48 VDC nominal, +50%, -12.5%) is first converted to a
regulated 400 VDC using a high frequency 3-terminal Push-Pull DC-DC converter. The DC-DC
conversion stage consists of a high-frequency transformer. Isolation is provided for safety, system
protection, and to meet the stringent FCC Class-A standards. The 400V DC-DC converter output
stage. To obtain independent single phase outputs, two half-bridge inverters are used. An output
LC filter stage is employed to produce a low THD AC waveform. Low loss, high switching
Page 3 of 40
frequency MOSFET and IGBT switches have been employed to achieve a higher efficiency,
manufacturability and mass production. Another unique aspect of the design is the use of the
TMS320C2407 DSP to control the inverter. The DSP reduces printed circuit board layout
complexity. Readily programmable, the DSP adds flexibility and intelligence to implement
various control aspects by means of software. (See Appendix D for DSP cost information)
Two sets of lead-acid batteries are provided on the 200V DC bus to supply sudden load
demands. By floating the standby battery off the 400V instead of at the 48V level, we avoid
processing the battery power via two stages. Efficient and smooth control of the power drawn
from the fuel cell and the high voltage battery is achieved by controlling the front end DC-DC
Page 4 of 40
TR1
i i L1 I
1:5 D1 L DC
+ i
A
T1 i
C
Fuel Cell Input K1
i C2 L S1 L2
48VDC I in T1 D1 D3 b
S3
N2 A
A C4 3
V V N
in N1 batt i
AO
+ AC Output
C1
V N 120/240V , 60 Hz
- DC
N1 L B C5 B
C3 b i
BO 3
N2 N
V i
D4 D2 batt L3 B
I K2
T2
- S2 S4
T2
Page 5 of 40
Battery Backup
48VDC / 400VDC, 40KHz PUSH PULL CONVERTER 120V/240VAC, 20KHz PWM INVERTER
Note: Components shown in dotted boxes are not considered for cost evaluation
variable limiter -
DC-DC Converter
Current
Feedback
Voltage
Feedback
Figure 2 shows the block diagram for the current control of the DC-DC converter. The
Power Available signal (analog) from the fuel cell is used to adjust the current limit setting of the
DC-DC converter. This ensures that the power drawn from the fuel cell does not exceed its
capability. The remaining power is then provided by the battery backup system (Figure 1). The
inverter, on the other hand, determines the actual power drawn by the loads and communicates to
the fuel cell to either increase or decrease its power output. This ensures that the fuel cell has
sufficient time to adjust its power generation to meet the changes in load demand.
In this section design of the DC-DC Converter is detailed. Figure 1 shows the circuit
diagram of the push-pull DC/DC converter. Fuel cell output is connected to the DC/DC
converter as shown. MOSFETs T+ and T- are turned on and off alternately at a switching
frequency of 40kHz.
The power output Po of the inverter is 10000W. Assuming an efficiency of 95% for the
Page 6 of 40
10000W
Pin 11050W (1)
0.95 0.95
Designing for the low input line condition (Vin=42VDC), input current Iin from the fuel cell is,
11050W
I in 263 A (2)
42V
The push pull DC/DC converter shown in Figure 1 comprises of two switches, T+ and T-. At the
maximum duty ratio of 0.45, rms current rating IT of the switches are,
IRFP260N (200V, 50A) MOSFETs with 4 devices in parallel in each leg are then chosen.
For obtaining an output voltage of 400VDC for the push-pull converter, a turns ratio of K=5 is
selected for the transformer. Center taps are available on both the primary and secondary sides as
shown in Figure 1.
The VA rating of the transformer is defined as the sum of the total primary and secondary
1
Vin I in I
VATr 2 Vin K in 2 1.5Vin I in 1.5 42 263 16600W 17.0kVA
2 2 2 K
(4)
Diode ratings:
The reverse blocking voltage is equal to the DC link voltage 400V. Since each diode is clamped
to the mid-point of the DC-link (200V), each diode can be rated for 300V.
Page 7 of 40
I in
ID 37.2 A (5)
K 2
The DC-DC Converter uses the 3-terminal push-pull topology to boost the 48V from the fuel cell
means of a high speed PWM controller UC3825B (datasheet attached in Appendix C). The special
features of this controller are: suitability for current control; soft start; over current and under
voltage protection; low propagation delay; high current dual outputs and low cost.
Current mode control has numerous advantages over simple voltage mode control,
including making the converter respond faster to load changes. In particular the UC3825B is
suitable for the fuel cell inverter application because it allows direct control over the power drawn
from the fuel cell. The error amplifier output in the outer voltage loop defines the level at which
the primary current (in the inner current loop) will regulate the pulse width and output voltage.
Pulse-by-pulse symmetry correction is a feature of current mode control and thus is essential for
Timing section:
From the UC3825B data sheet, for a maximum duty cycle of 0.9, we have
3V 3
RT 2 3k
10mA(1 DMAX ) 10 (1 0.9)
1.6 DMAX 1.6(0.9)
CT 12nF
RT F 3(103 )(40)(103 ) (6)
Page 8 of 40
Pin 11050W
The primary current under minimum fuel cell input voltage (42V) conditions, Iin is
I in 263 A
263
I in ,rms 0.9 278 A (7)
0.9
Current sensing:
To obtain 1.0V at 400A, current sensing resistor Rs = 0.0025 is used. We shall use 4 power
resistors rated 0.01 , 75W in parallel (See DC-DC converter schematic in Appendix A).
Accounting for voltage drops on the secondary side, the transformer secondary voltage is 410V.
Hence a transformation ratio of 1:10 is selected. This would result in a transformer turns ratio of
10500
Io 26 A (8)
400
dt 11.25s
L Vsec 410V 576H 600 H (9)
dI 8A
Slope compensation is required to compensate for the peak to average differences in primary
current as a function of the pulse width. The downslope of the inductor current is,
dI 8A
0.71 A s (10)
dt 11.25s
This value when reflected to the primary side (multiplying by the transformation ratio) yields
0.71 10 7.1 A s
Page 9 of 40
Equivalent ramp downslope voltage VSL available across the sense resistor is,
1.8V
VOSC 0.08V s (12)
22.5s
If the amount of inductor downslope voltage to be added to the oscillator waveform is 75%, then
Input Capacitor :
Selecting a proper input capacitor C1 (Figure 1) contributes to the reduction in fuel cell input
Assuming a square wave input current, for a duty ratio of 0.9, the peak current I,
263
I 292 A (13)
0.9
I c ,rms I rms
2
I avg
2
92 A (15)
Based on the rated ripple current, 4 Rubycon Aluminum electrolytic capacitors 22000F, 100V
The simulation results for a 10kW load on the system are presented in Appendix B. Vds1,
Vds2 are the drain to source voltage across the MOSFETs T1 and T2 respectively. VDC is the
output voltage.
Page 10 of 40
3.2 Inverter Design Procedure for the 10kW TAMU Fuel Cell Inverter System
The schematic of the DC-AC Inverter circuit is shown in Figure 1. The inverter produces
two single-phase outputs, Phase-A and Phase-B. It is comprised of two half bridge inverters each
supplying a separate single-phase load at 120VAC, 60Hz. Consider the case when Phase-B is not
loaded and Phase-A is supplying full load (5000VA). The peak amplitude of the fundamental
frequency component is the product of ma and ½VDC, where ma is the modulation index. A
The fundamental component of the inverter Phase-A output voltage VAO is,
VDC
V AO ,1 ma sin( 1t ) 0 ma 1 (16)
2
0.9
sw1 0.5 sin1t higher frequency terms (17)
2
The Phase-A output current (iAO) is assumed to contain fundamental and third harmonic
current components due to presence of nonlinear load. The current iAO can be expressed as,
isA sw1 i AO
2 3
I 1 sin(1t 1 ) I 3 sin(31t 3 ) ... (19)
2 2
0.9 0.9
2 I 1 cos 1 cos(21t 1 3I 3 cos 3 cos(31t 3 ) ...
2 2
Assuming the load current iA to consist of only fundamental (I1) and third harmonic component
(I3), we have,
I A,rms I 12 I 32 (20)
Page 11 of 40
Further, assuming I3=0.7 I1 (which is typical of a single phase rectifier type nonlinear load) we
have,
I A,rms 1.22 I1
Since
5000
I A, rms 41.7 A (21)
120
41.7
I1 34 A (22)
1.22
Therefore, the largest component of the DC-link capacitor current ic is the fundamental frequency
1
ic ,rms I 1 17 A (23)
2
ic ,rms
Vc (24)
C
ic ,rms 17
C 4500 F (25)
Vc 10 2 60
Panasonic Electrolytic capacitors rated 100V, 4500F are selected for this design.
The rms current isA is 41.7A. Thus, rms current rating IT of each switch is
41.7
IT 30 A (26)
2
Page 12 of 40
3.3 DC-AC Inverter Output Filter Design Procedure
Figure 3 shows the topology for the output L-C filter. A transfer function is developed
from the schematic. The assumptions used in the analysis are, the output filter is lossless and the
jnX L
Vi,n -jXC
Vo,n ZL1n
n
The transfer function for this type of filter is described by the equation
Vo , n jX C Z L , n
Hn . (27)
Vi , n nX L X C jZ L , n (n 2 X L X C )
Where
Hn - transfer function
Z L,n - impedance
n - harmonic order
For H 1 1 ; or X L X C , then
jX C Z L ,1
H1 1. (28)
jZ L ,1 X C
Page 13 of 40
At no load, Z L ,1 , therefore equation (27) is
XC 1
Hn (29)
n XL XC
2
X
n2 L 1
XC
1 X 34.333 (30)
0.03 L
X X n2
n2 L 1 C
XC
Non-Linear Load
An equivalent circuit used in finding filter characteristics for a non-linear load is shown in
Figure 4.
jhXL
-jXC
Vh Ih
h
jhX L X C
Vh Ih . (31)
X C h2 X L
Where
Vh - equivalent voltage
h - harmonic order
Ih - current at h harmonic
Page 14 of 40
hX L
Vh Ih . (32)
2 XL
1 h
XC
XL X
Here is very small making h 2 L 1 , therefore
XC XC
Vh hX L I h (33)
V3 3X L I 3 V3
, where THD is 0.03 or 3% . Inductor impedance can be found by
V1 V1 V1
0.03 V1
XL (34)
3* I 3
fr XC n2
56.89 . (35)
f1 XL 34.333
f r 3413 Hz
XL
L (36)
2 f1
Where
L - inductance
f1 - fundamental frequency
Page 15 of 40
where f 1 60Hz , the inductance will be L 123H .
To find the capacitor impedance use the equation (30), to get X C 148.9 , then using
1
C (37)
2 f1 X C
where
C - capacitance
XC - capacitor component of impedance
f1 -fundamental frequency
4.0 Schematics
The following detailed schematics are attached in Appendix A.
A4. Inverter voltage and current sensing and protection circuitry (Sheet 1)
A5. Inverter voltage and current sensing and protection circuitry (Sheet 2)
Page 16 of 40
5.0 Bill of Materials
In this section, a detailed bill of materials is developed for the DC-DC converter and DC-
AC inverter subsystems. The components in the bill of materials are shown in schematics in
Appendix A.
Table 1: Bill of Materials for DC/DC Converter, Bulk Capacitors and its associated control
& protection circuitry (refer Figures A1-A2 in Appendix A)
Page 17 of 40
Table 2: Bill of Materials for DC/AC Inverter, Output Filter and its associated control &
protection circuitry (refer Figures A3-A5 in Appendix A)
Page 18 of 40
Table 3: Bill of Materials for DSP Control Board
(refer DSP Schematics in Appendix A)
and the faculty advisors, the team was able to make well-informed design decisions to
aggressively lower the cost of the final 10kW design and 1.5kW prototype. The TAMU fuel cell
inverter team’s approach to reducing the cost of the inverter by reducing the number of high cost
switching devices by adopting push-pull topology, using a low cost PWM DC-DC controller and
By use of the push–pull topology the number of MOSFETs was minimized to half that
needed by a full bridge topology. IGBT’s were reduced in the inverter by use of the half bridge
topology as opposed to the full bridge topology. The analog PWM controller provided a low cost
solution to control of the DC-DC converter. It provides a single chip control solution opposed to
complex discrete analog hardware. DSP control of the DC-AC inverter provides sophisticated
control at low cost. Further, the DSP enables software control of the inverter and adaptability for
stand-alone and utility interface modes. Software control translates into efficiency in human
capital reducing costs of analysis, troubleshooting, development and manufacturing of the fuel
Page 19 of 40
cell inverter. The use of the DSP allows a seamless interface with other components of a power
management system, saving integration time and human resources. The topology of the TAMU
Fuel cell Inverter System employs a high voltage battery floating on the DC-link. This approach
does not add any additional power processing cost for load management.
The cost for the power components of the TAMU Fuel Cell Inverter system were
calculated by developing the cost of the DC-DC converter and the DC-AC inverter and adding
the two components together. The cost analysis was based on the schematic shown in Figure 1
and the 10kW design procedure detailed in this report. The results of the cost analysis for the
DC-DC converter are seen on the normalized spreadsheet Table 4 and the results of the DC-AC
As per the cost analysis spreadsheet provided by the 2001 Future Energy Challenge
Committee, the cost of the DC-DC converter was $598.09. The cost of the DC-AC inverter
$198.69. The total cost of the TAMU Fuel Cell System was $796.78. It should be noted that the
cost analysis spread sheet (Tables 4 & 5) do not give the absolute cost and assumes a fixed cost
for control and packaging. These costs are highly dependent on the type of design and the
The TAMU inverter control is based on a low cost DSP (TMS320C24X). Our design and
implemented on this DSP platform. Appendix D details a press release from Texas Instruments
and lists a cost of $2.98 for the TMS320C24X DSP employed in the TAMU inverter design.
The TAMU Fuel cell Inverter Team believes that with a detailed analysis of the control
circuit and the ancillary components, this design can be mass produced and marketed for an
Page 20 of 40
2001 FUTURE ENERGY CHALLENGE
Page 21 of 40
2001 FUTURE ENERGY CHALLENGE
OTHER (EXPLAIN)
TOTAL 198.69
Table 5: DC-AC Inverter Subsystem Costs
7.0 Conclusions
This report has discussed the design methodology and cost analysis for the 10kW Texas
A&M Fuel Cell Inverter System. The topology and control strategy for this design has been
adopted keeping in mind the specific objectives of the 2001 Future Energy Challenge Committee.
Keeping the cost of the product low and obtaining the best performance for the given cost have
been the most important objectives that were pursued throughout this design procedure.
However, we believe that with sophisticated manufacturing techniques available today in the
Page 22 of 40
Appendix A
SCHEMATICS
Page 23 of 40
A B C D E
CN2
OH2 1 YELLOW
2 NC
3 NC
P15 4 NC
+BUS V 5 RED
N15
BUS FB P15A 6
4 7 GREEN 4
10k 8
30k 100uF YELLOW
P15 9
18k -BUS V 10
1 16 11 RED
2 INV VREF 15
42k 12 WHITE
3 NI VCC 14 0.1uF
120pF EA OUT OUTB SD 13 GREEN
4 13 DC VREF 14 NC
3k RT 5 CLOCK VC 12 0.1uF
RT PGND BUS FB 15 CYAN
R 6 11
7 CT OUTA 10
RAMP GND CN2
4.8nF 8 9
CT SS ILIM/SD
THD1
UC3825B
10k 1 2
OH2
SD
3 OHD5R13-90M 3
30k
30k
L6A
300uH
25A + BUS V
D1 D4 D6 R8
40A 40A 15A 56k
3
3
300V 300V 1200V C3
Q1 Q2 Q3 7W
4500uF
J1 54A 54A 54A 250V
1 C20
1 2 200V 1 200V 1 200V 1 Q4
54A 0.1uF C4
JUMPER R1 R2 R3 R4 1200V 4500uF
PC1 200V
P15A D2 D3 R9 56k 250V
10 10 10 10
2
2
1 8 40A 40A
2 7 300V 300V 7W
2 3 6 2
4 5 C22 L6A
0.1uFR12 300uH
- BUS V
50V 47K 1 T2
R22 HCPL-3120 25A
47K CN1 C5 C6
J3 C3 5
1 2 10uF 300uF 4700uF
1
100Vdc 100Vdc 100V 2 6
JUMPER 2
4 x 0.01ohm, 3 4 8
C7 C8 C4
75W each 4 4700uF
10uF 300uF 7
100Vdc 100Vdc 100V
J2 3
1 2 C10
3
3
JUMPER Q7 R10 150pF
Q5 Q6 Q8 500
54A 1000V
PC2 P15A 54A 54A 54A 10W
200V 200V 200V
1 8 1 1 1 1 200V
1 2 7 1
3 6
R5 R6 R7 R8
4 5 C21 Title
10 10 10 10
2
+
-15V
R14 6.8k
R1 U1C R7 10K
75k,1W POT1 2k
11
LF347 R13
POT1 2k 33k
5 + C1
9 -
8 5
R2 10 + U1B
470pF
11
1.2k LF347 -15V
R5 10k 6 - R15 1k
4
7 U2
R3 5 +
LF356
4
5
1.2k C2 +15V R6 10k
+
470pF R26 R16 2 - Power IN
2.2k 6
4
10k CN1
3 +
R4 Voltage Feedback 48G 1
75k, 1W U1D 2
11
7
1
LF347 +15V 3
13 - 48V 4
-200V 14 +15V
POT1 4 PIN HEADER
-Vbw 12 + +15V
10k
Outputs
CN2
4 4
4
D2
R9 470k LED +15V 1
CN5 G 2
FB 3
+15V 1 C4 SD
D1 4
G15 2 0.1uF OT
1N753 U1A 5
ON 3
6.2V R12 6
OV 4 LF347
4
+15V
OT 5 R10 10k 1.2k 6 PIN HEADER
3
+200V 6 Thermal Protection +
1
N 7
-200V D5 2 LCD
8 - Overvoltage Protection
R11 10k CN3
LED
8 HEADER 48V 1
11
+15V 48G 2
+200V 9V 3
R22 48G 4
R21
1.2k 4 PIN HEADER
1.5k ***Note: Reset pushbutton is
R17
10k,2W on the faceplate.
3 3
Switches
+15V CN4
9V
D7
1N747A
3.6V + C4
0.1uF
D8
1N753A
1 6.2V
GND
Title
1
FIG. A2: DC-DC CONVERTER VOLTAGE FEEDBACK AND PROTECTION SCHEMATIC
A B C D E
A B C D E
1
10 6
1
G+ C1 C2
HIN VB 1uF 1uF
11 5 50V R13
SHUTDOWN SD Vs 50V FR104
10k
2
12 3 OUTPUT TO FILTER
G- LIN VCC
2
13 2 FR104
VSS COM
1
LO CR6
FR104
IR2110
Q2
1
C14 C7 R4
1 2 IGBT
10uF 0.1uF
10 600V,35A
50V 50V
1
2
R10
3 FR104 10k 3
E9 RAILNEG
J1
+15V 1 2
3 4
G+ 5 6
G- 7 8
SHUTDOWN 9 10
11 12
drnI 13 14 RET
1 1
Title
FIG. A3: INVERTER POWER CIRCUIT AND GATE CONTROL
R16
680
4
OPAMP1
3
+
5 2
-
1
+3.3V 5
D0
+5VDSP Optocouplers
11
R19
ISO1 +15V2
-15V 390
8
2 7
R20
U8A C7 1k
Voltage Sensor 1 R5 10k 1 2 3 6
0.1uF
35V
PWM2
+15V 3.3V
5 Gate Driver 1
74LS14
POT1 50k D1 6N137
ISOAMP1
4
R4 10k OPAMP1
38 19 5
FB HI +
7
R1 R2 VSENSE1 +5VDSP
3 6
4 470k 470k 1 -IN
+IN LO
18 R3 10k -
D2 4
Vout1 R21
2 15V 50k POT2
ISO2 +15V2
11
IN COM 20 3.3V 390
+15DC 8
36 2 7
37 +VISO 22
-VISO PWR RET C1 -15V R22
C2 22p U8B
0.1uF C8 1k
35V
AD202JN 35V 0.1uF
3 4 3 6 35V
PWM1
5 Gate Driver 2
POT3 50k D3
ISOAMP2 +5VDSP
4
R8 10k OPAMP1
38 19 10
FB HI +
8
R6 R7 VSENSE2 R23
3 3
1 -IN 18 R9 10k
9
-
D4 390
ISO3 +15V2 3
470k 470k +IN LO 8
Vout2
2 15V 50k POT4 2 7
11
IN COM 20 3.3V R24
+15DC U8C C9
36 1k
37 +VISO 22 0.1uF
-VISO PWR RET C3 -15V 5 6 3 6
C4 22p 35V
PWM4
0.1uF 35V
AD202JN 35V 5 Gate Driver 3
74LS14
6N137
+15V
T1 +15V -15V 3.3V
R25
1 ISO4 +15V2
+ 390 8
4
2 3 OPAMP2
R13 10k R26
M +
1 6 U8D C10 1k
2 2
-
-15V R11 C5 2 7 0.1uF
3 680p
-
5 ISense1
82 + 9 8 3 6 35V
- 35V D6 PWM3
50k
3.3V 5 Gate Driver 4
11
LA 55-P
4
POT5 74LS14
6N137
-15V
+15V
+15V
T1 +15V -15V 3.3V
1
+
4
2 10 OPAMP2
M + R17 10k
8 13
1 1
-
-15V R15 C6 9 14
3 680p
-
12 ISense2 Title
- 82 +
35V D8 FIG. A4: INVERTER VOLTAGE & CURRENT SENSING AND PROTECTION CIRCUITRY (1 OF 2)
POT6
3.3V
11
LA 55-P
Size Document Number Rev
4
50k
Custom 2.02
-15V
+15V
Date: Friday, August 31, 2001 Sheet 4 of 5
A B C D E
A B C D E
J1
JP3
+15V 1 2
34 33 ISENSE2 3 4
32 31 ISENSE1 Gate Driver 1 5 6
30 29 Gate Driver 2 7 8
28 27 SHUTDOWN 9 10
26 25 VSENSE2 Temperature1 11 12
24 23 +15V
13 14
5 22
20
21
19
VSENSE1
Inverter A Header 5
18 17
16 15 POT
14 13 50k
12 11
10 9 ***There is only one SHUTDOWN
8 7 PWM4 OPAMP6D signal shared between both
6 5 PWM3 R37 inverter boards.
11
PWM2 10k LF347
4 3
PWM1 13 U?A
2 1 -
14 1 U?C
R36 NOR1
DSP Header 12 + 3 8 +15V
2 10 1 14
150 ohm 9 2 13
J2
3 12 C17
4
4071
+15V 1 2 Current sensor1 4 11 104
4071
3 4 5 10
Current sensor2 6 9
Gate Driver 3 5 6
Gate Driver 4 Current sensor return 47 ohm 7 8
7 8 D13 +15V
SHUTDOWN 9 10 1N4148
4 Temperature2 11
13
12
14
MC14001BCP
4
R43 +15V JP?
Inverter B Header 680
1
2
3
4
D9 +15V
C14 R44 HEADER 4
1N4148
221 47k
ISense1
D10 POT
1N4148 R27 10k R28 10k R33 R34 50k
***Pins 3 and 4 connect to an external NOPB switch on the
10k 10k
+15V Inverter Box interface. Pins 1 and 2 connect to an LED on
R37 OPAMP6C the Inverter Box interface for shutdown notification.
11
OPAMP6B
10k LF347
11
OPAMP6A LF347
R29 10k 9
11
D13
-
LF347 C13
1N4148 6 - R35 10k 8
152 R36
3 C11
152
2 -
1
R31
5 +
7
10k
10 +
D15 3
3 +
C15 1N4148
10k
D14
4
R32 221
R30 1N4148 C12
4
10k +15V
4
10k 221
U?B
6N137
5
4
6
R19 390
4071
Fuel cell input
D?
ISO1 +15V
DIODE ZENER 8
2 7
R20
1 C7
0.1uF
10k
Title
1
3 6 35V FIG. A5: INVERTER PROTECTION CIRCUITRY (2 OF 2)
5
Size Document Number Rev
6N137 A4 {Doc} 0
A B C D E
1 2 3 4
11
DSP 16 ohm C4
TLV5619 / 296-1925-5 U2 0.1uF
80 127 19
A0 D0 D0
VDD
78 130 20 C3
A1 D1 D1
74 132 1 470pF
A2 D2 D2
71 134 2
A3 D3 D3
Emulation and test / JTAG 68 136 3 12
A4 D4 D4 VREF
trst 1 64 138 4 13 DACout
TRST* A5 D5 D5 AnalogOut
tck 135 61 143 5 14
TCK A6 D6 D6 GND
tms 144 57 5 6
TMS A7 D7 D7
36 53 9 7 L3
TMS2 A8 D8 D8
tdi 139 51 13 8 Fbead
LDAC*
TD1 A9 D9 D9
tdo 142 48 15 9
WE*
TD0 A10 D10 D10
PD*
CS*
emu0 90 45 17 10
EMU0 A11 D11 D11
emu1 91 43 20
EMU1 A12 D12
39 22
A13 D13
DSP 34 24
A14 D14
18
we 17
16
15
R3 R2 31 27
A15 D15 5V
10k 10k
DSP
TMS320LF2407, 144 pin device broken into its functional blocks
B 33V B
33V
H3
tms 1 2 trst R1
3
2
1
tdi 3 4 10k J7
5 6 5V C30
5V R4 .1uF
tdo 7 8
tck 9 10 10k
11 12
3
2
1
3
2
1
emu0 13 14 emu1 J4 J3
Header / s2212-07
A A
Title
TAMU Fuel Cell Inverter / DSP Layout
Size Number Revision
Letter
Date: 3-Sep-2001 Sheet of
File: C:\Mark\ee405\doe_ieee\protel\doe_v2.ddb Drawn By: M. Yeary
1 2 3 4
1 2 3 4 5 6
C24
33V U6 C26
D 1 D
c1+
C25 0.1uF 3 2 0.1uF
c1- v+
R8 6
v-
1.62k 4 0.1uF
c2+ H5
0.1uF 5
c2-
U12 33V C27 1
1 14 6
1A VCC
2 13 iopa0 11 14 2
1B 4B T1in T1out
rs 3 12 iopc0 10 7 7
2
1Y 4A T2in T2out
4 11 3
2A 4Y
5 10 13 12 8
2B 3B R1in R1out
6 9 8 9 4
2Y 3A R2in R2out
7 8 9
GND 3Y
15 16 5
VSS VCC 5V
J9 74LCX08
jumper max232 a23303 / rs232header
1
5V
R10
4.7k
5V
C C
R12
U4 1.62k
TPS7333 / 296-8066-5 U10
1 8 20
GND RESET* VCC
5V 2 7 H4 1 19
EN* FB/NC NC BE* C29 D2
3 6 33V pwm1 1 2
IN OUT 0.1uF LM4040AIM3-4.1
4 5 pwm2 3 4 iopa1 2 18
IN OUT A0 B0
pwm3 5 6 clkin 3 17
A1 B1
pwm4 7 8 4 16
A2 B2
pwm5 9 10 cap1 5 15
A3 B3
pwm6 11 12 cap2 6 14
A4 B4
13 14 iopc4 7 13
A5 B5
15 16 iopc5 8 12
A6 B6
Text DACout 17 18 pdpa 5V 9 11
A7 B7
19 20
21 22 10
GND
23 24
25 26 PI5C3245
27 28
29 30
31 32
33 34
B B
Header
5V U7
L4 R13
8 5
Fbead Vcc out
1 4 33
C28 nc GND
0.1uF xc263
decoupling caps
cc1
cc2
cc3
cc4
A A
Title
TAMU Fuel Cell Inverter / DSP Layout
Size Number Revision
B
Date: 3-Sep-2001 Sheet of
File: C:\Mark\ee405\doe_ieee\protel\doe_v2.ddb Drawn By: M. Yeary
1 2 3 4 5 6
Appendix B
SIMULATION RESULTS
Page 31 of 40
Simulation Results for the inverter system on 10kW load are presented here.
Page 32 of 40
Figure B3: Currents of the DC-DC converter (contd.)
where, I(D1) – Current through diode D1
Io+, Io- – Output currents of the DC-DC converter
Page 33 of 40
Appendix C
UC3825B DATASHEET
Page 34 of 40
application UC1823A,B/1825A,B
INFO UC2823A,B/2825A,B
available
UC3823A,B/3825A,B
High Speed PWM Controller
FEATURES DESCRIPTION
• Improved versions of the The UC3823A & B and the UC3825A & B family of PWM control ICs are
UC3823/UC3825 PWMs improved versions of the standard UC3823 & UC3825 family. Performance
enhancements have been made to several of the circuit blocks. Error ampli-
• Compatible with Voltage or
Current-Mode Topologies fier gain bandwidth product is 12MHz while input offset voltage is 2mV. Cur-
rent limit threshold is guaranteed to a tolerance of 5%. Oscillator discharge
• Practical Operation at Switching current is specified at 10mA for accurate dead time control. Frequency ac-
Frequencies to 1MHz curacy is improved to 6%. Startup supply current, typically 100µA, is ideal
• 50ns Propagation Delay to Output for off-line applications. The output drivers are redesigned to actively sink
current during UVLO at no expense to the startup current specification. In
• High Current Dual Totem Pole addition each output is capable of 2A peak currents during transitions.
Outputs (2A Peak)
Functional improvements have also been implemented in this family. The
• Trimmed Oscillator Discharge Current UC3825 shutdown comparator is now a high-speed overcurrent comparator
• Low 100µA Startup Current with a threshold of 1.2V. The overcurrent comparator sets a latch that en-
sures full discharge of the soft start capacitor before allowing a restart.
• Pulse-by-Pulse Current Limiting While the fault latch is set, the outputs are in the low state. In the event of
Comparator continuous faults, the soft start capacitor is fully charged before discharge
• Latched Overcurrent Comparator With to insure that the fault frequency does not exceed the designed soft start
Full Cycle Restart period. The UC3825 Clock pin has become CLK/LEB. This pin combines
the functions of clock output and leading edge blanking adjustment and has
been buffered for easier interfacing.
(continued)
BLOCK DIAGRAM
UDG-95101
* Note: 1823A,B Version Toggles Q and Q are always low
CONNECTION DIAGRAMS
ELECTRICAL CHARACTERISTICS: Unless otherwise stated, these specifications apply for TA = –55°C to +125°C for
the UC1823A,B and UC1825A,B; –40°C to +85°C for the UC2823A,B and UC2825A,B; 0°C to +70°C for the UC3823A,B and
UC3825A,B; RT = 3.65k, CT = 1nF, VCC = 12V, TA = TJ.
PARAMETER TEST CONDITIONS MIN TYP MAX UNITS
Reference Section
Output Voltage TJ = 25°C, Io = 1mA 5.05 5.1 5.15 V
Line Regulation 12 < VCC < 20V 2 15 mV
Load Regulation 1mA < IO < 10mA 5 20 mV
Total Output Variation Line, Load, Temp 5.03 5.17 V
Temperature Stability TMIN < TA < TMAX (Note 1) 0.2 0.4 mV/°C
Output Noise Voltage 10Hz < f < 10kHz (Note 1) 50 µVRMS
Long Term Stability TJ = 125°C, 1000 hours (Note 1) 5 25 mV
Short Circuit Current VREF = 0V 30 60 90 mA
2
UC1823A,B/1825A,B
UC2823A,B/2825A,B
UC3823A,B/3825A,B
ELECTRICAL CHARACTERISTICS: Unless otherwise stated, these specifications apply for TA = –55°C to +125°C for
the UC1823A,B and UC1825A,B; –40°C to +85°C for the UC2823A,B and UC2825A,B; 0°C to +70°C for the UC3823A,B and
UC3825A,B; RT = 3.65k, CT = 1nF, VCC = 12V, TA = TJ.
PARAMETER TEST CONDITIONS MIN TYP MAX UNITS
Oscillator Section
Initial Accuracy TJ = 25°C (Note 1) 375 400 425 kHz
Total Variation Line, Temperature (Note 1) 350 450 kHz
Voltage Stability 12V < VCC < 20V 1 %
Temperature Stability TMIN < TA < TMAX (Note 1) 5 %
Initial Accuracy RT = 6.6k, CT = 220pF, TA = 25°C (Note 1) 0.9 1 1.1 MHz
Total Variation RT = 6.6k, CT = 220pF (Note 1) 0.85 1.15 MHz
Clock Out High 3.7 4 V
Clock Out Low 0 0.2 V
Ramp Peak 2.6 2.8 3 V
Ramp Valley 0.7 1 1.25 V
Ramp Valley to Peak 1.6 1.8 2 V
Oscillator Discharge Current RT = Open, VCT = 2V 9 10 11 mA
Error Amplifier Section
Input Offset Voltage 2 10 mV
Input Bias Current 0.6 3 µA
Input Offset Current 0.1 1 µA
Open Loop Gain 1V < VO < 4V 60 95 dB
CMRR 1.5V < VCM < 5.5V 75 95 dB
PSRR 12V < VCC < 20V 85 110 dB
Output Sink Current VEAOUT = 1V 1 2.5 mA
Output Source Current VEAOUT = 4V –0.5 –1.3 mA
Output High Voltage IEAOUT = –0.5mA 4.5 4.7 5 V
Output Low Voltage IEAOUT = 1mA 0 0.5 1 V
Gain Bandwidth Product F = 200kHz 6 12 MHz
Slew Rate (Note 1) 6 9 V/µs
PWM Comparator
RAMP Bias Current VRAMP = 0V –1 –8 µA
Minimum Duty Cycle 0 %
Maximum Duty Cycle 85 %
Leading Edge Blanking R = 2k, C = 470pF 300 375 450 ns
LEB Resistor VCLK/LEB = 3V 8.5 10 11.5 kohm
EAOUT Zero D.C. Threshold VRAMP = 0V 1.1 1.25 1.4 V
Delay to Output VEAOUT = 2.1V, VRAMP = 0 to 2V Step (Note 1) 50 80 ns
Current Limit/Start Sequence/Fault Section
Soft Start Charge Current VSS = 2.5V 8 14 20 µA
Full Soft Start Threshold 4.3 5 V
Restart Discharge Current VSS = 2.5V 100 250 350 µA
Restart Threshold 0.3 0.5 V
ILIM Bias Current 0 < VILIM < 2V 15 µA
Current Limit Threshold 0.95 1 1.05 V
3
UC1823A,B/1825A,B
UC2823A,B/2825A,B
UC3823A,B/3825A,B
ELECTRICAL CHARACTERISTICS: Unless otherwise stated, these specifications apply for TA = –55°C to +125°C for
the UC1823A,B and UC1825A,B; –40°C to +85°C for the UC2823A,B and UC2825A,B; 0°C to +70°C for the UC3823A,B and
UC3825A,B; RT = 3.65k, CT = 1nF, VCC = 12V, TA = TJ.
PARAMETER TEST CONDITIONS MIN TYP MAX UNITS
Current Limit/Start Sequence/Fault Section (cont.)
Over Current Threshold 1.14 1.2 1.26 V
ILIM Delay to Output VILIM = 0 to 2V Step (Note 1) 50 80 ns
Output Section
Output Low Saturation IOUT = 20mA 0.25 0.4 V
IOUT = 200mA 1.2 2.2 V
Output High Saturation IOUT = 20mA 1.9 2.9 V
IOUT = 200mA 2 3 V
UVLO Output Low Saturation IO = 20mA 0.8 1.2 V
Rise/Fall Time CL = 1nF (Note 1) 20 45 ns
UnderVoltage Lockout
Start Threshold UCX823B and X825B only 16 17 V
Stop Threshold UCX823B and X825B only 9 10 V
UVLO Hysteresis UCX823B and X825B only 5 6 7 V
Start Threshold UCX823A and X825A only 8.4 9.2 9.6 V
UVLO Hysteresis UCX823A and X825A only 0.4 0.8 1.2 V
Supply Current
Startup Current VC = VCC = VTH(start) –0.5V 100 300 µA
Icc 28 36 mA
Note 1:Guaranteed by design. Not 100% tested in production.
APPLICATIONS INFORMATION
OSCILLATOR Oscillator
The UC3823A,B/3825A,B oscillator is a saw tooth. The
rising edge is governed by a current controlled by the RT
pin and value of capacitance at the CT pin. The falling
edge of the sawtooth sets dead time for the outputs. Se-
lection of RT should be done first, based on desired
maximum duty cycle. CT can then be chosen based on
desired frequency, RT, and DMAX. The design equations
are:
3V
RT =
(10mA)(1 – D MAX )
(1.6 • DMAX )
CT =
(RT • F )
4
Appendix D
Page 39 of 40
Texas A&M University
2001 Future Energy Design Team
Final Report
Faculty Advisors
Student Members
ii
Faculty Advisors
____________________________________ ___________________________________
Dr. Prasad Enjeti Dr. Mark Yeary
Department of Electrical Engineering Department of Electrical Engineering
e-mail: enjeti@ee.tamu.edu e-mail: mbyeary@ee.tamu.edu
_______________________________________ ______________________________________
Dr. Jo Howze Dr. Charles Culp
Department of Electrical Engineering Department of Mechanical Engineering
e-mail: howze@ee.tamu.edu e-mail: cculp@esl.tamu.edu
iii
Report Authors
___________________________________ ___________________________________
Rajesh Gopinath Matthew Webster
Department of Electrical Engineering Department of Electrical Engineering
______________________________________ ______________________________________
Phillip Briggs Douglas Becker
Department of Computer Engineering Department of Chemical Engineering
____________________________________ ___________________________________
Steven Campbell Jon Burghardt
Department of Electrical Engineering Department of Electrical Engineering
____________________________________ ___________________________________
Samsung Kim Chiranjib Mukherjee
Department of Electrical Engineering Department of Electrical Engineering
___________________________________
Justin Busse
Department of Electrical Engineering
iv
Table of Contents
1. Summary ..................................................................................................................................1
2. Introduction.............................................................................................................................2
v
4.1 Tracking Chart & Budget..............................................................................................36
4.2 DC-DC Converter Costs................................................................................................40
4.3 DC-AC Inverter Costs....................................................................................................41
5. Demonstration of Operational success of the 1.5kW Prototype ............................42
Design of the Battery Backup System: ...........................................................................44
6. Responsibility Matrix & Organizational Approach .....................................................46
6.1 Institutional Commitment and Sources of Added Support ................................47
6.2 Impact on Undergraduate Education........................................................................47
7. Nomenclature .......................................................................................................................50
9. Bibliography..........................................................................................................................52
vi
List of Figur es
Figure 1: Block Diagram of the TAMU Fuel Cell Inverter ......................................................... 6
Figure 4: Motorola SG3525A Control chip for the TAMU DC-DC Inverter............................ 10
Figure 8: Equivale nt Circuit for Single-Phase Inverter Output Filter Stage and Load....... 22
Figure 14: DC Input into to the DC-AC Inverter and a Single Phase AC Output................. 30
Figure 18: Quantity and Power Schematic and Rating Take-Off Sheet............................... 39
Figure 20: DC Input into to the DC-AC Inverter and a Single Phase AC Output................. 43
vii
List of T ables
viii
1 . S u mma r y
This report describes the development of a low cost fuel cell inverter with DSP control to meet
the 2001 Future Energy Challenge competition. A one-year project under EE-405 Electrical Design
Laboratory course to address the 2001 Future Energy Challenge was launched with undergraduate student
participation. The Texas A&M team was comprised of competent senior undergraduate students along
The proposal outlines the technical approach to achieve the objectives proposed by the 2001
Future Energy Challenge organizing committee. The Texas A&M team believes it has developed an
efficient and effective inverter system. The team has developed a unique digital signal processor (DSP)
control mechanism for DC-AC control, an efficient push pull topology DC-DC converter and a rigorous
cost reduction approach for the 2001 Energy challenge inverter project.
A low cost Texas Instruments, TMS320F2407, DSP provides the control scheme for the DC-AC
inverter system. The DSP provides closed loop control for the DC-AC converter allowing easy
compliance to the total harmonic distortion (THD) specification of less than 5%. The DSP allows
convenient communication between the fuel cell and the inverter, and, through the RS232 port, allows
communication of information to data collection software or to the Internet. Since the DSP is
programmable, control algorithms are easily updated as opposed to traditional hard wire devices.
The Texas A&M team provides a rigorous cost savings approach by reducing the number of
(IGBT) in the design. Incorporating many small-power rated resistors and fewer power resistors
enhances cost savings. The push-pull DC-DC converter topology provides isolation for safety, suitable
boosting of the fuel cell voltage to 400 volts, reduced cost and reduced size of the energy storage
1
2. Intr oduction
Distributed power systems including fuel cells, microturbines, flywheels and wind turbines offer
a potential increase in energy efficiency by localizing power generation and eliminating the need for line
transmission of electricity [1]. Even though these environmentally friendly, highly efficient energy
resources are promising, several barriers must be overcome. A Department of Energy (DOE) study,
Making Connections, completed in May 2000 addressed the technical, business practice, and regulatory
barriers affecting distributed power systems. Since the barriers have been identified, rapid progress has
The Future Energy Challenge 2001 has identified the fuel cell as a distributed energy technology
that will soon be affecting the energy market. One of the main barriers for fuel cell technology is the cost
of manufacturing and the cost of power conditioning and control. Currently, fuel cell production costs are
decreasing, and have nearly achieved energy costs that are competitive with local utility rates. To further
assist the reduction of cost, the price of the power-conditioning portion of the fuel cell system must also
decrease, while at the same time increasing efficiency, reliability, and power quality. Lower cost will
enable the fuel cell systems to achieve a production cost at a more competitive rate than that offered by
many local utility companies, thus triggering rapid penetration into the utility market.
The 2001 Future Energy Challenge has resolved that one of the main components in the power
conditioning system is the inversion of direct current (DC) power from the fuel cell to consumer usable
alternating current (AC). The challenge requires the inverter design to be small, efficient, environmentally
compatible and low cost. A low cost inverter approach will help enable small-scale fuel cell system
commercialization and will encourage the development of distributed power systems. The 2001 Energy
Challenge invited participants to design and develop a low cost fuel cell inverter system that will perform
Fuel Cells for distributed power have many advantages. Environmental acceptability, efficiency,
distributed capacity; fuel flexibility and cogeneration are reasons why the fuel cell should be promoted as
the next generation of power. The following is a list of advantages of fuel cells:
• Environmental Acceptability - Because fuel cells are so efficient, CO2 emissions are reduced
for a given power output. Fuel cell power plants are projected to decrease CO2 emissions by a significant
amount in the next few years. The fuel cell is quiet, emitting only 60 decibels at 100 feet. Emissions of
SOx and NOx are 0.003 and 0.0004 pounds/megawatt-hour respectively. Fuel cells theoretically can be
designed as water self-sufficient. Since fuel cell exhaust is primarily water and CO2 natural gas fuel cell
power plants have a blanket exemption from regulations in California's South Coast Air Quality
Management District. These emission restrictions are possibly the strictest in the nation.
• Efficiency - Dependent on type and design, the fuel cells direct electric energy efficiency
ranges from 40 to 60 percent low heating value (LHV). The fuel cell operates at near constant efficiency,
independent of size and load. Fuel cell efficiency is not limited by the Carnot Cycle. For the fuel cell/gas
turbine systems, electrical conversion efficiencies are expected to achieve over 70 percent (LHV). When
by-product heat is utilized, the total energy efficiency of the fuel cell systems approach 85 percent.
• Distributed Capacity - Distributed generation reduces capital investment and improves the overall
conversion efficiency of fuel to end use electricity by reducing transmission losses. In high growth, or
remotely located load demands, distributed generation can reduce transmission and distribution problems
by reducing the need for new capacity or siting new lines. Presently, 8-10 percent of the generated
electrical power is lost between the generating station and the end user. Distributed generation will result
3
in many smaller units distributed throughout the United States. Many smaller units are statistically more
reliable since the probability of all distributed units failing at once is negligible.
• Permitting - Permitting and licensing schedules are short due to the ease in siting. In fact, natural
gas fuel cell power plants have been exempt from many of California's environmental regulations.
• Modularity - The fuel cell is inherently modular. The fuel cell power plant can be configured in a
wide range of electrical outputs, ranging from a nominal 0.025W to greater than 50-megawatt (MW) for
the natural gas fuel cell to greater than 100-MW for the coal gas fuel cell.
• Fuel Flexibility - The primary fuel source for the fuel cell is hydrogen, whic h can be obtained from
natural gas, coal-gas, methanol, landfill gas, and other fuels containing hydrocarbons. This fuel flexibility
means that power generation can be assured even when a primary fuel source is unavailable.
• Cogeneration Capability - High-quality heat is available for cogeneration, heating, and cooling.
Fuel cell exhaust heat is suitable for use in residential, commercial, and industrial cogeneration
applications .
general, fuel cells produce a rectified voltage from an electrochemical reaction between a hydrogen-rich
fuel gas and an oxidant (air or oxygen). The principal by-products are water, carbon dioxide, and heat.
Fuel cells are similar to batteries in that both produce a DC voltage by using an electrochemical process.
Two electrodes separated by an electrolyte make up an anode and a cathode pair called a cell. Groups of
cells are called stacks and produce useable voltage and power output. Unlike batteries, however, fuel cells
do not release stored energy; instead they convert energy from hydrogen-rich fuel directly into electricity.
Fuel cells operate as long as they are supplied with fuel. Further, fuel cells have a large time constant
(several seconds) to respond to an increase or decrease in power output. In view of this, a stand-alone fuel
4
cell power system requires some amount of battery backup to accommodate fluctuating electric loads.
TAMU Fuel Cell inverter incorporates this feature and is detailed in later parts of this report.
The inverter and the fuel cell have some unique interdependencies. The inverter and the fuel cell
must work together to produce AC power and therefore must communicate with each other. For the 2001
Future Energy Challenge, four basic controls were required: A digital, 0-5 volt, on/off request from
inverter to fuel cell, a 0-5V analog signal to the fuel cell requesting power, (5V corresponds to 1500W in
2001 Challenge), a 0-5 volt, analog output proportional to the power available and a digital, ready/trip, 0-
The amount of energy a fuel cell can produce is dependent on the total potential of the stack and
the current demanded from the stack. The fuel cell will only provide current in the amount available from
the total chemical reaction within the fuel cell stack. This reaction is dependent on the quantity of fuel
and oxidant available to the fuel cell stack. In general, the fuel cell stack must have the fuel and oxidant
available prior to any load increase. The fuel cell controller will control the high fuel utilization and the
low fuel utilization limits. This leading indicator characteristic required by the fuel/oxidant flows
requires a signal of load increase prior to the fuel cell actually seeing the load increase. A decrease in
fuel/oxidant flows is not as critical and can be reduced directly as load reduces. A digital, 0-5 volt, on/off
request from inverter to fuel cell will tell the fuel cell to turn on or off. The TAMU fuel cell inverter
generates a 0-5V signal to the fuel cell requesting power from a minimum condition (idle) to a maximum
level. A 0-5 volt, “power-available”, analog signal from fuel cell to the inverter, as an indicator, is
The fuel cell must not exceed its maximum allowable limits of heat and load. In the case of over
heating or short circuit seen by to fuel cell a digital, ready/trip, 0-5 volt signal from fuel cell to Inverter
will tell the inverter if the fuel cell is ready. Since the fuel cell response to an increase in power is large
(several seconds), the TAMU fuel cell inverter incorporates battery backup system for sudden load
Figure 1 shows the block diagram for the TAMU fuel cell inverter. In general an inverter system
consists of a DC-DC boost circuit, a DC-AC inverter circuit and a filter. This section will briefly describe
how the boost circuit works, how an inverter creates an AC output from a DC source, what types of
+ +
48V Load
- -
Vo*
SG3525A DSP
TMS 320C2407 io
The DC input from the fuel cell (48 VDC nominal, +50%, -12.5%) is first converted to a regulated
400 VDC using a high frequency DC-DC converter. The DC-DC conversion stage consists of a high-
frequency transformer. Isolated primarily for safety, system protection, and to meet the stringent FCC
Class-A standards. The 400V DC-DC converter output is converted to 120V/240V, 50/60 Hz, single -
phase AC by means of a pulse width modulation (PWM) driven IGBT, inverter stage. An output-LC
filter stage is employed to produce a low THD-AC waveform. Low loss, high switching frequency
MOSFET and IGBT components have been employed to achieve a higher efficiency, lower size and
volume of the fuel cell inverter system. The circuit topology of the TAMU inverter system is shown in
Figure 2 below.
6
1:5 L IDC
+
T+ AC Output
Iin C Lb TA+ TB+ 120/240V , 60 Hz
FUEL CELL N2 D1 D3 LF
VC+ ia
Vin Vbatt A
N1
42-72V DC + C VDC 240V, 60Hz
48V nom. LF ib
-
N1 C Lb B
N2 VC - Va
D4 D2 Vbatt CF C F
- TA - TB - Vb 120V, 60Hz
T-
L
N
Battery Backup
48VDC / 400VDC, 40KHz PUSH PULL CONVERTER 120V/240V, 20kHz PWM INVERTER
DC-DC converter and inverter topologies were designed to achieve ease of manufacturing and
facilitate production in large volume. Another unique aspect of the design is the use of the
TMS320C2407 DSP to control the inverter. The DSP reduces printed circuit board layout complexity.
Readily programmable, the DSP adds flexibility to implement various control aspects by means of
software. In addition, the DSP incorporates imbedded intelligence into the design.
The TAMU fuel cell inverter employs a push-pull type DC-DC converter to suitably boost the
fuel cell voltage from 48V to 400V. Figure 3 shows the topology of the push pull DC-DC converter.
The push-pull, full bridge and flyback converters belong to the family of isolated buck converters.
This family of converters may be used in conjunction with a high frequency transformer to boost the
output voltage with the additional advantage of providing isolation between the input and output stages.
Isolation of the input and output stages provide safety of personnel accessing the output terminals and
enhance short circuit protection for the inverter. The DC-DC converter operates at high switching
frequency (40 kHz), which produces high frequency AC across the isolation transformer. The secondary
output of the transformer is rectified and filtered to produce 400VDC. The design is rated for 10 kW and
7
consists of parallel-connected MOSFETs, a full-bridge rectifier, a Motorola SG3525A control chip for
feedback control, snubber circuitry, a high frequency transformer, a coupled inductor and bulk capacitors.
The output voltage is regulated by means of feedback control employing a low cost Motorola
I in 1:K ID L IO
> > >
IT
<
D1 D4 C
T+
N2
Vin + N1
PWM RL
- CONTROL VO
N1 C
T- N2 D3 D2
The push-pull converter consists of two forward converters driven by anti-phase inputs. The two
diodes D1 and D2 in the secondary of the transformers act as both forward and flywheel diodes. Ideally,
conduction times of T+ and T- are equal and the transformer is driven symmetrically. The primary side
conduction losses are lower since at any given instant only one transistor is connected in series with the
DC source. Since in the full bridge push-pull converter, both the halves of the secondary winding
conduct, the turn ratio (N2/N1) can be minimized, reducing the transistor currents.
The transistors T+ and T- are switched alternately with a pulse width modulated signal to produce
a high frequency AC at the input of the transformer. A center-tapped secondary is used. The neutral of
the center-tapped secondary is connected to the center point of the bulk capacitor. When T+ is on, D1
and D2 conduct whereas D3 and D4 are reverse biased. This results in voltage
N2
vL = 2 V −V 0 < t < t ON (1)
N1 D 0
where
8
vL - inductor voltage
N1 - primary turns
N2 -secondary turns
VD -input voltage
Vo - dc output voltage
t - time
and the inductor current increases linearly. During the interval ∆ when both the switches are off, the
The next half cycle consists of tON (during which T- is on) and the interval ∆. The waveforms
1
t ON + ∆ = Ts (2)
2
V0 N2
=4 D 0 < D < 0 .5 (3)
VD N1
where D = tON /Ts is the duty ratio of the switches T+ and T- and has the maximum value of 0.5.
The duty ratio D through the controller regulates the output voltage V0 . An efficient control scheme will
eliminate the effect of disturbances, i.e. reject the input power supply variations and transient load
changes.
A challenge that arises with the push-pull topology is that the transformer core may saturate if the
characteristics of the forward-voltage drop and conduction times of the transistors are not precisely
matched. Small imbalances can cause the DC component of voltage applied to the transformer-primary to
be nonzero. Consequently, during every switching period, there is a net increase in the magnetizing
9
current. If this imbalance continues, then the magnetizing current can eventually become large enough to
saturate the transformer. Core saturation results in rapid thermal runaway and destruction of one of the
transistors. To ensure that there is no significant imbalance between the two switch currents a coupled
inductor is employed on the secondary of the transformer. The coupled inductor balances the currents in
the two halves of the center-tapped transformer. Additionally, the choke filters out the switching
frequency components off the DC output current and balances the power output of each inverter phase,
(which are specified to be capable of being loaded independently), and helps generate a clean 400VDC
A low cost Motorola SG3525A Pulse Width Modulator controller is used for the control of the
DC-DC converter. The SG3525A is particularly suited for the push-pull converter application because it
has two output terminals. The two output terminals work perfectly with the two transistors used in this
converter design. The block diagram of the Motorola SG3525A is shown below in Figure 4.
Figure 4: Motorola SG3525A Control chip for the TAMU DC-DC Inverter
10
The Motorola SG3525A provides this design with other features that enhance control and safety.
The SG3525A provides for an input under-voltage lockout that automatically shuts off the chip in case of
low voltage. The modulator has a soft start capability, which allows it to be protected from capacitor
inrush currents.
The feedback board is a voltage divider that scales the 400V output to a level that can be
managed by the pulse width modulation (PWM) controller. Since the ground is at the midpoint of the
400V, the output voltage appears as a +200V signal and a -200V signal. Both of these voltages are scaled
down to a few volts by a resistive divider. The parallel combination of the resistors at the +200V divider
is equal to the parallel combination of the resistors at the -200V level. The schematic for the feedback
N15 C8
0.1uF
50V
U9B
11
6 - TA75074P R18
R1 R2 R12 7
5
+BUS V +
22 K
200K 200K 39 K
4
C23 U9A
R13 C7
4
470pF P15 3 TA75074P
5.1 K 50V 0.1uF + 1
50V 2
- BUSFB
11
C34 R17
R14
470pF
7.5 K 16 K R21
50V
U9C 16 K
11
R20 R22
22 K 5.2 K
A small positive signal and a small negative signal are produced. The signals are the same
polarity in order that they can be added together. The added signal is buffered from the resistive divider
so other circuit resistances do not affect the scaling factor. This is achieved by using a unity-gain op-amp
11
for the positive signal and an inverting unity-gain op-amp for the negative signal. These signals are
mixed and fed into a non-inverting op-amp with a variable gain such that the controller sees a single
ended 6V signal at the feedback input when the output voltage is 400V.
temperatures or shut down conditions in the circuit to prevent damage to the DC/DC boost stage. A user
signaled shutdown and fuel cell signaled shut down are provided for as well.
Op-amp comparators and sensors are used to monitor operation and provide a shutdown signal to
the controller. The protection circuit uses a reference voltage for each signal to set the maximum limit
and a signal representing the measured quantity. Whenever any of the measured signals exceeds their
reference, the controller SD (shutdown) pin is pulled low which blocks the gating signals to the transistors
and shuts down the converter. Indications for over current, thermal overload and fuel cell interrupt
conditions are provided through light emitting diodes (LED) mounted on the faceplate.
Each signal is fed to a separate comparator with a separate reference voltage. When the measured
quantity exceeds the maximum threshold, the comparator output is pulled high. This forward biases a
series LED, which is mounted on the front of the enclosure as a visual indicator. Each signal has an op-
amp/LED combination, and the cathodes of all the LEDs are connected together to the base of a bipolar
junction transistor (BJT) pull-down transistor. If any signal exceeds the threshold the LED lights and the
base gets a high signal. The collector is pulled to ground, which sends a low voltage to the shutdown
While the references are obtained with potentiometers connected to the control supply, the signals
need some external circuitry. A DC current sensor is needed on the output line of the converter to provide
a voltage proportional to the current. The over current threshold is set to 110% of the full load current.
Any current above 110% of full current will shut down the converter.
12
A negative temperature coefficient (NTC) thermistor is mounted on the heatsink for thermal
protection of the converter. As the temperature varies, the resistance of the thermistor will change and
force a small differential voltage at the output of the Wheatstone bridge. This voltage needs to be
buffered and converted to a single -ended voltage with amplitude of a few volts. The circuit is designed to
The fuel cell or a user-defined shutdown is simple to accomplish. Because this signal will be a
relatively low voltage (~5V), it will need boosting to properly light its LED. Feeding this signal directly
to a comparator and providing a reference of 1V accomplished the user defined shut down. Whenever the
signal is high, the comparator will provide +15V to light the LED. If the signal is low, the comparator
will provide -15V to keep the LED reverse biased and unlit.
switching on the DC-DC board, the output of the DC-DC inverter exhibits some switching noise (40kHz).
The noise is present in the signal sent to the DC-DC controller and in the DC voltage sent to the DC-AC
inverter. The Texas A&M team’s design includes filtering of both the control signal and the DC voltage
Switching noise should be removed from the control signal sent back to the comparator so the
accuracy of the control loop is not compromised. Two ceramic capacitors are included in the circuit prior
to the resistive divider of the feedback board. The capacitors filter out the switching noise from the
feedback circuit. Wire leads that implement the summing of the signal can pick up noise easily, therefore
the wire leads are as short as possible. These features provide a clean DC control signal.
13
3.4.7 DC-DC Converter DC- Link Design
Two high frequency film capacitors rated at 0.22µF capacitors provide the filtering for the DC
voltage sent to the DC-AC inverter. The series capacitors remove the 40 kHz voltage noise and provide
the DC-DC converter with a clean 400 V output. The bulk (electrolytic) capacitors provide additional
conditioning for the DC-AC inverter and serve to balance the ±200V voltages for the single -phase
inverters. In addition, two 47KΩ, 2W resistors are connected across the DC link for safe discharge of the
3.4.8 DC-DC Converter Design For The 10kW TAMU Fuel Cell Inverter System
Figure 3 in the beginning of this section shows the circuit diagram of the push-pull DC/DC
converter. Fuel cell output is connected to the DC/DC converter as shown. The operation of the push
pull converter is described in detail in section 3.4.1 above. MOSFETs T+ and T- are turned on and off
10000 W
Pin = = 11111 W (4)
0 .9
Designing for the low input line condition (Vin =42VDC), input current from fuel cell,
11111 W
I in = = 265 A (5)
42V
The push pull DC/DC converter shown in Figure X. comprises of two switches, T+ and T-. At the
MOSFETs rated 100V, 100A with 2 devices in parallel in each leg are selected.
For obtaining an output voltage of 400VDC for the push-pull converter, a turns ratio of K=5 is selected
for the transformer. Center taps are available on both the primary and secondary sides as shown in Figure
The VA rating of the transformer is defined as the sum of the total primary and secondary
1 Vin Iin I
VATr = ⋅ ⋅ 2 + Vin ⋅ K ⋅ in ⋅ 2 = 1. 5 Vin ⋅ Iin = 1. 5 ⋅ 42 ⋅ 265 = 16695 W ≅ 17. 0kVA (7)
2 2 2 K
Voltage ratings of the transformer are selected as: Primary voltage=80V, Secondary voltage=400V
Diode ratings:
I in
ID = = 37 .5 A (8)
K⋅ 2
The PWM controller SG3525A is used for the control of the push-pull DC-DC converter. The block
diagram of the SG3525A is shown in Figure 6. The error amplifier used for implementing the closed loop
voltage control is a part of the SG3525A. The resistors and capacitors shown below are external to the
chip and were selected as follows. Phase compensation is achieved by a type-2 amplifier.
15
C2
R2 C1
Vfdbk R1
- Vout
Rbias +
Vref
To achieve a phase boost of 82° and dB gain (G) of 2.83 at the center frequency (f) of 455Hz, we select
π boost
k = tan + 45 = 14. 3 (9)
180 2
1
C2 = = 453 pF ≅ 470 pF
2π f G k R1
C1 = C 2( k 2 − 1) = 0 .092 µF ≅ 0 .1µF (10)
k
R2 = = 54 kΩ ≅ 51 .1kΩ
2 π f C1
For the oscillator section, a timing resistor RT and capacitor CT are selected to obtain a switching
16
Component Type Rating Quantity
The voltage ripple on the 400VDC bus as obtained by simulation is less than 1%.
The DC-AC subsystem consists of the circuitry between the DC-DC converter and the load. A
IDC is
+
ic AC Output
TA+ TB+ 120/240V , 60 Hz
LF ia
VC+ i sA i sB i oA
A
C
400VDC O
VDC 240V, 60Hz
LF ib
i oB
B
C
VC - Vb
CF CF 120V, 60Hz
TA - TB - Va
-
17
Since the DC-DC converter maintains equal ±200V on the dc-link capacitors, two inverter legs are
sufficient to generate 120V/240Vac output. The inverter has two PWM modules, which can each
accomplish a task such as centered and/or edge-aligned PWM generation. The 400 V output of the DC-
DC boost is used across two parallel switching legs. Each leg consists of two IGBT’s connected in series
across the 400 V input. The IGBTs are switched by the DSP. The DSP uses PWM by means of a
software-controlled algorithm to emulate a sine wave to determine when to open and close the gates.
An LC filter is used to filter out harmonics above 60 Hz out of the PWM output. The voltage sensor
senses the output voltage across the external load. This sensed voltage is fed back into the DSP for use in
the PWM algorithm. If the voltage is too high the voltage will be decreased and if the voltage is to low it
will be increased. The current sensor serves as over-current protection for the load. This subsystem can
be broken down into two main components, the control circuitry and the passive output AC filter.
3.5.1 Inverter Design Procedure for the 10kW TAMU Fuel Cell Inverter System
The inverter produces two single -phase outputs Phase-A and Phase-B. It is comprised of two half
bridge inverters each supplying separate single -phase loads at 120VAC, 60Hz. Consider the case when
Phase-B is not loaded and Phase-A is on full load (5000W). The peak amplitude of the fundamental
frequency component is the product of ma and ½ VDC, where ma is the modulation index. A modulation
V DC
(VAO )1 = m a ⋅ sin(ω1t ) 0 < ma < 1 (11)
2
0.9
sw1 = 0.5 + sin ω1t + higher frequency terms (12)
2
and the Phase-A output current is assumed to contain fundamental and third harmonic component due
to nonlinear load.
18
ioA = 2 I1 sin(ω1t − φ1 ) + 3 I 3 sin(3ω1t − φ3 ) + ... (13)
If Irms is the rms value of the Phase-A output current, neglecting higher frequency terms, we have
Assuming I3 =0.7 I1 which is typical of single phase rectifier type nonlinear loads,
I rms = 1 .22 ⋅ I 1
5000
I rms = = 41 .7 A
120
which gives
41 . 7
I1 = = 34 A (16)
1 .22
Therefore, the largest component of the capacitor current ic is the fundamental frequency current, the rms
1
ic , rms = ⋅ I1 = 17 A (17)
2
ic, rms
∆Vc = (18)
ωC
ic , rms 17
C= = ≅ 4500 µF (19)
ω ∆ Vc 10 ⋅ 2π ⋅ 60
19
Inverter switch ratings:
The rms current isA is 41.7A. Thus, rms current rating of each switch is
41 .7
IT = = 30 A (20)
2
Digital systems are becoming more ubiquitous in industry as a consequence of automated design
tools becoming more prevalent. Also, the accuracy that digital systems offer may be efficiently verified
Unlike analog circuits, sophisticated algorithms can be implemented and updated digitally in a
short period of development time. Furthermore, digital circuits are less likely to be influenced by
temperature, aging, process technology, or chip layout. In contrast, analog circuits typically require
additional tuning circuits. In order to avoid use of tuning circuits and non-linearities associated with
A Texas Instruments TMS320C2407 DSP platform was implemented to obtain the closed loop
control and PWM functions via software that maximize overall performance of the fuel cell inverter
system, while allowing the low cost objective to be achieved. The TMS 320C240X is a high-speed
processor designed for power electronic control. The DSP has on-chip memory to store and run the
program. Essential to the decision to implement DSP control, the cost of the Texas Instruments
20
The DSP system has a high-speed A/D converter, 9 PWM output channels and serial
communication capabilities. In addition, the TMS 320C240X series contains a 10-bit analog-to-digital
converter (ADC) having a minimum conversion time of 500 ns that offers up to 16 channels of analog
input. The auto sequencing capability of the ADC allows a maximum of 16 conversions to take place in a
single conversion session without any central processing unit (CPU) overhead [2]. This capability is
important for exploring ideas such as sensing outputs and inputs, programmable dead band to prevent
shoot-through faults, and synchronized analog-to-digital conversion [2]. By implementing the control via
DSP, the proposed approach will offer increased flexibility, insensitivity to temperature drifts and will
minimize component cost. Here, control is defined to mean how the DSP is used to: 1) modulate the
firing angles of the IGBT’s that will be used in the inverter stage, 2) provide supervisory protection
against system over-voltage, over-current, and over-temperature conditions, and 3) examine critical fuel
cell integration control parameters, 4) and provide tight output voltage regulation to meet THD
specifications under varying linear and nonlinear loading conditions. Appendix C shows the code
The system involves high-speed feedback loops that force the actual output voltage, Vout , to
follow the reference sinusoidal voltage Vref. Vref is 170V and is internally generated by the DSP
The voltage feedback configuration takes Vo (output of inverter) and divides out sin(ωt). This
gives DC component or the peak voltage for the output. The peak is then compared to the 120√(2)V or
~170V. The error is given to the PI controller where a voltage compensation amount is output into the
summer. The summer adds the correction to the reference and multiplies sin(ωt) back into the DC output
making it an AC signal again. The AC signal is then entered into a voltage limiter, which constrains the
voltage to a range of 0.0 to +3.3V. All intermediate values are linearly scaled into the range +3.3V to 0V.
21
The signal has a DC bias (offset) of 1.65V. This provides a range that covers the whole output spectrum
and limits it to a more manageable range for the DSP. The DSP can handle an input ranging from 0V to
3.3V.
In this section, a closed-loop control strategy to maintain sinusoidal voltage at the output
terminals for varying loads is discussed. According to the specifications the 120V and 240V terminals of
the inverter could be connected to linear and nonlinear loads. In addition, the load can be considerably
unbalanced between the two outputs. Figure 8 shows the equivalent circuit of the output filter (LC) stage
of the inverter. Section 3.5.6 of this report details the output filter selection procedure along with a design
example. Figure 9 on the next page shows the block diagram of the output voltage control.
iL L io
ic
vinv C vo
Figure 8: Equivalent Circuit for Single -Phase Inverter Output Filter Stage and Load
The output voltage of the inverter is sampled and transformed to the synchronous reference frame
in order to cancel higher harmonic components employing low pass filter (LPF). The transformation
sin θ cosθ
T(θ ) =
− cosθ sin θ , (21)
22
sin θ − cosθ
T(θ ) −1 =
cosθ sin θ . (22)
The transformation matrix T can be used for a system which might have two phases with 90° phase shift
to be represented by dc component. Since the output voltage is only available, another phase input is
e*
determined by v od cos ωt . As shown in Figure 9, the control system is designed on the synchronous
reference frame to increase the performance for nonlinear loads. Inverter input command is obtained by a
repetitive controller with predicted output voltage. During the fundamental cycle period, PI controller
output is calculated, applied to the next cycle period by integrating all the previous controller output.
i
veod* = 120 2 ∫0 ℜinv(i + 1)n −1 dt
*
M sin ωt
vo veod +
LPF _ PI ℜ*inv (i )n −1 ℜ*inv(i + 1)n +
v*o cos ωt T (θ ) e
voq T (θ )−1 ∫ +
LPF _ PI v*inv(i + 1) n
+
e*
voq =0
*
v inv ( i + 1 ) n = M sin ω t + ℜ *inv ( i + 1 ) n (23)
ℜ*inv (i + 1) n = I ℜ ∫ ℜ*inv (t ) n −1 dt
i
0 (24)
Where, M is the modulation index, n denotes the step of a fundamental period; i represents the sampling
23
Results:
Figure 10 and Figure 11 show the simulation results of the inverter output voltage control
scheme. It is clear from the results that the inverter output voltage maintains a sinusoidal wave shape
Current and temperature protection are important for the output load and the inverter
itself. High current and high temperature can cause unexpected and serious damage to the fuel
into 1V. A full bridge rectifier and a capacitor are used to obtain a DC level corresponding to the
Another op-amp uses the voltage and compares it to a reference voltage by means of a
comparator configuration. If the voltage measured is larger than the reference, then a signal is
heat sink per leg of the inverter. The thermistor is used in a Wheatstone bridge. The voltage
difference across the bridge proceeds to an op-amp in a voltage follower configuration. Another
op-amp compares it to a reference voltage. If the voltage measured is larger than the reference, a
A manual push button reset is used to restart the system after the current or the
temperature has reached its limit and the gate driver has shutdown.
The output filter of the power inverter is used to smooth out the waveforms generated from our
DC-AC stage. If monitored, the pure output of the inverter is a square wave with varying duty cycles.
This signal contains many unwanted frequencies including multiples of the 20kHz pulse width modulation
(PWM) switching frequency. Total harmonic distortion (THD) can be affected by these harmonics,
therefore, the THD of the inverter can be reduced by using an output filter. The THD requirement for this
25
design requires the system to have a THD < 5%. This low-pass filter is designed to meet the THD and
power requirements of this project. The following section will describe the procedure used to find the
the schematic. The assumptions used in the analysis are, the output filter is loss less and the third
jnXL
Vi,n -jXC
Vo,n ZL1n
n
The transfer function for this type of filter is described by the equation
Vo , n jX C ⋅ Z L, n
Hn = =− . (25)
Vi , n nX L X C + jZ L ,n ( n 2 X L − X C )
Where
Hn - transfer function
26
Z L, n - impedance
n - harmonic
− jX C ⋅ Z L ,1
H1 ≤ ≅ 1. (26)
− jZ L ,1 ⋅ X C
XC 1
Hn = − = (27)
n X L − X C n 2 ⋅ L −1
2
X
XC
1 X 23.222 (28)
≤ 0.045 = L ≥
X XC n2
n 2 ⋅ L −1
XC
jhXL
-jX C
Vh Ih
h
jhX L ⋅ X C
Vh = ⋅ Ih . (29)
X C − h2 X L
Where
27
Vh - equivalent voltage
h - harmonic
Ih - current at h harmonic
hX L
Vh = ⋅ Ih . (30)
2 XL
1−h
XC
XL X
Here is very small making h 2 L << 1 ∴
XC XC
Vh ≤ hX L ⋅ I h (31)
V3 3 X L ⋅ I 3 V3
= , where THD is = 0.03 or 3% . Inductor impedance can be found by
V1 V1 V1
0.03 ⋅V1
XL = (32)
3* I 3
Let f s be defined as the switching frequency and f 1 be defined as the fundemental frequency.
fs X
Then for f s = 20kHz , f 1 = 60 Hz , and n = = 333.33 , L ≥ 2.09 x10 − 4 the filter resonant
f1 XC
frequency f r can be found with
fr XC n2
= ≤ ≤ 69 .17 . (33)
f1 XL 23 .22 2
fr ≈ 4150 Hz
28
The 10 KW inverter (5 KW per Phase) with V1 = 120V , produces I rms = 41.67 A , I 3 = 25.95 A . Use
XL
L= (34)
2π f1
Where
L - inductance
f1 - fundemental frequency
To find the capacitor impedance use the equation (28), to get X C = 221.26 , then using
1
C= (35)
2π f1 ⋅ X C
where
C - capacitance
f1 -fundemental frequency
Figure 14 below shows the DC input into to the DC-AC inverter and a single phase AC output of
the inverter stage of the 1.5kW prototype. With the DC input voltage at 390V, the voltage that appears
across the drain and source of the IGBT S1 is shown above. The ACRMS waveform shown is the
sinusoidal output voltage VAO after the filter stage. This voltage equal to 121 volts RMS, which meets
29
the required specification of 120 ± 6% volts. This output is realized with a modulation index (M) equal to
0.88. The measured frequency of the AC output is within the 60 ± 0.1 Hz requirement.
Figure 14: DC Input into to the DC-AC Inverter and a Single Phase AC Output
Figure 15 below shows the two PWM gating signals driving the IGBT of one phase of the
inverter. Channel 1 shows the PWM output from the DSP and Channel 2 shows the same PWM output
from an isolation opto-coupler. This image highlights clean PWM switching of the IGBT via the opto-
isolators controlled by the DSP. This clean switching translates to a cleaner output signal, which is
necessary to meet the THD < 5% specification. Please note that both Figure 14 and Figure 15 were
30
Figure 15: Two PWM Gating Signals Leading to One IGBT
31
3.6 Output: Monitoring and Computer Interface Via RS-232
The inverter uses an RS-232 cable to link the DSP’s serial port to a Windows PC. This interface
is used to transfer voltage and current values from the inverter to the PC so they can be displayed for real
time viewing. Data is also stored into text files for data processing. An overview is given in Figure 16
below.
Data is output from the DSP in a predefined transfer protocol that can be recognized by the PC.
In the DSP, the data is coded according to the protocol, and then transferred out of the SCI pin of the chip.
After the data reaches the other end of the cable, the software on the PC reads the correct serial port and
inputs the data into the computer’s software. The data is stored into a temporary 32 bit buffer. The
software reads the buffer and collects the data before the next piece of data is transferred. When new data
is transferred, it is stored in the same buffer as the old data, and the old data is destroyed. After the
software reads and stores the data, it must decode the data in order to know what it is and how it is to be
used. This is done with algorithms that are written based on the transfer protocol that is discussed in the
next section. When all of the needed data values are coded, transferred, read into the computer port, and
32
3.6.1 Transfer Protocol
The protocol that is used to transfer the data is designed to use as few data transfers as necessary,
but enough to accurately monitor the data. The data to be transferred is the Leg A voltage and current,
The voltages on Leg A and B will almost always be about 120 V, it was decided that the range of
measurable values that could be represented were from 0 to 180. Likewise, the currents of each leg will
almost always be between 0 and 12.5 A. It was decided to use a range of 0 to 55 to represent the current.
If physical values go beyond this range, only the upper values, (55 for current and 180 for voltage) will be
processed.
The RS-232 transfer protocol has a maximum data transfer of 8 bits per transfer. This does not
including the other start, stop, and parity bits. Within these 8 bits is the information that will be used to
identify the data as well as the actual value of the data. This can be more clearly seen in Table 3 below.
33
Using this method, each of the four values that are transferred has 10 total bits worth of data. The
10 bits are enough to approximate the actual current and voltage values. The formula’s used to code and
Coding
upper range = 180 for voltage values and 55 for current values
Decoding
Approximately every .5 to 1 second, the DSP will code all 4 data values and then break up the
coded values according to the table above. The extra bits that represent what the actual data is will also
be added to the 8-bit transfer block. Then all 8 of the 8 bit blocks will be sent to the computer to be
Once all of the data values are transferred and decoded, it is up to the software to display the data,
write the data to files, and perform all of the computations that are on the display. Given the four data
values that are transferred from the DSP, the display software also computes the maximum and average
values that were seen for the current, voltage, and power, and also computes the current power. This can
be seen more clearly in Figure 17 below which illustrates the TAMU Fuel Cell Inverter system under
non-operating conditions.
34
Figure 17: Display of RS-232
As the DSP keeps updating the display, the inverter’s operators can see real time data of its
performance. There are also limit arrows placed on the Current and Voltage data fields. These limits are
set to give an up or down arrow if the current value falls out of the range of the limits. Currently the
current limit is set to go off if the value goes above 12.5 A or equals zero, and the voltage limits will go
off if the voltage falls out of the range of 120 V plus or minus 6%. These limits can also be interpreted
and an overall status of the inverter given at the lower right of the display. The software has the ability to
periodically write data to output text files. These files can later be used to import into other data
processing programs. For example, data can be imported into Excel and the data plotted in graphs.
Test software had to be written, before the interface could be implemented. Software modules
were written using a Microsoft Visual C compiler. Software was written to simulate the DSP coding the
35
4. Cost Evaluation
4.1 Tracking Chart & Budget
The cost evaluation of the Texas A&M Fuel Cell Inverter considers two types of costs. First, the
cost of the development process and second the cost evaluation based on the normalized system cost
projection worksheet provided for by the 2001 Future Energy Committee. The development budget
helped guide the design team in organizing and prioritizing time and resources for optimum use. The
working budget allowed team members to make the best cost-design decisions by showing the team
actual costs of the specific components for the fuel cell inverter.
The development budget, in Table 4 on the next page, was prepared at the initiation of the Texas
A&M 2001 Future Energy Challenge team. The budget takes in consideration a team of nine
undergraduate students and three graduate students who designed the inverter system. Funds and supplies
were provided from the Department of Electrical Engineering at Texas A&M, Texas Instruments, 3M,
Toshiba, Lucent Technologies, and Reliant Energy. The chart outlines how time was spent. Hours have
been estimated.
With the practical experience gained by the working budget, the team’s industry partners and the
faculty advisors the team was able to make well-informed design decisions to aggressively lower the cost
of the final 10kW design and 1.5kW prototype. The TAMU fuel cell inverter team’s approach to reducing
the cost of the inverter by reducing the number a high cost switching devices by adopting push-pull
technology, using a low cost Motorola SG3525A PWM DC-DC controller and including an efficient DSP
36
Materials E s t i m a t e d L a b o r
Hours @ $35/hr
Labor
Research $0 800 $28,000
Design $0 5000 $175,000
Debugging $0 1900 $66,500
Prototype Testing $0 200 $7,000
Final Assembly $0 100 $3,500
Equipment
Computer $2,000 0 $0
Software $1,000 0 $0
DSP Bundle $3,990 0 $0
Books/References $1,000 0 $0
Parts
Inverter (DSP Design) $200 0 $0
DSP Chip $100 0 $0
Board $500 0 $0
Inverter Electronic Components $500 0 $0
Energy Source (Fuel Cell) $4,000 0 $0
Travel
Company Presentations $1,000 200 $7,000
Orlando, Florida $5,000 30 $1,050
West VA $10,000 70 $2,450
Miscellaneous
Materials $750 0 0
By use of the push–pull topology the number of MOSFETs was minimized to half needed by a
full bridge topology. IGBT’s were reduced in the inverter by use of the half bridge topology opposed to
the full bridge topology. The Motorola SG3525A PWM controller provided a low cost solution to control
of the DC-DC converter. It provides a single chip control solution opposed to complex discrete analog
hardware. DSP control of the DC-AC inverter provides efficiency of time and control. Readily
programmable, the DSP enables easy design changes to account for differing power applications.
Program capability translates into efficiency in human capital reducing costs of analysis, troubleshooting,
development and manufacturing of the fuel cell inverter. The use of the DSP allows a seamless interface
with other components of a power management system, saving integration time and human recourses. The
topology of the TAMU Fuel cell Inverter System employs a high voltage battery floating on the on the
DC-link. This approach does not add any additional power processing cost for sudden load management.
37
The cost for the power components of the TAMU Fuel Cell Inverter system were calculated by
developing the cost of the DC-DC converter and the DC-AC inverter and adding the two components
together. The Costs were based on the schematic and the quantity and power ratings take-off sheet shown
in Figure 18. The results of the cost analysis for the DC-DC converter are seen on the normalized
spreadsheet Table 5 and the results of the DC-AC costs are seen in Table 6.
The cost of the DC-DC converter was $290.51. The cost of the DC-AC inverter $206.79. The
total cost of the TAMU Fuel Cell System was $497.30. Following the cost analysis guild and spreadsheet
provided by the 2001 Future Energy Committee, the TAMU Fuel cell Inverter Team believes the power
processing components proposed by this design can be produced for less than the required $500.00
38
TR1
1:5 L1 I DC
+ ia
T1
Fuel Cell Input K1
48VDC Iin D1 D3 C2 Lb S1 S3 L2
N2 C4 3 A
Vin N1 Vbatt N
+ AC Output
C1 120/240V , 60 Hz
- VDC
N1 Lb C5 B
C3 3
N2 Vbatt ib N
D4 D2 L3 K2
- S2 S4
T2
Battery Backup
48VDC / 400VDC, 40KHz PUSH PULL CONVERTER 120V/240VAC, 20KHz PWM INVERTER
Note: Components shown in dotted boxes are not considered for cost evaluation
39
Component Designation Rating Quantity
40
4.3 DC-AC Inverter Costs
OTHER (EXPLAIN)
TOTAL 206.79
41
5. Demonstr ation of Oper ational success of the 1.5kW Pr ototype
In this section an overall summary of the preliminary results obtained for our 1.5kW inverter
prototype is presented.
DC-DC converter: Figure 3. shows the DC-DC converter topology. Figure 19 shows the test
results and the input & output waveforms. The 42VDC input from the high current DC power supply is
at 49V. Figure 19 shows the DC-DC converter output voltage Vd1 and Vd2 obtained during the
DC-AC Inverter: Figure 20 below shows the DC input into to the DC-AC inverter and a single
phase AC output of the inverter stage of the 1.5kW prototype. With the DC input voltage at 390V, the
42
voltage that appears across the drain and source of the IGBT S1 is shown above. The ACRMS waveform
shown is the sinusoidal output voltage VAO after the filter stage. This voltage equal to 121 volts RMS,
which meets the required specification of 120 ± 6% volts. This output is realized with a modulation index
(M) equal to 0.88. The measured frequency of the AC output is within the 60 ± 0.1 Hz requirement.
Figure 20: DC Input into to the DC-AC Inverter and a Single Phase AC Output
Figure 21 below shows the two PWM gating signals driving one IGBT of one phase of the
inverter. Channel 1 shows the PWM output from the DSP and Channel 2 shows the same PWM output
from an isolation opto-coupler. This image highlights clean PWM switching of the IGBT via the opto-
isolators controlled by the DSP. This clean switching translates to a cleaner output signal, which is
necessary to meet the THD < 5% specification. Please note that both Figure 20 and Figure 21 were
43
Figure 21: Two PWM Gating Signals Leading to One IGBT
manage sudden load changes at the inverter output. Figure18 shows the TAMU Fuel Cell inverter
integrated with battery backup system on the DC-link. The TAMU Fuel Cell inverter employs a ±192V
battery backup system connected to the DC-link. A string of sixteen (16), 12V, 1.2Ah YUASA sealed
rechargeable lead-acid batteries are employed to form +192V. Another string of sixteen batteries of the
same rating form –192V. The entire string of 32 batteries are connected to the 400V DC-link via two
inductors (Figure. X). The string of 32 batteries provides 460.8Wh capacity in the DC-link to support
load increases. The purpose of the inductors is to block the DC-AC inverter ripple current from flowing
into the battery circuit. Since the DC-DC converter stage regulates the inverter to ±200V, ±192V battery
44
bank will essentially float on the DC-bus. In the event of sudden load increase, the ±192V battery bank
Below are photographs of the experimental setup. Figure 22 is of the DC-DC boost converter. Figure 23
45
Figure 23: Single Phas e DC-AC inverter
The Texas A&M Fuel Cell Inverter was a combined effort of undergraduate, graduate, faculty
and industry personnel primarily organized by Texas A&M faculty advisors; Drs. Enjeti, Yeary, Howze
and Culp. The team approach to the design solutions was implemented over several semesters. A wide
range disciplines including electrical engineering, computer engineering, chemical engineering and
mechanical engineering were utilized. A substantial commitment from all members of the design team
was required and a thank-you is forwarded to all those who committed funding, resources and time.
46
6.1 Institutional Commitment and Sources of Added Support
The Texas A&M team was fortunate to have secured the sponsorship and commitment from a
variety of corporations for the project. The resources and technical direction from industry professionals
proved useful in determining the feasibility, manufacturability, and cost factors involved in the teams
decisions. Several companies such as Toshiba, Texas Instruments, Reliant Energy, 3M, were helpful in
their support.
The undergraduate students involved in the challenge were participants by virtue of enrollment in
the required electrical engineering senior design course at Texas A&M University. Students taking
electrical engineering directed studies course that involves research have also participated. The students
in the electrical engineering directed studies course earned three hours of course credit towards the their
undergraduate degrees. The students received an enriched experience of the engineering project process
including exposure to practical research, design, and manufacturing issues and techniques. The
opportunity to present their ideas and designs to industry was a valuable experience for all the students.
Undergraduate students worked in tandem with graduate students throughout the project. The
teams were lead by the graduate students under the supervision of the faculty advisors. The design
development teams are shown below in Figure 24 (the names of graduate level students are in Italics).
47
DSP/Inverter Team Integration Team:
Phillip Briggs Matt Campbell
David Leschberg Jared Machala
Matthew Webster Wes Weibel
Lori Dalton Cody Sicking
Oscar Montero Douglas Becker
Sansung Kim Rajesh Gopinath
48
A plan for the work over the duration of the project was written in the form of a Gantt chart.
Gantt Chart
Date Activity
September Formation of Fall Student Team (consisting of an Inverter and DSP Design Teams)
2000 Commence Industrial Presentations for company sponsorship
Commence Proposal Preparation
Begin Research and Planning Phase
49
7. Nomenclatur e
Iin - input current
vL - inductor voltage VATr -Transformer VA rating
L - inductance
C - capacitance
f1 - fundamental frequency
Pin - power input
Vin - voltage input
50
8. List of Acr onyms
AC - alternating current
DC - direct current
MW - megawatt
51
9. Bibliogr aphy
National Fuel Cell Research Center at the University of California, Irvine. Fuel Cell Technology Comes of Age.
Available: http://www.nfcrc.uci.edu/journal/article/fcarticleE.htm.
52
10. References
[1] Jiang, H.J., Qin, Y., Du, S.S., Yu, Z.Y., Choudhury, S., DSP based Implementation of a Digitally-
Controlled Single Phase PWM Inverter for UPS, Telecommunications Energy Conference, 1998,
INTELEC. Twentieth International, 1999, Page(s): 221 -224
[2] Abdel-Rahim, N., Quaicoe, J.E. Multiple feedback loop control strategy for single-phase voltage
source UPS inverter, Power Electronics Specialists Conference, PESC '94 Record., 25th Annual IEEE ,
1994 , Page(s): 958 -964 vol.2
[3] Abdel-Rahim, N.M., Quaicoe, J.E. Analysis and Design of a Multiple Feedback Loop Control
Strategy for Single-Phase Voltage-Source UPS Inverters, Power Electronics, IEEE Transactions on
Volume: 11 4 , July 1996 , Page(s): 532 -541
53
11. Appendices
54
Appendix A: Schematics for the TAMU Inverter
55
56
57
58
59
Appendix B: Schematics for DSP Control
60
61
62
63
Appendix C: DSP code (All .c and .h files)
evmgr2407.c
/*----------------------------------------------------------------------|
| |
| File: evmgr2407.c |
| Target Processor: TMS320LF2407 |
| Compiler Version: 6.6 |
| Assembler Version: 6.6 |
| Date: 11/2/00 |
| Programmer: SSKIM |
|---------------------------------------------------------------------|*/
#include "LF2407.h"
#define FREQIN4 ( 0 )
#define FREQIN2 ( ( PS0 ) )
#define FREQIN1_33 ( ( PS1 ) )
#define FREQIN1 ( ( PS1 ) | ( PS0 ) )
#define FREQIN_8 ( ( PS2) )
#define FREQIN_66 ( ( PS2) | ( PS0) )
#define FREQIN_57 ( ( PS2) | ( PS1) )
#define FREQIN_50 ( ( PS2) | ( PS1) | ( PS0) )
Continued
64
2001 Future Energy Challenge Competition
Thank You !
And our Sponsors:
2001 Future Energy Challenge
Competition
Undergraduate Student Team Members
Matt Campbell Andy Hale Cory Cress
Jon Burghardt David Leschber Cody Sicking
Phillip Briggs Wes Weibel Gary Tobola
Matthew Webster Nick Denniston David Payne
Jared Machala Matt Campbell David Leschber
Wes Weibel Nick Denniston Dao Le
Douglas Becker Justin Busse Randall Jones
Steven Campbell Lori Dalton Mike Spence
Mark Arldt
2001 Future Energy Challenge
Competition
Faculty Advisors
Dr. Prasad Enjeti
Email: enjeti@tamu.edu Tel: 979-845-7466
Dr. Mark Yeary
Dr. Jo Howze
Dr. Charles Culp
Development of a Low Cost Fuel Cell Inverter
with DSP Based Control
DC-DC
Converter DSP Control
DC – AC Board
Inverter
Inverter Performance
Single Phase on 500 W Resistive Load
THD < 1%
Vol. Reg.= ±1%