Micronote 1823 - Comparative Analysis of 150W RF VDMOS Transistors for 100MHZ Applications

Application Note 1823
Comparative Analysis of 150W RF VDMOS
Transistors for 100MHz Applications
Introduction
Four RF transistors, all rated for 150W and commonly used for industrial applications up to 100MHz,
are evaluated and compared, they are: BLF177 from NXP Semiconductors, SD2941-10 from
STMicroelectronics, and VRF151 and VRF152 from Microsemi Corporation Power Products Group
(MPPG).
The maximum ratings and key parameters of these transistors are first compared and discussed. For
the RF performance, all four transistors are load-pulled at 100MHz using a common test fixture as
well as the same load-pull system at MPPG to eliminate uncertainties due to various application
circuits and variable test conditions.
Through the current studies, we strive to demonstrate that devices from MPPG not only outperform
others, but that MPPG also offers higher voltage versions of otherwise very similar device,
i.e., VRF151, which is more versatile than others by operating the device at higher voltage around
60V to 65V for higher output power capabilities, or at 45V to 50V for improved ruggedness and
reliability at similar output power level as others.
Maximum Ratings of the RF VDMOS Transistors
Maximum ratings of the four RF transistors under study are listed in Table 1. The importance and
implication of these ratings are discussed in details next.
Table 1 · Maximum rating table for the 150W RF VDMOS transistors under study.
Maximum Rating Table
Symbols
V(BR)DSS
Unit
[V]
Description
Drain-Source Breakdown Voltage
VRF152 BLF177 SD2941-10
130
125
130
VRF151
170
ID,max
PD,max
[A]
[W]
Continuous Drain Current
Total Power Dissipation
20
300
16
220
20
389
16
300
Tj,max
VGS,max
[°C]
[V]
Junction Temperature
Gate-Source Voltage
200
±40
200
±20
200
±20
200
±40
Drain-Source Breakdown Voltage - V(BR)DSS
V(BR)DSS is primarily determined by the dopant concentration in the starting epitaxial wafer, onto which
the RF MOSFETs are fabricated. The performance of the transistor is strongly influenced by the
choice of the starting epitaxial wafer in three areas: 1) drain-source turn-on resistance RDS(on),
2) operating voltage VDD(op), and 3) ruggedness and long-term reliability of the device.
RDS(on) increases with V(BR)DSS, and decreases with increasing die size. Figure 1 depicts the typical
normalized relationship between RDS(on) and V(BR)DSS for MOSFETs. The effect of RDS(on) on RF
performance will be discussed in more details in the next section. In short, lower RDS(on) is desirable
for higher output power and efficiency.
The operating voltage VDD(op) is highly influenced by the circuit topology as well as the circuit design.
As a rule of thumb for Class-A, B, AB, and C operations, the operating voltage is generally kept
around one third of V(BR)DSS. Applying this empirical rule, the operating voltage for VRF152, BLF177,
and SD2941-10 is generally recommended around 45V-50V, while for VRF151, it can be operated at
55V-65V. Higher operating voltage enjoys the benefit of delivering more output power to a fixed load.
October, 2014
© 2014 Microsemi Corporation
1
Comparative Analysis of 150W MOSFETs for 100MHz Applications
Figure 1 · Typical relationship between RDS(on) and V(BR)DSS normalized to 500V.
Device ruggedness is a measure of its survivability against mismatched loads, which results in
elevated device junction temperature due to increased dissipation from reflected power. Although
ruggedness and survivability are general, albeit vague terms, it is well understood that the device
ruggedness can be improved by enhancing its thermal performance as well as lowering the operating
output power into a fixed load - typically 50Ω. Thermal performance of the four transistors under study
will be discussed in detail later.
In RF generator terminologies, the incident power is the power entering the load, of which some will
be reflected back into the generator and the rest will be delivered to the load. If the generator and
load are conjugate-matched, all the incident power will be delivered to such load. Since mismatched
loads are often specified as the percentage reflection of the incident power, lowering the operating
output power into the matched load effectively eases the severity of the mismatch specification, thus
improves its ruggedness. For example, operating VRF151 at 60V offers 44% more output power
(~216W) than operating the same device at 50V (~150W), therefore, for the same 33% reflection the
1
reflected power are 70W vs. 50W for VDD(op) = 60V and 50V, respectively .
Devices with higher V(BR)DSS should be inherently more reliable as well if operated at the same
operating voltage. This is due to the simple fact that the reverse breakdown of the MOSFET may
weaken, or even destroy the device, and higher V(BR)DSS allows more headroom against voltage
excursion caused by some uncontrollable events during operation. Therefore, VRF151 offers better
long-term reliability than the others when operated at the same 45V to 50V.
In summary, the four RF transistors under study can be categorized into two groups by their
recommended operating voltages: (A) VDD(op) = 45V-50V for VRF152, BLF177 and SD2941-10, and
(B) VDD(op) = 45V-65V for VRF151. RF MOSFET with lower V(BR)DSS typically comes with lower RDS(on),
while higher V(BR)DSS affords higher operating voltage as well as more rugged and reliable than
devices with lower V(BR)DSS.
Maximum Continuous Drain Current - ID,max
All four RF transistors under study belong to a special class of MOSFET called vertical, doublydiffused MOS, or VDMOS. The maximum current rating of a VDMOS transistor, ID,max, is limited
mostly by die size and RDS(on), and to a lesser extent, influenced by the device structure, layout design
and fabrication processes. In fact, RDS(on) and die size are interdependent as RDS(on) goes down with
increasing die sizes. Since ID,max is determined by running the MOSFET continuously, the self-heating
effect is unavoidable such that the strong dependency of RDS(on) vs. temperature has to be taken into
consideration. As shown in Figure 2, there is about 20% increase in RDS(on) for every 25°C
temperature rise. If we neglect all the other factors except RDS(on) and PD,max, ID,max can be estimated
by running the transistor at rated power dissipation up to its rated temperature as
(1)
1
I D,max =
PD,max
R DS(on) (Tj = 200o C)
Ruggedness specifications are highly application dependent, examples of commonly adopted specifications are 1) withstand
33% reflection at full rated (incident) power for 10 minutes over all phase angles, and/or 2) withstand total (99%) reflection at 33%
of rated (incident) power for 10 minutes over all phase angles. Note that the rated incident power becomes the rated output power
into a matched load, and the phase angle specifies inductive or capacitive load conditions.
2
Maximum Ratings of the RF VDMOS Transistors
where RDS(on) at Tj = 200°C can be estimated from Figure 2, i.e., RDS(on) increases 3.6 times from 25°C
to 200°C. Table 2 lists both the estimated and reported ID,max values for the four transistors under
study, which shows that Equation (1) is fairly accurate in predicting ID,max for MPPG devices since the
temperature scaling factor of 3.6 is specific to MPPG's VDMOS technologies, while it understates
BLF177's and overstates SD2941-10's temperature scaling factors. In fact, the temperature scaling
factors can be estimated from the reported ID,max values for BLF177 and SD2941-10, respectively, as
16% and 25% increase in RDS(on) for every 25°C temperature rise, or factors of 2.9 and 4.9,
respectively, from 25°C to 200°C.
Figure 2 · Typical normalized relationship between RDS(on) and junction temperature Tj.
Table 2 · Reported and estimated ID,max by Equation (1).
Unit
VRF152
BLF177
SD2941-10
VRF151
PD,max (from Table 1)
RDS(on) @25°C (from Table 4)
Parameters
[W]
[Ω]
300
0.2
220
0.3
389
0.2
300
0.3
RDS(on) @200°C
ID,max, Estimated
[Ω]
[A]
0.72
20.4
1.08
14.3
0.72
23.2
1.08
16.7
ID,max, Reported
[A]
20
16
20
16
It is interesting to note that A) The difference in ID,max between VRF152 and VRF151 is mostly due to
RDS(on), which can be attributed to the difference in breakdown voltages, V(BR)DSS; and B) Although
VRF152 and BLF177 have similar V(BR)DSS, the fact that BLF177 has lower ID,max due to higher RDS(on)
can be explained by its smaller die size, which will be discussed in more detail in the next section.
Total Power Dissipation - PD,max
Total power dissipation is derived from two other parameters, they are: the maximum junction
temperature Tj,max and the junction-to-case thermal resistance of the device RθJC. Conventionally, the
package case temperature of 25°C is assumed in such calculation. By definition, the total power
dissipation can be expressed as
(2)
PD,max = (Tj,max − 25o C)/R θJC
The estimated and reported PD,max for the four transistors under study are listed in Table 3 confirming
that Equation (2) is the formula used in deriving the power rating. In short, all four transistors have
ample thermal capabilities to dissipate partially or totally reflected power at 150W.
3
Comparative Analysis of 150W MOSFETs for 100MHz Applications
2
It should be noted that SD2941-10 uses a more expensive version of the SOE package, i.e., the
thermally enhanced M174 package, which is available to the other two manufacturers as well, who
have deemed it unnecessary to further enhance the thermal capabilities of the device with the added
costs. Finally, the 26% higher thermal capability of VRF152 compared to BLF177 can be attributed to
the die size difference between these two devices, which will be discussed in the next section.
Table 3 · Reported and estimated PD,max by Equation (2).
Parameters
Tj,max (from Table 1)
Unit
[°C]
VRF152
200
BLF177
200
SD2941-10
200
VRF151
200
RθJC (from Table 4)
PD,max, Estimated
[°C/W]
[W]
0.6
292
0.8
219
0.45
389
0.6
292
[W]
300
220
389
300
PD,max, Reported
Maximum Gate-Source Voltage - VGS,max
VGS,max is a measure of the dielectric strength of the gate oxide against large gate voltage excursions.
Higher VGS,max rating implies more rugged gate oxide construction. It is no accident that devices from
MPPG are rated for the highest gate-source voltage in both polarities since MPPG's core strength is
in high-voltage VDMOS device design and construction. In fact, MPPG is the only supplier among the
three manufactures of VDMOS transistors, who also offers RF MOSFETs with 500V to 1200V
breakdown voltages in its ARF family of transistor products.
It should be noted that it is not uncommon for a VDMOS transistor to experience a large voltage
excursion at its gate during operation due to the feedback coupling capacitance, Crss, which will be
discussed in the next section. Therefore, higher VGS,max is desirable for additional headroom against
such voltage excursions.
Key Parameters of the RF VDMOS Transistors
As part of the current analysis, transistor samples are first de-capsulated, and their die sizes are
measured as shown in Figure 3. Since the die sizes and wire-bonding schemes of VRF151 and
VRF152 are identical, only VRF152 is de-capsulated and shown. As seen from Figure 3, BLF177 has
the smallest die area, followed by SD2941-10 and VRF151/VRF152. Also interesting to note is that
BLF177 employs two MOSFET dies, while a single die is used in both SD2941-10 and
VRF151/VRF152, respectively.
Table 3 lists the key parameters of the four RF transistors under study along with test conditions,
under which these parameters were measured. It should be noted that common test conditions are
listed in Table 3 unless they are tested under different conditions. Finally, only NXP reports RDS(on)
directly on the datasheet, while VDS(on) is reported by MPPG and STMicroelectronics, which are then
converted to RDS(on) by taking the ratio of VDS(on) and the fixed sensing drain current ID.
Gate-Source Threshold Voltage - VGS(th)
VGS(th) and RDS(on) are two of the most important parameters to consider if multiple MOSFETs are
paralleled in an application for higher output power capacities. As shown in Figure 4, the gate-source
threshold voltage of these RF MOSFETs decreases with increasing temperature, or a negative
temperature coefficient, below the crossover point. If a few of these MOSFETs are paralleled in the
application, the negative temperature coefficient promotes thermal runaway, i.e., a hotter device
would be turned on more easily than others resulting in uncontrollable self-destruction.
2
Stripline Opposed Emitter (SOE) Package - Originally developed by Motorola for bipolar transistors, SOE packages provides low
inductance leads, which interfaces well with microstrip lines on circuit boards or substrates. The two emitter leads provide good
collector-to-base isolation, as well as promote symmetry in board layout when combining devices to obtain higher output power.
This package is also widely adopted for power MOSFETs with gate and drain leads in the place of base and collector leads,
respectively, and two source leads in the place of emitter leads. There are two SOE package types, namely, stud-mounted and
flange-mounted packages. Only the flange-mounted SOE packages are discussed in this Application Note.
4
Key Parameters of the RF VDMOS Transistors
Figure 3 · De-capsulated RF VDMOS transistors under study with measured die sizes.
Table 4 · Key parameters table of the 150W RF VDMOS transistors under study.
Key Parameters Table
Sym
A
VGS(th)
RDS(on)
gfs
Unit
Description
2
[mm ] Die Size
[V]
[Ω]
[S]
Gate-Source Threshold
Voltage
Drain-Source On-State
Resistance
VRF152 BLF177 SD2941-10 VRF151
Typ.
20.30
17.40
17.86
20.30
Min.
2.9
2.0
1.5
2.9
Typ.
3.6
Max.
4.4
4.5
Typ.
0.13
0.20
Max.
0.20
0.30
0.20
0.30
Min.
5.0
4.5
5.0
5.0
Typ.
6.2
6.2
6.0
3.6
4.0
Test Conditions
MSC: VDS = 10V; ID = 100mA.
NXP: VDS = 10V; ID = 50mA.
ST: VDS = 10V; ID = 250mA.
4.4
0.20
Forward Transconductance
MSC/ST: VGS = 10V; ID = 10A.
NXP: VGS = 10V; ID = 5A.
VDS = 10V; ID = 5A.
IDSS
[mA]
Drain-Source Leakage
Current
Max.
0.05
2.5
0.05
1.0
MSC: VGS = 0V; VDS = 100V.
NXP/ST: VGS = 0V; VDS = 50V.
IGSS
[µA]
Gate-Source Leakage
Current
Max.
1.0
1.0
0.25
1.0
MSC/NXP: VDS = 0V; VGS = ±20V.
ST: VDS = 0V; VGS = 20V.
Ciss
[pF] Input Capacitance
Typ.
383
480
415
375
VGS = 0V; VDS = 50V; f = 1MHz.
Coss
[pF] Output Capacitance
Typ.
215
190
236
200
VGS = 0V; VDS = 50V; f = 1MHz.
Crss
[pF]
Reverse Transfer
Capacitance
Typ.
20
14
17
12
VGS = 0V; VDS = 50V; f = 1MHz.
Junction to Case Thermal
Resistance
Max.
0.6
0.8
0.45
0.6
Typ.
1.8
1.3
1.6
1.9
RθJC [°C/W]
fT
GHz Cutoff Frequency
VDS=10V (Ciss' are scaled to 10V)
5
Comparative Analysis of 150W MOSFETs for 100MHz Applications
Figure 4 · Temperature dependency of the transfer characteristics.
It is desirable to have the crossover point at a lower drain current for a wider range of thermal stability
in paralleled configurations. Among the four RF transistors under study, VRF151 and VRF152 have
the lowest crossover point at about 8A compared to 10A for SD2941-10. Since only temperature
coefficient is reported on BLF177's datasheet up to about 2.5A, it is too far away for accurate
extrapolation to the point of vanishing temperature coefficient.
All four RF transistors can be ordered in matched VGS(th) groups, or bins, and their grouping and
marking information are provided in Table 4. It is seen that VRF151 and VRF152 have the narrowest
range of VGS(th) and the best group spacing, they are: 1.5V and 0.075V, respectively, while for BLF177
and SD2941-10, are 2.5V and 0.1V, respectively. Again, the narrower range and finer grouping of
VGS(th) are both highly desirable when transistors are paralleled in the application.
Table 5 · VGS(th) grouping and marking codes for the RF VDMOS transistors under study.
VGS(th) Binning Group/Code
VRF151 / VRF152
(20 Bins, 0.075V Spacing)
6
BLF177
(25 Bins, 0.1V Spacing)
Bin
Range [V]
Bin
Range [V]
Bin Range [V] Bin
A
2.900 2.975
R
3.875
3.950
A
2.00 2.10
B
2.975 3.050
S
3.950
4.025
B
C
3.050 3.125
T
4.025
4.100
C
D
3.125 3.200
W
4.100
4.175
E
3.200 3.275
X
4.175
4.250
SD2941-10
(25 Bins, 0.1V Spacing)
Range [V]
Bin
Range [V]
Bin
Range [V]
O
3.30
3.40
AA
1.50
1.60
J
2.80
2.90
2.10 2.20
P
3.40
3.50
BB
1.60
1.70
K
2.90
3.00
2.20 2.30
Q
3.50
3.60
CC
1.70
1.80
L
3.00
3.10
D
2.30 2.40
R
3.60
3.70
DD
1.80
1.90
M
3.10
3.20
E
2.40 2.50
S
3.70
3.80
EE
1.90
2.00
N
3.20
3.30
F
3.275 3.350
Y
4.250
4.325
F
2.50 2.60
T
3.80
3.90
A
2.00
2.10
P
3.30
3.40
G
3.350 3.425
Z
4.325
4.400
G
2.60 2.70
U
3.90
4.00
B
2.10
2.20
Q
3.40
3.50
H
3.425 3.500
H
2.70 2.80
V
4.00
4.10
C
2.20
2.30
R
3.50
3.60
J
3.500 3.575
J
2.80 2.90
W
4.10
4.20
D
2.30
2.40
S
3.60
3.70
K
3.575 3.650
K
2.90 3.00
X
4.20
4.30
E
2.40
2.50
T
3.70
3.80
M
3.650 3.725
L
3.00 3.10
Y
4.30
4.40
F
2.50
2.60
U
3.80
3.90
N
3.725 3.800
M
3.10 3.20
Z
4.40
4.50
G
2.60
2.70
V
3.90
4.00
P
3.800 3.875
N
3.20 3.30
H
2.70
2.80
Key Parameters of the RF VDMOS Transistors
Drain-Source On-State Resistance - RDS(on)
It has been discussed in the last section that RDS(on) increases with V(BR)DSS and temperature, see
Figure 1 and Figure 2, respectively; and decreases with die size. The relationship between RDS(on)
and die size can also be influenced by the design as well as fabrication processes of the device. The
RDS(on)-V(BR)DSS trend can be seen clearly between VRF152 and VRF151 with different breakdown
voltages. The lower RDS(on) of VRF152 compared to BLF177 reflects its bigger die size, while no such
trend is observed between VRF152 and SD2941-10.
Discussed in the last section, RDS(on) is the determining factor for the current capacity of the device.
As predicated by Equation (1), VRF152 and SD2941-10 both have higher current capability of 20A,
while VRF151 and BLF177 are rated 4A lower due to their higher RDS(on).
Finally, It is interesting to note that VRF151 has the same RDS(on) as BLF177, while VRF151's V(BR)DSS
is much higher than BLF177. Taking VRF151's lower RθJC into consideration, which will be discussed
shortly, it implies that VRF151 can be operated at higher drain voltages for higher output power
without compromising the long-term reliability of the device.
Forward Transconductance - gfs
The forward transconductance, gfs, is a static parameter determined by measuring the rate of change
in drain current in response to the change in gate voltage at fixed drain-source voltage and drain
current. It can be shown that there is a positive correlation between gfs and the RF gain. Since RF
gain depends on the application circuit as well as test conditions, it will be discussed in more detail
when load-pull results are presented in the next section. For practical purposes, the forward
transconductance of the four RF VDMOS transistors under study are very similar with the exception
of BLF177, whose minimum gfs, is 10% lower than the other three.
Leakage Currents - IGSS & IDSS
The gate-source leakage current IGSS, measured in µA, are all very small for 150W power transistors,
and the slightly smaller value for SD2941-10 really offers no measurable benefit in typical
applications.
The differences in drain-source leakage currents, measured in mA, on the other hand, are not so
subtle with BLF177 being 50 times more leaky than VRF152 and SD2941-10. More importantly,
recognizing that drain-source leakage current grows exponentially with drain-source voltage, the IDSS
of VRF152 is measured at a VDS of 100V, while both BLF177 and SD2941-10 are measured at 50V,
therefore, VRF152 has the lowest IDSS among the three RF transistors with similar V(BR)DSS, and
BLF177 is about two orders of magnitude more leaky than VRF152.
Input, Output and Reverse Transfer Capacitances - Ciss, Coss & Crss
As discussed in Appendix A, the capacitances of a VDMOS device can be represented by: (A) the
terminal capacitances, Cgs, Cds and Cgd commonly used in circuit analysis; (B) the measurable
capacitances, Ciss, Coss and Crss; or (C) the physical capacitances, Cox1, Cox2, Cdrift and Cbody. The
terminal and physical capacitances are illustrated in Figure 5.
7
Comparative Analysis of 150W MOSFETs for 100MHz Applications
Figure 5 · Terminal and physical capacitances of a VDMOS device.
The terminal capacitances can be expressed in terms of measurable capacitances as well as physical
capacitances such as
(3)
Cgs = Ciss − Crss = Cox1
(4)
Cds = Coss − Crss = Cbody
(5)
Cgd = Crss
 1
1
= 
+
 Cox2 Cdrift



−1
The gate-source capacitance, Cgs, is the capacitance of the gate oxide (ox1), which needs to be
charged to turn on the MOSFET. Cgs can be readily obtained by subtracting Crss from Ciss. The drainsource capacitance, Cds, is the junction capacitance of the body diode, whose reverse breakdown
voltage determines the drain-source breakdown voltage V(BR)DSS of the MOSFET. It should be noted
that Cgs is relatively independent of the drain-source bias voltage VDS, while Cds strongly depends on
VDS. Similarly, Cds is readily calculated by subtracting Crss from Coss. In amplifier operation, Cds is seen
as the reactance of its output impedance.
The gate-drain capacitance, Cgd, is the same capacitance as Crss as measured, which consists of the
stacked capacitances of the gate oxide (ox2) and the capacitance due to the space charge in the drift
region under bias conditions. The gate-drain capacitance, Cgd, or the reverse transfer capacitance,
Crss, plays dual roles in the operation of the MOSFET: (A) it is the augmented "Miller" capacitance
during turn-on of the MOSFET; and (B) it is responsible for the feedback or coupling between the
input and output terminals of the MOSFET in a common-source configuration.
As shown in Figure 6, there are three distinctive segments in the charging curve of a typical
MOSFET. The corresponding transients of gate-source voltage, drain-source voltage and current are
also illustrated in the Figure 6. At t=0, the drain is already positively biased, and a constant gate
voltage is suddenly connected to the gate with common source ground. Reacting to the applied gate
signal, the gate-source voltage, VGS, rises at a constant rate until t2. Within this period, VGS first
passes the threshold voltage VGS(th), hereafter the drain-source current increases rapidly from t1 to t2,
while the drain-source voltage VDS drops only slightly during the same period. During t2, both Cgs and
Cgd are charged to the total amount of Qgs until the so-called "Miller" plateau is reached at t2. Since
Cgs is typically much larger than Cgd, most of charge goes to Cgs before t2.
8
Key Parameters of the RF VDMOS Transistors
Figure 6 · Transient behavior and Miller effect during turn-on of MOSFET.
The Miller effect simply states that the coupling impedance between the input and output of a voltage
amplifier is augmented by the gain of itself. As seen in Figure 6, the gain of the MOSFET reaches its
peak at t2 when IDS reaches its saturation point. At this point, Cgd begins its rapid ascension due to
the Miller effect. It is this rapid rate of changing capacitance that results in a sudden increase in the
observed capacitance at the gate, which is reflected by the gentler slope in the charging curve since
the capacitance is inversely proportional to the slope of the charging curve. The partition of charge
also reverses during this period since it is the rate of changing capacitance that dictates the draw of
charge such that most charge is injected into Cgd.
The Miller plateau ends when the drain-source voltage reaches its on-state value, which is set by the
product of IDS and the on resistance of the MOSFET RDS(on). Beyond t3, the gain of the MOSFET
saturates and no more increase of Cgd is possible such that the total capacitance seen at the gate is
again the sum of two constant capacitances Cgs and Cgd, albeit a larger Cgd than before, thus a less
steep slope.
In amplifier operation, the coupling between input and output is highly undesirable since it serves as a
negative feedback resulting in less gain of the device, or "making it harder to drive", i.e., more input
power and current are required for the same output power.
In summary, the desirable attributes of RF MOSFET capacitances include: (A) low Ciss (Cgs+Cgd) for
fast turn-on, (B) low Crss (Cgd) for less pronounced Miller effect as well as less negative feedback, and
(C) low Cds (Coss-Crss) for smaller size of output matching inductor or transformer.
The measurable capacitances of the four RF VDMOS transistors under study are plotted against the
drain-source voltage in the same scale for clear comparison in Figures 7 and 8. Ranking of the
measurable capacitances from the lowest to the highest values are: (A) Ciss - VRF151, VRF152,
SD2941-10, and BLF177, (B) Coss - BLF177, VRF151, VRF152, and SD2941-10, C) Crss - VRF151,
BLF177, SD2941-10, and VRF152.
The most significant differences among these four RF VDMOS transistors are: A) BLF177's Ciss is the
worst by a large margin; B) To a less extent, SD2941-10 has the largest Coss and VRF152 has the
largest Crss, and C) VRF151 ranks first in both Ciss and Crss and second in Coss.
9
Comparative Analysis of 150W MOSFETs for 100MHz Applications
Figure 7 · Comparison of measurable capacitances between VRF152, BLF177 and VRF151.
Figure 8 · Comparison of measurable capacitances between VRF152, SD2941-10 and VRF151.
Junction-to-Case Thermal Resistance - RθJC
The junction-to-case thermal resistance is the limiting factor for the maximum power dissipation of the
device, and is mostly determined by the properties of the material stack including the MOSFET and
packaging materials. More specifically, the junction-to-case thermal resistance is the sum of the
thermal resistance of each layer in the material stack such as
(6)
R θJC = ∑
i
di
A ik i
th
where di, Ai and ki are, respectively, the thickness, area and thermal conductivity of the i layer in the
material stack.
10
RF Performance of the VDMOS Transistors
The same M174 package is adopted in VRF151, VRF152 and BLF177, while a thermally enhanced
version of M174 is used for SD2941-10. As seen in Table 4, BLF177's RθJC is 33% larger than
VRF151 and VRF152, which is mostly due to the difference in die sizes. SD2941-10 has a similar die
size as BLF177, whose smaller thermal resistance is mostly due to the thermally enhanced, and a
more expensive version of the M174 package. In fact, MPPG used to offer VRF151E, which adopts
the same thermally enhanced M174 package. VRF151E has been obsoleted from MPPG's RF
product offering since the power dissipation of the regular VRF151 is more than adequate in most
applications, and the added cost could not be justified.
Transistor Cutoff Frequency - fT
The cutoff frequency of a MOSFET is defined as the frequency at which its current gain becomes
unity, or 0dB, and can be expressed as
(7)
fT =
g fs
g fs
=
2π (Cgs + Cgd ) 2πCiss
Intuitively, (A) a higher starting gain at lower frequency is one of the prevailing conditions for higher
fT, and the forward transconductance gfs is closely related to current gain at DC, and (B) the input
capacitance Ciss limits fT since it takes longer to charge up a larger Ciss to turn on the MOSFET. From
Table 4, it is seen that VRF151 and VRF152 have the highest fT at 1.9GHz and 1.8GHz, respectively,
and BLF177's fT is the lowest at 1.3GHz.
RF Performance of the VDMOS Transistors
The RF performance of the four VDMOS transistors under study are evaluated and compared using
the same load-pull system in a common water-cooled test fixture as shown in Figure 9. This load-pull
system consists of a source tuner (LFT-013006 from Focus Microwave, 60-130MHz) and a load tuner
(CCMT-101 also from Focus Microwave, 0.1-1.1GHz), which are both automatically controlled by
software supplied by Focus Microwave. The load-pull test fixture comprises a socket for the M174
package, biasing circuit, quarter-wave transmission-line pre-matching circuits at both input (gate) and
output (drain) terminals of the transistor, and N-type input and output connectors. The text fixture
PCB is directly mounted on a copper heat-sink with forced cooling by a chiller. A thin layer of thermal
grease is first applied onto the heat-sink surface in the M174 socket, and the DUT is then mounted
onto the socket by mounting screws. Before the insertion of a new DUT, the residual grease in the
socket is thoroughly wiped clean, and the grease application and DUT mounting are carefully carried
out for consistency. The quarter-wave pre-match circuits are used to increase the impedance at the
tuner for better accuracy, and their circuit contribution is de-embedded during the calibration of the
load-pull system.
Figure 9 · The load-pull system and test fixture used in assessing the RF performance of the
four RF VDMOS transistors under study.
11
Comparative Analysis of 150W MOSFETs for 100MHz Applications
In a fully automated load-pull system, the source and load tuners can be adjusted to search for
impedance states that meet the preset load-pull objective. The load-pull objective can be a simple
one such as the optimal states for (A) output power, (B) gain, (C) efficiency, or (D) linearity. If the
load-pull objective is set loosely, e.g., setting output power below the maximum saturated power,
multiple impedance states can be found, which forms a contour on the input and output Smith Charts.
On the other hand, no impedance state will be found if the objective is set beyond the capability of the
device. By setting the load-pull objective closer and closer to the capability of the device, these loadpull contours shrink to a single point, which represents the optimal impedance states for such
objective. Composite load-pull object is also possible by assigning weights to the four simple
objectives above. Finally, the source and load tuners are adjusted separately and iteratively with one
tuner at a fixed impedance when the other is being adjusted until the desired impedance states are
found.
The First Load-Pull Experiment - Po,max & PAE > 65%
In the first load-pull experiment, the load-pull objective is set for optimal output power with a minimum
power added efficiency (PAE) of 65%. Two units of each of the four transistors under study are loadpulled at 100MHz in pulsed mode with 50% duty cycle, and cooled by water at 25°C. It should be
noted that the impedance states are conventionally reported in terms of the input impedance of the
transistor and load impedance presented to the transistor during load-pull. Table 6 lists the load-pull
results of the first experiment for each of the two units of the four RF VDMOS transistors under study.
Also listed in Table 6 are the average impedances of the two units of the four RF VDMOS transistors.
Table 6 · Optimal impedance states of the first load-pull experiment.
Po,max/PAE > 65%
VRF151
VRF152
BLF177
SD2941-10
Zin
ZL
DUT1
4.05 - j6.11
1.76 + j3.80
DUT2
4.05 - j6.11
2.08 + j3.68
AVE
4.05 - j6.11
1.92 + j3.74
DUT1
2.86 - j6.71
2.13 + j3.93
DUT2
2.86 - j6.71
2.13 + j3.93
AVE
2.86 - j6.71
2.13 + j3.93
DUT1
3.62 - j5.02
2.13 + j3.93
DUT2
3.62 - j5.00
2.13 + j4.00
AVE
3.62 - j5.00
2.13 + j3.97
DUT1
4.14 - j6.07
2.23 + j3.47
DUT2
4.14 - j6.07
2.23 + j3.47
AVE
4.14 - j6.07
2.23 + j3.47
Power sweep experiments are also performed for each of the eight transistors at their respective
optimal impedance states as listed in Table 6. Figures 10-12 show, respectively, the results of power
sweep, gain vs. output power, and PAE vs. output power between VRF151/VRF152 and
BLF177/SD2941-10. Key observations from the first load-pull experiment are in order:
(1) The RF performance of VRF152 and BLF177 closely resemble each other with noticeable
difference in input impedance Zin and slightly higher and flatter gain of VRF152 than BLF177.
(2) The RF performance of VRF151 and BLF177 closely resemble each other as well with
noticeable difference in input impedance Zin, slight difference in load impedance ZL, and
slightly higher gain and PAE of BLF177 than VRF151.
(3) The gain of SD-2941-10 is significantly lower as well as least flat among the four VDMOS
transistors under study.
The variations in the resulting optimal load-pull impedance between two units of the same transistor is
most significant in the load impedance of VRF151, however, the variations in power sweep results
between the two units of the same transistor with similar optimal load-pull impedances are almost as
significant as the former.
12
RF Performance of the VDMOS Transistors
Figure 10 · Power sweep results of VRF151, VRF151, BLF177 and SD2941-10.
Figure 11 · Gain vs. output power results of VRF151, VRF152, BLF177 and SD2941-10.
13
Comparative Analysis of 150W MOSFETs for 100MHz Applications
Figure 12 · PAE vs. ouptut power results of VRF151, VRF152, BLF177, and SD2941-10.
From the first load-pull experiment, it is seen that VRF151, VRF152 and BLF177 are interchangeable
in a given RF power generator design since the load impedances are very close. The input
impedances are noticeably different among these three transistors, which may impact the stability
performance if one transistor is dropped into the circuit designed and optimized for another.
Therefore, it is recommended that the input matching circuit be modified when replacing an
incumbent transistor with another. The best way to assess the required circuit modification for a dropin replacement is to examine the input and output matching circuits from de facto 50Ω standard to the
input and output impedances of the transistors of interest.
Adopting the most simplistic L-network matching circuit, Figure 13 shows the input matching circuit
design for the four transistors under consideration using the ADS simulator from Keysight (Agilent)
Technologies, and Figure 14 shows the frequency responses of the input matching circuits. It should
be noted that the quality (Q) factor for the inductors and capacitors used in the circuit simulation are,
respectively, 50 and 250.
The input matching circuit parameters along with the insertion loss of the matching circuit at 100MHz
are listed in Table 7. It is clear that the same input matching topology can be used for all the four
transistors, while the optimal circuit parameters differ by 10% in inductance and 21% in capacitance.
The output matching circuits for all four transistors are also analyzed similarly using the L-network as
the matching circuit. Figure 15 shows the output matching circuit design using the ADS simulator, and
Figure 16 shows the frequency responses of the output matching circuits. Under the same matching
topology, the output matching circuit parameters along with the insertion loss of the matching circuit
at 100MHz are listed in Table 8. These output circuit matching parameters only differ by about 6% to
7% among the four transistors reflecting the fact that the optimal load impedances from the first loadpull experiment are very close among the four transistor under study.
14
RF Performance of the VDMOS Transistors
Figure 13 · Input matching circuit design for the first load-pull experiment using ADS
simulator
Figure 14 · Frequency responses of the input matching circuits for the first load-pull
experiment.
Table 7 · Input matching circuit parameters and insertion loss at 100MHz for the first load-pull experiment.
Pmax/PAE > 65%,
Input Matching
Transistor
Zin
L (nH)
C (pF)
IL (dB)
VRF151
4.05 - j6.11
31.5
105.0
0.47
VRF152
2.86 - j6.71
29.4
125.6
0.60
BLF177
3.62 - j5.00
28.6
111.7
0.48
SD2941-10
4.14 - j6.07
31.6
103.7
0.46
15
Comparative Analysis of 150W MOSFETs for 100MHz Applications
Figure 15 · Output matching circuit design for the first load-pull experiment using ADS
simulator.
Figure 16 · Frequency responses of the output matching circuits for the first load-pull
experiment.
Table 8 · Output matching circuit parameters and insertion loss at 100MHz for the first load-pull experiment.
Pmax/PAE > 65%,
16
Output Matching
Transistor
ZL
L (nH)
C (pF)
IL (dB)
VRF151
1.92 + j3.74
21.4
155.6
0.66
VRF152
2.13 + j3.93
22.5
147.5
0.63
BLF177
2.13 + j3.97
22.6
147.5
0.62
SD2941-10
2.23 + j3.47
22.1
144.5
0.59
RF Performance of the VDMOS Transistors
The Second Load-Pull Experiment - 100% Po,max
To further assess the RF performance between VRF151, VRF152 and BLF177, a second load-pull
experiment designed to search for the maximum output power at 100MHz in a pulsed mode operation
with 20% duty cycle under active cooling by chilled water. It should be noted that SD2941-10 is
dropped from this study due to its inferior RF performance.
The results of the second load-pull experiment are listed in Table 9. Unlike the first load-pull
experiment, where only input impedances shows noticeable differences, variations in both input and
load impedances are observed in the second load-pull experiment. The power sweep results of the
three transistors at their respective optimal impedance states as listed in Table 9 are plotted in Figure
17. Again, the RF performance of these three transistors is fairly comparable with some noticeable
differences in:
(A). The gain of BLF177 is slightly higher ~ 0.2 dB higher than VRF151 and 0.5 dB higher than
VRF152,
(B). The PAE of VRF152 is the highest, which in part is due to its lower RDS(on) than the other two
transistors.
The 1-dB compression point of VRF152 is the lowest, which implies the worst linearity performance
among the three transistors.
Table 9 · Optimal impedance states of the second load-pull experiment.
100% Pout
Zin
ZL
VRF151
DUT1
2.54 - j3.45
3.28 + j1.78
VRF152
DUT1
2.45 - j3.42
2.36 + j2.07
BLF177
DUT1
3.67 - j3.72
3.05 + j1.75
Figure 17 · Power sweep results of the second load-pull experiment.
17
Comparative Analysis of 150W MOSFETs for 100MHz Applications
Using the same L-network as the input/output matching circuits, the optimal impedances of the
second load-pull experiment are also matched to the de facto 50Ω standard. The matching circuit
parameter along with the insertion loss at 100MHz are listed in Tables 10 and 11, which indicates that
a simple value change of matching components suffices to preserve the optimal performance of any
of these three transistors when dropped into the circuit designed and optimized for the other.
Table 10 · Input matching circuit parameters and insertion loss at 100MHz for the second load-pull experiment.
100% Poptm ,
Input Matching
Transistor
Zin
L (nH)
C (pF)
IL (dB)
VRF151
2.54 - j3.45
23.0
135.1
0.55
VRF152
2.45 - j3.42
22.7
137.7
0.56
BLF177
3.67 - j3.72
26.6
111.3
0.44
Table 11 · Output matching circuit parameters and insertion loss at 100MHz for the second load-pull
experiment.
100% Poptm ,
Output Matching
Transistor
ZL
L (nH)
C (pF)
IL (dB)
VRF151
3.28 + j1.78
22.4
118.9
0.43
VRF152
2.36 + j2.07
20.2
141.2
0.53
BLF177
3.05 + j1.75
21.8
123.6
0.45
It is also of interest to assess the RF performance if either VRF151 or VRF152 is dropped into the
circuit designed and optimized for BLF177 without any circuit modification. Table 12 and 13 list the
results of such exercises by presenting the optimal impedances for BLF177 to VRF151 and VRF152,
respectively, and adjusting the input drive until the output power reaches the rated power of 150W.
Table 12 · Results of drop-in of VRF151 into the circuit optimized for BLF177 in the second load-pull
experiment.
Device
Zin
ZL
Po (dBm)
(~150W)
PAE (%)
Gain (dB)
P1dB
(dBm)
BLF177
3.66 - j3.71
3.05 + j1.75
51.75
53.10
18.93
53.26
VRF151
2.54 - j3.45
3.28 + j1.78
51.70
55.23
19.64
53.70
VRF151 w BLF177 Zoptm 3.66 - j3.71
3.05 + j1.75
51.84
57.89
19.47
n/a
Table 13 · Results of drop-in of VRF152 into the circuit optimized for BLF177 in the second load-pull
experiment.
18
Device
Zin
ZL
Po (dBm)
(~150W)
PAE (%)
Gain (dB)
P1dB
(dBm)
BLF177
3.66 - j3.71
3.05 + j1.75
51.75
53.10
18.93
53.26
VRF152
2.45 - j3.40
2.36 + j2.07
51.75
59.76
18.49
52.42
VRF152 w BLF177 Zoptm
3.66 -j3.71
3.05 + j1.75
51.78
55.43
18.25
n/a
Concluding Remarks
Interestingly, the "drop-in" of VRF151 into the circuit optimized for BLF177 in the second load-pull
experiment results in both higher gain and PAE than BLF177 in its own optimal circuit, which
indicates that VRF151 and BLF177 are fairly interchangeable in their respectively optimized circuit
without any change. For stability considerations, however, it is recommended that the input matching
be modified to its own optimal impedance state for the broadest region of stability during operation.
Similar "drop-in" of VRF152 into the optimal circuit for BLF177 in the second load-pull experiment
results in higher PAE but slightly lower gain than BLF177 in its own optimal circuit, which again
indicates both VRF152 and BLF177 are also interchangeable. Again, modification on the input
matching circuit is recommended for stability considerations.
Concluding Remarks
The maximum ratings and key parameters of the four RF VDMOS transistors under study are
discussed and compared. For a quantitative analysis, ranking is assigned for each parameter
according to its merit with a score of 4 assigned to the highest ranking transistor and progressively
decreasing scores towards the lower rankings. A figure of merit (FM) is then computed for each
transistor by averaging its score over the 14 categories included in the analysis. Table 14 lists the
rankings of the 4 transistors in each of the 14 categories as well as the FM for each of the 4
transistors. It should be noted: (A) Instead of ranking the absolute value of VGS(th), the range and bin
spacing are ranked since narrower range and spacing are both desirable as discussed earlier, (B)
Although the test conditions for VGS(th), RDS(on), IDSS, RθJC may not be identical among the four
transistors, it wouldn't changed the rankings of these parameters had they been tested under the
3
same conditions , and (C) The same rankings in V(BR)DSS and gfs are given to BLF177 as SD2941-10
and VRF152 since the differences between them are small.
From Table 14, it is not surprising that VRF151 scores the highest figure of merit since it ranks first in
eight and second in 6 categories out of the 14 parameters analyzed. On the other hand, BLF177
ranks poorly among the four RF VDMOS transistors under study, which only ranks first in 2 of the 14
parameters. Also, VRF152 is closely behind VRF151 followed by SD2941-10.
The RF performance of the four VDMOS transistors is also evaluated with the load-pull experiments.
The load-pull results strongly suggest that VRF151 and VRF152 are interchangeable with BLF177.
For best stability performance, it is recommended that the input matching circuit be optimized for the
respective transistor, which often only involves a simple value change of the matching components in
the same matching circuit topology. Interchanging with SD2941-10 is not recommended since its RF
gain is significantly lower and least flat vs. output power than the other three transistors.
3
For VGS(th), instead of absolute values, only range and group spacing are used in ranking, which are relatively independent of
different sensing currents, at which VGS(th) is measured. For RDS(on), they are measured at different drain currents of 5A and 10A at
a common gate-source voltage of 10V, which are in the linear region of the ID-VD curve well below saturation, where the variation in
RDS(on) is very small. For IDSS, both VRF151 and VRF152 are measured at higher drain-source voltage at 100V, and their assigned
ranking would be the same had they were measured at a lower drain-source voltage of 50V. For RθJC, whose measurement is very
sensitive to the positioning of the thermal couple underneath the device, it is assumed the best practice as well as optimal values
are reported by the manufacturers of these transistors.
19
Comparative Analysis of 150W MOSFETs for 100MHz Applications
Table 14 · Ranking and figure of merit of the four RF VDMOS transistors under study.
RF VDMOS Transistor
V(BR)DSS
ID,max
PD,max
VGS,max
VGS(th) Range
VGS(th) Bin Spacing
RDS(on)
gfs
IDSS
Ciss
Coss
Crss
RθJC
fT
Figure of Merit
20
VRF151
VRF152
BLF177
SD2941-10
[V]
170
130
125
130
Ranking
1
2
2
2
[A]
16
20
16
20
Ranking
2
1
2
1
[W]
300
300
220
389
Ranking
2
2
3
1
[V]
±40
±40
±20
±20
Ranking
1
1
2
2
[V]
1.5
1.5
2.5
2.5
Ranking
1
1
2
2
[V]
0.075
0.075
0.1
0.1
Ranking
1
1
2
2
[W]
0.3
0.2
0.3
0.2
Ranking
2
1
2
1
[S]
5
5
4.5
5
Ranking
1
1
1
1
[mA]
1
0.05
2.5
0.05
Ranking
2
1
3
1
[pF]
375
383
480
415
Ranking
1
2
4
3
[pF]
200
215
190
236
Ranking
2
3
1
4
[pF]
12
20
14
17
Ranking
1
4
2
3
[°C/W]
0.6
0.6
0.8
0.45
Ranking
2
2
3
1
[GHz]
1.9
1.8
1.3
1.6
Ranking
1
2
4
3
3.6
3.3
2.6
3.1
Appendix A - Capacitances of a VDMOS Device
In device circuit models, capacitances of a transistor are conventionally specified as the capacitance
between any of the two terminals of the transistor, as shown in Figure A1 for a VDMOS device.
Although these terminal capacitances are useful in circuit analysis, they are very difficult to measure
directly.
Figure A1 Capacitances of a VDMOS device.
The so-called input, output and reverse transfer capacitances, Ciss, Coss and Crss, on the other hand,
are readily measurable. Figures A2 and A3 illustrate the circuits for measuring these capacitances. It
is important to note that these capacitances are highly dependent on the bias voltage at the drain
terminal, and such biasing circuit needs to be included in any of the measurement circuits.
Figure A2 Measurement Circuit for Ciss and Crss.
The input capacitance, Ciss, is defined as the capacitance between the gate and source terminals
when the drain terminal is dynamically shorted to the source terminal. By properly placing the RF
choking inductor and the DC blocking capacitor, the drain terminal is biased at VDD, while no DC
current is drawn from the DC source. In addition, the drain terminal is dynamically (AC) shorted to
ground as shown in Figure A2. The input impedance and capacitance at the gate terminal are
defined, respectively, as
(A1)
(A2)
Z iss = R iss + jX iss =
Ciss =
vg
ig
−1
2πf o X iss
From Figure A2, it can be seen that Ciss is the equivalent of Cgs and Cgd in parallel such as
(A3)
October, 2014
© 2014 Microsemi Corporation
Ciss = Cgs + Cgd
21
Comparative Analysis of 150W MOSFETs for 100MHz Applications
Also from Figure A2, the reverse transfer impedance and capacitance at the same gate terminal are
defined, respectively, as
(A4)
(A5)
Z rss = R rss + jX rss =
Crss =
vg
id
−1
2πf o X rss
Note that the drain current, id, can be measured at the external current path dynamically shorting the
drain and source terminals as shown in Figure A2, and Crss is identical to Cgd, i.e.,
(A6)
Crss = Cgd
Figure A3 Measurement circuit for Coss.
Finally, from Figure A3, the output capacitance is defined as the capacitance at the drain terminal
when the gate terminal is shorted to the source terminal. Since the gate and source terminals are DC
isolated internally, a simple DC short suffices in this case. The output impedance and capacitance at
the drain terminal are defined, respectively, as
(A7)
(A8)
Z oss = R oss + jX oss =
Coss =
vd
id
−1
2πf o X oss
As seen from Figure A3 Coss is the equivalent of Cds and Cgd in parallel such as
(A9)
Coss = Cds + Cgd
Moving beyond the terminal capacitances, Cgs, Cds and Cgd, and measurable capacitances, Ciss, Coss
and Crss, it is also insightful to see how these capacitances reside in the physical structure of a typical
VDMOS device. Figure A4 shows a conceptual cross-section of a VDMOS device along with the DC
bias conditions to activate the inversion channels for the electron current flow from source to drain
terminals.
With the aid of Figure A5, the physical capacitors within the VDMOS structure can be readily
identified: (A) The overlapped area between polysilicon and P-body and N-source regions forms the
capacitor Cox1, (B) The drain is connected to source via the drift region and the body diode at the Pbody and N-drift junction, and (C) the gate and drain are connected via the capacitor Cox2, which is
nd
the overlapped area between polysilicon and N-drift region, and a 2 capacitor Cdrift, which is formed
due to the space-charge in the drift region when drain is positively biased. It should be noted that both
Cox1 and Cox2 are relatively independent of the bias voltage at the drain, while both Cdrift and Cbody are
strongly dependent on the drain bias voltage.
22
Appendix A - Capacitances of a VDMOS Device
Figure A4 Conceptual cross-section of a typical VDMOS structure.
Figure A5 Cross-sectional view of the VDMOS capacitances.
Inspecting circuits in Figures A1 and A5, one quickly recognizes that
(A10)
Cgs = Cox1
(A11)
 1
1
+
Cgd = 
 Cox2 Cdrift
(A12)
Cds = Cbody



−1
From Equations A3, A6, A9 and A10-A12, the measurable capacitances and physical capacitances
can be linked via the following equations:
(A13)
(A14)
(A15)
 1
1
+
Ciss = Cox1 + 
 Cox2 Cdrift



Coss = Cbody
 1
1
+ 
+
 Cox2 Cdrift



Crss
 1
1
= 
+
 Cox2 Cdrift



−1
−1
−1
23
Comparative Analysis of 150W MOSFETs for 100MHz Applications
Acknowledgements
The load-pull system at MPPG was set up by highly competent RF engineers, Dick Frey and John
Soliday, with a combined RF transmitter/PA expertise of more than 60 years. Dick Frey designed and
executed the first load-pull experiment, and John Soliday did the second one. This work simply is not
possible without their expert contribution and consultation.
W. Albert Gu
Power Products Group
Microsemi Corporation
24
Microsemi Corporation (NASDAQ: MSCC) offers a comprehensive portfolio of semiconductor
solutions for: aerospace, defense and security; enterprise and communications; and industrial
and alternative energy markets. Products include high-performance, high-reliability analog and
RF devices, mixed signal and RF integrated circuits, customizable SoCs, FPGAs, and
complete subsystems. Microsemi is headquartered in Aliso Viejo, Calif. Learn more at
www.microsemi.com.
Microsemi Corporate Headquarters
One Enterprise, Aliso Viejo CA 92656 USA
Within the USA: +1(949) 380-6100
Sales: +1 (949) 380-6136
Fax: +1 (949) 215-4996
© 2014 Microsemi Corporation. All rights reserved. Microsemi and the Microsemi logo are trademarks of
Microsemi Corporation. All other trademarks and service marks are the property of their respective owners.
1823-A/10.14