aei systems component test and model summary · file: aei systems component test and model...
TRANSCRIPT
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
1.0 Component Test and Model Summary This document contains results for various tests performed on magnetics (88uH inductor, seven
output flyback power transformer), passive and active components (capacitor, 2N6059,
IRHNM57110), and various ICs (LM120/LM140), along with their SPICE correlations. This
report gives a good indication of AEi System’s ability to accurately model components and ICs.
1.1 Individual Component Testing
Inductors
A 88uH inductor is tested for impedance over frequency and saturation characteristics using the
Picotest J2130A DC bias injector.
Figure 1.1-1: Simplified diagram of setup used to test the impedance of each inductor.
J2130A
DC BiasInjector
bode_100AC = 1
DC
Inductor
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-2: Impedance (in dB) and inductance of the 88uH inductor.
From these plots, the inductance, DCR, capacitance, and effective parallel resistance can be
extracted for modeling purposes.
Figure 1.1-3: Effective 88uH inductor model.
1
DCR54m
1
88uH
Cp14p
Rp100k
L1 VLz1
I1AC = 1
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-4: Bench and simulation results for the 88uH phase and impedance.
The bench and simulation data are overlayed for each inductor; these overlay plots show that the
model performs exceptionally well over a wide frequency range. However, these models do not
account for the saturation characteristics of each inductor. The tabulated results and graphs are
shown below.
Figure 1.1-5: 88uH inductance as DC current increases.
3 z_bench 13 phase_bench 14 phase_sim 15 z_sim
100 1k 10k 100k 1Meg 10Megfrequency in hertz
-40.0
0
40.0
80.0
120z_
sim
, z_b
ench
in d
B(v
olts
)
-80.0
-40.0
0
40.0
80.0
phas
e_si
m, p
hase
_ben
ch in
deg
rees
plot
1
13
3
14
15
88uH InductorCurrent (A) mag (Ω) @80kHz Inductance (H)
0.01 42.85 85.25E‐060.1 42.782 85.11E‐060.2 42.742 85.03E‐060.4 42.645 84.84E‐060.6 42.56 84.67E‐060.8 42.455 84.46E‐061 42.34 84.23E‐06
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-6: Graphical representation of the 220uH inductance vs. DC current.
A 6th order polynomial equation is used to curve-fit the points up to 1A of DC current. The 6th
order equations are then used to generate a saturable core inductor model to replace the ideal
inductor used in the SPICE models. The results of the model saturation with current are
overlayed with the measurement results to show the nearly perfect correlation between the two.
Figure 1.1-7: Bench and simulation results of the 88uH inductor saturation curve.
y = 4E‐05x6 ‐ 0.0001x5 + 0.0001x4 ‐ 7E‐05x3 + 2E‐05x2 ‐ 3E‐06x + 9E‐05R² = 1.0001
84.00E‐06
84.20E‐06
84.40E‐06
84.60E‐06
84.80E‐06
85.00E‐06
85.20E‐06
85.40E‐06
0 0.2 0.4 0.6 0.8 1 1.2
88uH
88uH
Poly. (88uH)
1 simulated_88uh 2 measured_88uh
100m 300m 500m 700m 900mi1 in amperes
84.0u
84.4u
84.8u
85.2u
85.6u
sim
ulat
ed_8
8uh,
mea
sure
d_88
uh in
vol
tsPl
ot1
21
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Multi‐Winding Power Transformers
The power transformer used in the main converter is tested, and a sophisticated “mesh model” is
created. DC resistance of each winding is tested using the 4-wire resistance measurement
capability of the Picotest M3500A 6½ Digit Multimeter. The inductance of the primary winding
is measured using the Bode 100 Vector Network Analyzer (VNA).
Figure 1.1-8: Schematic of the power transformer windings.
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-9: Impedance of the primary winding of the transformer (N7). The winding measures 78uH.
The turns ratios are measured and compared to the calculated values.
compared to
compared to
compared to
compared to
compared to
compared to
compared to
-20
-10
0
10
20
30
40
50
60
70
-300u
-200u
-100u
0
100u
200u
300u
102 103 104 105 106 107
TR1/
dBTR
2/H
f/HzTR1: Mag(Impedance) TR2: Ls(Impedance)
nN1 1.295:=
N7N1
1.294=
nN2 1.295:=
N7N2
1.294=
nN3 0.564:=
N7N3
0.564=
nN4 0.564:=
N7N4
0.564=
nN5 0.479:=
N7N5
0.478=
nN6 0.917:=
N7N6
0.917=
nN8 1.101:=
N7N8
1.1=
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
The measured values are in good agreement with the calculated ratios. Next, the leakage
inductances are measured between the windings. Measurements of voltage and current are
performed with the OMICRON Lab Bode 100 network analyzer using the Gain function at the
71.5 kHz switching frequency. The ratio of V/I is measured with voltage probe on Ch2
(reference) and the Tektronix P6022 current probe on Ch1. All windings other than the winding
where voltage is being measured are shorted during each measurement.
These values are then each used in the equation below, to calculate the leakage inductances.
After each leakage inductance is calculated, the values are placed in a subcircuit along with the
turns ratios, primary inductance, and DCRs of each winding to create the Mesh model of the
transformer.
LN7N1VN7IN1
2π f nN7⋅ nN1⋅:=
LN7N1 1.354 10 6−×=
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-10: Final SPICE “Mesh” model of the power transformer.
This model is an extremely high-fidelity model that yields hyper-accurate switching simulation results as shown below.
2 1
N7RATIO = 1
1
LN777.83u
1 123
Lleak216.6n
3
RN721m
1 26
LN7N11.354u
11
1 5
LN7N21.593u
1 8
LN7N37.807u
1 13
LN7N430.82u
1 14
LN7N546.14u
1 15
LN7N65.918u
16 1
LN7N8790.5n
6
RN142m
5 9
N2RATIO = 0.772
1
1
9
RN248m
8 11
N3RATIO = 1.773
11
RN398m
1
1
13 18
N4RATIO = -1.773
26 6
N1RATIO = -0.772
1
1
14 19
N5RATIO = -2.088
1
1
15 20
N6RATIO = -1.091
1
1
1621
N8RATIO = -0.908
16 16
18
RN4102m
19
RN5123m
20
RN670m
21
RN845m
1
1
26
5
LN1N21.577u
5
5
26
26
2626
26
8
LN1N33.575u
8
8
26
26
2626
26
13
LN1N416.77u
13
13
26
26
2626
26
14
LN1N516.78u
14
14
26
26
2626
26
15
LN1N69.742u
15
15
26
2626
16
LN1N81.705u
55
88
1313
1414
1515
2626 26 2626
26
5
5
55
5
16
LN2N82.111u
5
5
55
5
15
LN2N65.785u
5
5
55
5
14
LN2N521.93u
5
5
55
5
13
LN2N46.668u
5
8
55
LN2N33.054u
5
5
8
8
88
8
16
LN3N813.85u
8
8
88
8
15
LN3N658.42u
8
8
88
8
14
LN3N525.65u
8
13
88
LN3N48.502u
8
8
15
151515
14
141414
13
131313
8
888
13
131313
14
141414
15
151515
161616 16161616
16
13
13
1313
13
16
LN4N813.65u
13
13
1313
13
15
LN4N630.42u
13
14
1313
LN4N514.63u
13
13
14
141414
15
151515
1616
14
14
1414
14
16
LN5N875.25u
14
15
1414
LN5N66.178u
14
14
15
16
1515
LN6N810.12u
15
15
15
151515
1616 1616
55
88
1313
2626
1414
1515
PIN1
PIN2
PIN3
PIN4
PIN5
PIN6
PIN7
PIN8
PIN9
PIN10
PIN11
PIN12
PIN13
PIN14
PIN15
PIN16
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-11: Simulated ripple voltages compared to the bench measurements.
The simulation of the output voltage ripple compares very well to the bench measurements.
Transformer models for power supplies such as multi-winding flybacks must be made with this
level of fidelity. The results for transient and frequency domain (stability) simulations using less
accurate models will not be correct.
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Capacitors
Bench measurements of the capacitor’s impedance magnitude, impedance phase shift, and ESR
are obtained to create an AC only frequency dependent model of the capacitor. This model is not
used in transient simulations.
Figure 1.1-12: 220uF output capacitor model configuration (AC Domain model) based on bench measurements. Voltage sources are used to tolerance resistor equations. The equations are used to fit the
model to measured bench data.
res7res8
V5 V9
4
5
C11237u
5
6
L1111.0196n
4
6
Itest
Impedance
R18R=(exp((log(0.105)+(log(f req/(2*pi))-log(3000))*((log(0.04)- log(0.105))/(log(300000)-log(3000))))/log(e))-0.015)*V(res7)+1n
R9R=(exp((log(0.05)+(log(f req/(2*pi))-log(7000000))*((log( 0.07)- log(0.05))/(log(20000000)-log(7000000))))/log(e))+1n)*V( res8)+1n
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-13: Overlay of the bench measurement of the 220uF output capacitor impedance magnitude, impedance phase shift, and ESR and the simulated result using the AEi Systems’ model.
Figure 1.1-14: 220uF capacitor model obtained from the manufacture vs. the same bench data as shown in
the previous figure. This model is not recommended for analysis.
1 benchdb 2 benchp 3 benchre 4 simdb 5 simp 6 simre
100 1k 10k 100k 1Meg 10Meg 100Megfrequency
-360
-180
0
180
360
sim
p, b
ench
p
50.0m
150m
250m
350m
450msi
mre
, ben
chre
-75.0
-55.0
-35.0
-15.0
5.00
sim
db, b
ench
dbPl
ot1
14
36
25
1 ph_v(1) 2 db_v(1) 3 PHASE(VOUT) 4 VDB(VOUT)
100 1k 10k 100k 1Meg 10Meg 100Megfrequency in hertz
-25.0
-15.0
-5.00
5.00
15.0
db_v
(1),
VD
B(V
OU
T) in
dB
-80.0
-40.0
0
40.0
80.0
PH
AS
E(V
OU
T), p
h_v(
1) in
deg
rees
Plo
t1
1342
SpiTan ModelVendor Model
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Linear Regulators
Seven linear regulators were obtained from the manufacturer in order to get a statistic set of data.
An individual model of each regulator is created and tested for accuracy, to ensure reliable
simulation results when the part is simulated in the system.
Each linear regulator is tested with the same input voltage as seen in the system and over a broad
current range. In each test, the output impedance is measured using the Picotest J2111A Current
Injector coupled with the Bode 100 VNA.
Figure 1.1-15: Typical bench setup for output impedance and non-invasive phase margin measurements.
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-16: Schematic of output impedance measurement of the LM1X0 linear regulators.
The bench test and simulation results for each regulator are shown below. Note the measured
effective inductance vs. the simulated effective inductance of each regulator. The effective
inductance created by the control loop of each regulator defines the AC and transient
performance of the regulator. Each regulator is tested over current to characterize the dynamic
impedance. Creating a model with performance similar to the real part ensures accuracy in all
simulation conditions, not just at a single operating point.
N1214
V8-15V
VN12in
Gnd
out14
LM120-12
14 14
J2111A25mAAC = 1
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Regulator Code Load Current(A) Frequency(Hz) Magnitude Magnitude (model) Eff. Inductance (H)Eff. Inductance (H)
(model)
25.00E‐03 20.00E+03 0.168703 0.197 1.34E‐06 1.57E‐06125.00E‐03 20.00E+03 0.07058 0.075 561.66E‐09 596.83E‐09225.00E‐03 20.00E+03 0.058824 0.0605 468.11E‐09 481.44E‐09525.00E‐03 20.00E+03 0.050509 0.0497 401.94E‐09 395.50E‐0925.00E‐03 30.00E+03 0.164836 0.1801 874.48E‐09 955.46E‐09145.00E‐03 30.00E+03 0.067453 0.068 357.85E‐09 360.75E‐09265.00E‐03 60.00E+03 0.096894 0.115 257.02E‐09 305.05E‐09505.00E‐03 60.00E+03 0.086723 0.098 230.04E‐09 259.95E‐0925.00E‐03 30.00E+03 0.202648 0.222 1.08E‐06 1.18E‐06175.00E‐03 30.00E+03 0.075407 0.0789 400.05E‐09 418.58E‐09325.00E‐03 30.00E+03 0.067457 0.0662 357.87E‐09 351.20E‐09625.00E‐03 30.00E+03 0.060588 0.0583 321.43E‐09 309.29E‐09
25.05E‐03 50.00E+03 0.3654 0.353712 1.16E‐06 1.13E‐0650.81E‐03 50.00E+03 0.205822 0.216301 655.15E‐09 688.51E‐09126.84E‐03 50.00E+03 0.134198 0.133822 427.17E‐09 425.97E‐09412.52E‐03 50.00E+03 0.107021 0.0948756 340.66E‐09 302.00E‐0925.12E‐03 30.00E+03 0.53467 0.432 2.84E‐06 2.29E‐0651.22E‐03 30.00E+03 0.276979 0.268383 1.47E‐06 1.42E‐06111.96E‐03 30.00E+03 0.181456 0.180237 962.65E‐09 956.19E‐09368.79E‐03 30.00E+03 0.136197 0.126429 722.55E‐09 670.73E‐0925.15E‐03 20.00E+03 0.412242 0.369939 3.28E‐06 2.94E‐0645.34E‐03 20.00E+03 0.249459 0.245317 1.99E‐06 1.95E‐06101.55E‐03 20.00E+03 0.157817 0.157306 1.26E‐06 1.25E‐06359.30E‐03 20.00E+03 0.112346 0.103504 894.02E‐09 823.66E‐09
Regulator Code Load Current(A) Frequency(Hz) Magnitude Magnitude (model) Eff. Inductance (H)Eff. Inductance (H)
(model)
25.00E‐03 20.00E+03 0.521982 1.03339 4.15E‐06 4.11E‐0650.00E‐03 20.00E+03 0.390884 1.08517 3.11E‐06 3.14E‐06
LM120H‐15(‐15V)
(500 mA)
LM120‐12(‐12V)
11506BYA
LM120‐5(‐5V)
11505SYA
LM120K‐15(‐15V)
11507BYA
LM140‐5(+5V)
LM140‐12(+12V)
10706SYA
10707SYA
LM140‐15(+15V)
10708SYA
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Various linear regulator overlays are shown below to demonstrate the accuracy of all aspects of
the model, including effective inductance, resonant frequency, Q of the resonant peak, and low
frequency impedance.
Figure 1.1-17: Overlay of the LM120-12 linear regulator output impedance (in Ohms) with a 25mA load.
Figure 1.1-18: Overlay of the LM120-12 linear regulator group delay with a 25mA load.
1 sim_magnitude 2 bench_magnitude
100 1k 10k 100k 1Meg 10Megfrequency in hertz
10m
20m
50m
100m
200m
500m
1
2
5
10
sim
_mag
nitu
de, b
ench
_mag
nitu
de in
ohm
sPl
ot1
21
2 bench_tg 4 sim_tg
100 1k 10k 100k 1Meg 10MegFREQUENCY in hertz
1p
10p
100p
1n
10n
100n
1u
10u
100u
1m
10m
sim
_tg,
ben
ch_t
g in
sec
onds
Plot
1
2
4
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-19: Overlay of the LM140-12 linear regulator output impedance (in Ohms) with a 25mA load.
Figure 1.1-20: Overlay of the LM140-12 linear regulator group delay with a 25mA load.
1 sim_magnitude 2 bench_magnitude
100 1k 10k 100k 1Meg 10Megfrequency in hertz
1m
10m
100m
1
10
100
benc
h_m
agni
tude
, sim
_mag
nitu
de in
ohm
sPl
ot1
12
1 bench_group_delay 4 sim_group_delay
100 1k 10k 100k 1Meg 10Meg 100MegFREQUENCY in hertz
10p
100p
1n
10n
100n
1u
10u
100u
1m
10m
benc
h_gr
oup_
dela
y, s
im_g
roup
_del
ay in
sec
onds
Plot
1
41
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-21: Overlay of the LM140-15 linear regulator output impedance (in Ohms) with a 45mA load.
Figure 1.1-22: Overlay of the LM140-15 linear regulator group delay with a 45mA load.
1 sim_magnitude 2 bench_magnitude
100 1k 10k 100k 1Meg 10Meg 100Megfrequency in hertz
1m
10m
100m
1
10
100
sim
_mag
nitu
de, b
ench
_mag
nitu
de in
ohm
sPl
ot1
21
1 bench_group_delay 3 sim_group_delay
100 1k 10k 100k 1Meg 10Meg 100Megfrequency in hertz
10p
100p
1n
10n
100n
1u
10u
100u
1m
10m
benc
h_gr
oup_
dela
y, s
im_g
roup
_del
ay in
sec
onds
Plot
1
31
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-23: Overlay of the LM120H-15 linear regulator output impedance (in Ohms) with a 25mA load.
Figure 1.1-24: Overlay of the LM120-15 linear regulator group delay with a 25mA load.
1 outz_simulation 2 outz_bench
10k 20k 50k 100k 200k 500k 1Megfrequency in hertz
100m
200m
500m
1
2
5
10
20
50
100
outz
_ben
ch, o
utz_
sim
ulat
ion
in o
hms
Plot
1
1
2
LM120H-15V Output Impedance 25mA Load1uF ceramic capacitor on the output
2 measured_group_delay 4 simulated_group_delay
10k 20k 50k 100k 200k 500k 1MegFREQUENCY in hertz
5.00u
15.0u
25.0u
35.0u
45.0u
mea
sure
d_gr
oup_
dela
y, s
imul
ated
_gro
up_d
elay
in s
econ
dsP
lot1
42
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
A small test circuit is assembled using regulators and capacitors provided by the program to test
a piece of circuitry as it exists in the real system, and to compare the model results to a full
circuit.
Figure 1.1-25: Test circuit used for testing the stability of the -12V linear regulator, U62.
Figure 1.1-26: Overlay of the LM120-12 linear regulator test circuit output impedance (in dB Ohms) with a
100mA load.
in
Gnd
out
U5LM120-15
in
Gnd
out
U10LM120-15
in
Gnd
out
U62LM120-12
VN3-19.5V
C1330uF
C210uF
C310uF
C41uF
Cceramic0.1uF
I1100mA
n12n15Vrf
n15V
Vn12
Vin
x18
2 n15_vout_bench 11 n15_vout_sim
100 1k 10k 100k 1Meg 10Megfrequency in hertz
-20.0
-10.0
0
10.0
20.0
n15_
vout
_sim
, n15
_vou
t_be
nch
in d
B(vo
lts)
Plot
1
2
11
Impedance in dBsim and bench are both at 100mA loads
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-27: Overlay of the LM120-12 linear regulator group delay with a 100mA load.
Looking at the peak in group delay that coincides with a peak in the output impedance, it is
possible to determine the bandwidth and phase margin (PM) of the regulator (LM120-12).
(Bandwidth of the circuit)
This calculation can be verified using a small signal load step to observe the oscillation in the
-12V output voltage. The load step can also be simulated and compared to the measured results
to further confirm the accuracy of the SPICE model.
2 tg_bench 4 tg_sim
100 1k 10k 100k 1Meg 10MegFREQUENCY in hertz
100p
1n
10n
100n
1u
10u
100u
1m
tg_b
ench
, tg_
sim
in s
econ
dsPl
ot1
4
2
Group Delay100mA load
Tg 5.89410 6−⋅:=
Freq 146.273103⋅:=
Q 93 10 6−⋅ 27 103
⋅, ( ) 7.889=
PM 9.954 106⋅ 81.780103
⋅, ( ) 20.907=
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-28: Measured load step response of the -12V linear regulator test circuit.
Figure 1.1-29: Simulated load step response of the -12V linear regulator test circuit.
1 voutn12 3 load_current
205u 225u 245u 265u 285utime in seconds
-40.0m
0
40.0m
80.0m
120m
load
_cur
rent
in a
mpe
res
-12.033
-12.023
-12.013
-12.003
-11.993
vout
n12
in v
olts
Plot
1
1
3
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Darlington Transistor
The 2N6059 is a Darlington structure NPN BJT. To properly model the part, it must be
understood how the part behaves at low currents and over frequency. At low currents, only one
of the transistors is turned on (Q4 shown below). Once enough current flows through the 70
Ohm resistor to generate approximately a 0.6V drop, transistor Q5 turns on. A small test circuit
is built and the forward transfer and PSRR (Power Supply Rejection Ratio) of the 2N6059 is
tested.
Figure 1.1-30: Subcircuit schematic of the 2N6059 Darlington transistor.
Figure 1.1-31: Test circuit used to test the AC forward transfer gain of the 2N6059.
1
R1070
1
1 1
R115.5k
1
Q42N6059b
1
Q52N6059aVc
Vb
111
1
Ve
Vc
3
V49.64716AC = 1
9
V115
R310k
4
C20.9u
VoutVout
4
5
R525m
5
L112n
Vb
I125m
8
R4350m
8
9
L220n
33
R150
3
Q12N6059
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-32: Bench measurements of the 2N6059 forward transfer gain at various load currents.
Figure 1.1-33: Simulation results of the 2N6059 forward transfer gain at various load currents.
1 6ma 2 7ma 3 8ma 4 9ma 5 10ma 6 11ma 7 12ma 8 13ma
100 1k 10k 100k 1MegFREQUENCY in hertz
-60.0
-40.0
-20.0
0
20.0
9ma,
8m
a, 7
ma,
6m
a, 1
3ma,
12m
a, 1
1ma,
10m
a in
dB
(vol
ts)
Plot
1
5678
1234
2N6059 Forward TransferAll measurements are loaded with the Current Injector exceptfor the 21mA measurement which is a purely resistive load
3, 4 and 5mA datapoints lie basically right on top 6mA and 7mA---- up to 5mA data is not shown here
1 6ma 2 7ma 3 8ma 4 9ma 5 10ma 6 11ma 7 12ma 8 13ma
100 1k 10k 100k 1Megfrequency in hertz
-60.0
-40.0
-20.0
0
20.0
10m
a, 1
3ma,
9m
a, 1
2ma,
8m
a, 1
1ma,
7m
a, 6
ma
in d
b(vo
lts)
Plot
1
12
6
3
7
4
85
Model Forward Transfer Results
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-34: Overlay of the measured and simulated PSRR of the 2N6059 test circuit.
A large signal simulator like Agilent’s ADS (unlike SPICE, which is a small signal simulator)
can simulate the pole frequency of the forward transfer of the 2N6059 while sweeping the load
current. Using the model created in SPICE, the low current point at which the second transistor
fully turns on can be observed.
3 bench_psrr 14 sim_psrr
100 1k 10k 100k 1Meg 10Meg 100Megfrequency in hertz
0
20.0
40.0
60.0
80.0
sim
_psr
r, be
nch_
psrr
in d
BPl
ot1
3
14
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-35: 2N6059 forward transfer test circuit simulated in ADS.
Figure 1.1-36: Simulated results of the 2N6059 forward transfer pole frequency vs. current.
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-37: Bench testing image of the 2N6059 transistor.
Power MOSFETS
This section contains a description of the IRHNM57110 (100V 100kRad Hi-Rel Single N-
Channel TID Hardened MOSFET) AEi Systems SPICE model performance compared to
measured test data for the part. Comparisons are also made to the SPICE model provided by IR
on their web site. Forward transfer and VGS vs. IS data are presented.
This model is made with the proviso that the MOSFET is to be used in a discrete LDO design,
converting a 3.3V input source voltage to a 2.5V output voltage. In this application, the dynamic
source impedance of the MOSFET forms a moving pole with the output capacitors, dependent
primarily on the load current. In order to properly simulate this circuit, the MOSFET model must
accurately portray the dynamic resistance (1/gfs) as a function of operating current.
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
The test setup output capacitor was measured in order to determine the actual capacitance, as
well as the parasitic resistance and inductance.
Figure 1.1-38: Schematic used for simulating the MOSFET’s forward transfer curve.
The IRHNM57110 is configured to have 3.3V on its drain and a variable gate voltage to keep the
source voltage at 2.5V. The forward transfer (Vs/Vg) is measured using an OMICRON Lab
Bode 100 VNA and Picotest J2130A DC Bias injector. A load current range of 1mA – 1.5A is
tested.
The SPICE subcircuit model structure is based on a template previously developed by AEi
Systems during its collaboration agreement with IR in 1998. The structure is configured for the
IR FET process.
AC plots (gain/phase), comparing the AEi Systems model performance to the bench data, are
shown next.
DC_BIAS_INJECTOR5.33756VAC = 1
Vin3.3
Vs
Vg
Vd
VsQAEi1DUT
Rload8
C21u
8
6
R225m
6
L120n
3.30V 2.20V
5.34V
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-39: Gain and Phase of the forward transfer with a 2.49k Ohm load or 1mA.
Figure 1.1-40: Gain and Phase of the forward transfer with a 1.25k Ohm load or 2mA
1 sim_phase 2 sim_gain 3 bench_phase 4 bench_gain
10 100 1k 10k 100k 1Meg 10Megfrequency in hertz
-70.0
-50.0
-30.0
-10.0
10.0be
nch_
gain
, sim
_gai
n in
db(
volts
)
-150
-50.0
50.0
150
250
benc
h_ph
ase,
sim
_pha
se in
deg
rees
Plo
t1
1
2
3
4
3.3Vin, 2.5Vout, 2.49k ohm load
1 sim_phase 2 sim_gain 3 bench_phase 4 bench_gain
10 100 1k 10k 100k 1Meg 10Megfrequency in hertz
-70.0
-50.0
-30.0
-10.0
10.0
benc
h_ga
in, s
im_g
ain
in d
b(vo
lts)
-150
-50.0
50.0
150
250
benc
h_ph
ase,
sim
_pha
se in
deg
rees
Plo
t1
1
2
3
4
3.3Vin, 2.5Vout, 1.25k ohm load
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-41: Gain and Phase of the forward transfer with a 649 Ohm load or 3.85mA
Figure 1.1‐42: Gain and Phase of the forward transfer with a 325 Ohm load or 7.7mA
1 sim_phase 2 sim_gain 3 bench_phase 4 bench_gain
10 100 1k 10k 100k 1Meg 10Megfrequency in hertz
-70.0
-50.0
-30.0
-10.0
10.0be
nch_
gain
, sim
_gai
n in
db(
volts
)
-150
-50.0
50.0
150
250
benc
h_ph
ase,
sim
_pha
se in
deg
rees
Plo
t1
12
3
4
3.3Vin, 2.5Vout, 649 ohm load
1 sim_phase 2 sim_gain 3 bench_phase 4 bench_gain
10 100 1k 10k 100k 1Meg 10Megfrequency in hertz
-70.0
-50.0
-30.0
-10.0
10.0
benc
h_ga
in, s
im_g
ain
in d
b(vo
lts)
-150
-50.0
50.0
150
250
benc
h_ph
ase,
sim
_pha
se in
deg
rees
Plo
t1
12
3
4
3.3Vin, 2.5Vout, 325 ohm load
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-43: Gain and Phase of the forward transfer with a 2.49k Ohm load. Results using IR’s model are shown and compared with the AEi Systems model and test data. IR’s model does not match the bench results (pink and orange traces).
The disparity in the IR model’s performance, with the pole being over a decade too high in frequency, is a function of the incorrect
modeling of the transconductance of the MOSFET. This would cause inaccurate AC analysis results, especially in the stability
assessment, if this model were used in a linear regulator simulation.
1 sim_phase_aei 2 sim_gain_aei 3 bench_phase 4 bench_gain 5 sim_phase_ir 6 sim_gain_ir
10 100 1k 10k 100k 1Meg 10Megfrequency in hertz
-70.0
-50.0
-30.0
-10.0
10.0si
m_g
ain_
ir, s
im_g
ain_
aei,
benc
h_ga
in in
dB
(vol
ts)
-150
-50.0
50.0
150
250
sim
_pha
se_i
r, si
m_p
hase
_aei
, ben
ch_p
hase
in d
egre
esP
lot1
3
4
1
2
5
6
3.3Vin, 2.5Vout, 2.49k ohm load
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
Figure 1.1-44: Forward Transfer gain comparison between the AEi Systems’ model and IR’s model at under various load conditions (1mA-7.7ma). This shows the large transconductance sensitivity to load current. The pole would continue moving at higher currents.
The incorrect performance of the poles moving in the IR model will greatly impact the transient step load performance of the model.
The transconductance of the MOSFET must be modeled correctly to see the proper transient response differences during the rising and
falling edges of a load current step.
1 aei_vdbs_2p49k_load 2 aei_vdbs_1p25k_load 3 aei_vdbs_649_load 4 aei_vdbs_325_load 5 ir_vdbs_2p49k_load6 ir_vdbs_1p25k_load 9 ir_vdbs_649_load 10 ir_vdbs_325_load
-70.0
-50.0
-30.0
-10.0
10.0
AE
i mod
el F
orw
ard
Tran
sfer
in d
b(vo
lts)
Plo
t1
1234
10 100 1k 10k 100k 1Meg 10Meg 100Megf i h t
-70.0
-50.0
-30.0
-10.0
10.0
irIR
mod
el F
orw
ard
Tran
sfer
in d
b(vo
lts)
Plo
t2
56910
Forward Transfer Gain of AEi 57110 MOSFET model
Forward Transfer Gain of IR 57110 MOSFET model
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
The DC characteristics of the MOSFET are characterized and compared to the bench
measurements below. The VGS of the FET is measured at various load currents to obtain
important information about the N, RS, VTO and RS terms of the SPICE model.
Figure 1.1-45: VGS vs. IS of the AEi Systems model, IR model and bench measurements over load current.
Figure 1.1‐46: Tabulated VGS vs. IS of the AEi Systems model, IR model and bench measurements over
load current. The IR model, set up mainly for hard switching applications, is not accurate.
0
1
2
3
4
5
6
1.0E‐3 10.0E‐3 100.0E‐3 1.0E+0 10.0E+0
VGS [V]
IS [A]
VGS vs IS
Bench
AEi
IR
Log. (Bench)
Log. (AEi)
Log. (IR)
Load Current (A) Vg (V) Vgs (V) Vg (V) Vgs (V) Vg (V) Vgs (V)Two 2.66k in parallel 1.9E‐3 5.754 3.254 5.74935 3.24935 7.21116 4.71116Four 2.66k in parallel 3.8E‐3 5.864 3.364 5.84766 3.34766 7.21713 4.71713150 ohm resistor 16.7E‐3 6.096 3.596 6.06263 3.56263 7.24113 4.7411399.57 ohm resistor 25.1E‐3 6.131 3.631 6.12362 3.62362 7.25225 4.7522521.92 ohm resistor 114.1E‐3 6.389 3.889 6.36805 3.86805 7.32955 4.829555.14 ohm resistor 486.4E‐3 6.601 4.101 6.78174 4.28174 7.53377 5.033772.57 ohm resistor 972.8E‐3 6.967 4.467 6.93484 4.43484 7.7454 5.24541.63 ohm resistor 1.5E+0 7.201 4.701 7.16522 4.66522 7.96495 5.46495
Bench Aei Model 6/21 IR Model
Component Testing and Model Correlation
File: AEi Systems Component Test and Model Summary.docx
Proprietary to AEi Systems, LLC Revision Date: 2/9/2012
The AC and DC performance of AEi Systems’ model is in excellent agreement with the acquired
bench data. The locations of the poles are at the correct frequencies and move appropriately with
varying load current. This allows the linear regulator stability to be properly represented in the
simulation of the full regulator application circuit at any operating current.
The VGS vs. IS performance follows the bench measurements with 1% accuracy which allows
accurately depiction of the DC and transient performance of the circuit.
The IR provided model proved to be deficient in all areas that were tested. The pole locations in
the forward transfer measurement are more than a decade greater in frequency than they should
be and the pole does not move correctly with respect to load current indicating both value and
topological deficiencies in the SPICE subcircuit structure.
The VGS vs. IS performance of the IR model was also inadequate as it did not line up with the
bench measurements and did not perform correctly over load current.
1.2 Conclusion All of the data presented above provides insight into the performance of the system’s power
supply. All of the components tested have been modeled and those models have been verified
against the measured results. The high levels of confidence in the models provide the basis for
accurate full-system simulations and analyses performed in the Worst Case Circuit Analysis.
As a general observation, the negative linear regulators has proven difficult to stabilize, as at low
currents the effective inductance of the regulator is quite large, and the effective Q of the
regulator at its bandwidth with no output capacitor is very high.