Professional Documents
Culture Documents
Reflection Oscillator
Reflection Oscillator
1 of 9
E-mail: john@rfic.co.uk
2 2 2
1 - S11 − S 22 +D From Table 1 we can see the value of K at 2GHz was
K= >1
2 S12 S 21 was 1.28 ie non-conditional stability. Therefore feed-
back is needed to reduce this below 1 and this is done by
adding shunt feedback in the form of an inductor as
where D = S 11S 22 − S12 S 21 shown in Figure 2.
Sheet
2 of 9
freq StabFact1
1.500GHz -0.914
2.000GHz -0.802
2.500GHz -0.792
If we add, for example a 5nH inductor to the emitter, we The value of k is now - 0.802 ie < 1 conditionally stable,
can analyse the new circuit on a Microwave/RF CAD but now the magnitudes of S22 is greater than 1 but still
such as Agilent ADS. Analysis gives us a set of modi- < 1 for S11. If we were to use the device as it stands it
fied S-parameters and a new value of k: would probably oscillate but only after a lot of tuning
and probably would not work over a wide temperature
freq StabFact1
1.500GHz 0.939 range.
2.000GHz 0.850
2.500GHz 0.898
Again we can add a small inductance to the base to at-
freq S(1,1) S(2,1) S(1,2) S(2,2) tempt to raise the magnitudes of S11 & S22 as shown in
1.500GHz
2.000GHz
0.429 / -11.658
0.429 / -26.387
1.646 / 65.202
1.431 / 57.680
0.213 / 87.637
0.314 / 91.284
0.799 / -18.613
0.820 / -25.714 Figure 4.
2.500GHz 0.418 / -17.443 1.220 / 52.442 0.381 / 80.178 0.789 / -30.357
dB(S(1,1)) We now have the values required for the basic oscillator
25
an ADS S-parameter simulation using the ‘osctest’ will
m1 confirm whether oscillation will occur. The simulation is
20
m1 shown in Figure 6.
freq=2.001GHz
dB(S(1,1))=20.344 R
15 R R1 V_DC
R3 R=200 Ohm SRC2
R=5370 Ohm
OscTest Vdc=10 V
OscTest1
10 Port_Number=1
DC_Feed
R DC_Feed1
Z=1.1 Ohm R2
Start=1.9 GHz R=500 Ohm
Stop=2.1 GHz DC_Feed
5 DC_Feed2 I_Probe
Points=101
Ic
0
1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 pb_hp_AT42085_19911003
vout
Q1
DC_Block
freq, GHz DC_Block1
CAPQ
C3
vb
Figure 5 Input return loss plot of the ADS sche- C=0.67 pF
Q=400.0 L Term
L1
matic shown in Figure 4, showing a high positive re- F=1000.0 MHz
Mode=proportional to freq
L=5.0 nH
Term2
Num=2
R=
turn loss required for the oscillator. Z=50 Ohm
DC_Feed C
DC DC_Feed3 C2
C=100 pF
DC S-PARAMETERS
The resulting modified S-parameters and K factor are DC1
S_Param
given in Table 3. SP1
Start=1.0 GHz
Stop=3.0 GHz
freq StabFact1 Step=
1.500GHz -0.856
2.000GHz -0.587
2.500GHz -0.429
Figure 6 ADS S-Parameter simulation of the basic
oscillator – see text for description of the simulation
freq S(1,1) S(2,1) S(1,2) S(2,2)
1.500GHz 2.763 / 155.978 3.906 / -45.647 0.762 / 163.487 1.860 / -29.955
features.
2.000GHz 10.475 / 45.778 13.207 / -171.... 4.471 / 60.654 5.210 / -149.885
2.500GHz 2.183 / 19.125 1.859 / 149.421 1.240 / 32.610 0.516 / -167.596 The S-parameter model has been replaced with a non-
linear model (Spice model) and as a result the DC bias
circuit is needed to ensure the bipolar transistor is biased
Table 3 Common-Base S-Parameters & K factor to 10mA with a Vce of 8V. The 200-ohm resistor is de-
with a 5nH inductor connected between the base signed to drop 4V to give a Vce of 8V. The emitter is
and ground, now showing >1 magnitudes of S11 &
grounded via the DC_feed which is a block to all RF.
S22 with the K factor < 1
The base inductor is RF grounded via the 100pF capaci-
The value of K is -0.58 ie conditionally stable. Therefore tor and this is where the base voltage bias is applied.
we have a device that is capable of oscillating when pre- The base resistors are set to give a 10mA collector cur-
sented with a resonant load on the emitter. rent.
If we connect a component ΓG ie on the emitter port, The ‘Oscport’ block will output the closed loop magni-
with a value of ~1∠-45 then this should resonate the tude and phase of the circuit over the frequency range
input. We can use a 50ohm load termination ΓL as the specified within the Oscport block. Note the magnitude
value of S22’ is > 1. If it was < 1 then we would have to has to be > 1 and have zero phase at the required fre-
modify the circuit configuration and/or vary the value of quency. The result of the simulation is shown in Figure
the base inductance until both | S11 | and | S22 | are > 1. 7.
1 1
C= = = 0.67pF
2πfX c 2π * 2E9 * 2.375 * 50
Sheet
4 of 9
R
R R1 V_DC
R3 R=200 Ohm SRC2
R=5370 Ohm Vdc=10 V
OscPort
Osc1 DC_Feed
R DC_Feed1
SP1.SP.S(1,1) m1 V=
R2
Z=1.1 Ohm
freq=2.095GHz NumOctaves=2
R=500 Ohm DC_Feed
SP1.SP.S(1,1)=1.023 / -0.389 Steps=10 DC_Feed2 I_Probe
FundIndex=1 Ic
MaxLoopGainStep=
m1
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 pb_hp_AT42085_19911003
Q1 vout
DC_Block
DC_Block1
CAPQ
C3
C=0.55 pF vb
Q=400.0 L Term
F=1000.0 MHz L1 Term2
Mode=proportional to freq L=5.0 nH Num=2
R= Z=50 Ohm
DC_Feed C
DC DC_Feed3 C2
C=100 pF
DC
freq (1.000GHz to 3.000GHz) DC1 HARMONIC BALANCE
HarmonicBalance
Figure 7 Polar magnitude/phase result of the S- HB1
Freq[1]=2.0 GHz
Paramter simulation using the ‘Oscport’ block Order[1]=3
shown in Figure 6.
The S-parameter simulation shows that we do achieve a Figure 8 Harmonic balance ADS simulation of the
magnitude of >1 at zero phase at 2GHz. Therefore the basic oscillator design.
next step is to perform a harmonic balance simulation to
predict the output power spectrum and phase noise per- The resulting frequency spectrum is shown in Figure 9
formance. The Harmonic balance ADS simulation is with the corresponding phase noise plot shown in
shown in Figure 8. Figure 10.
10
Noise(2) Select “vout”, Select Non linear Noise & Os-
5
cillator
0
Osc Osc Port name “Osc1” 0.0 0.5 1.0 1.5 2.0 2.5 3.0
harmindex
All other parameters can be left to their default values. harmindex HB.freq
0 0.0000 Hz
1 2.003GHz
2 4.006GHz
3 6.010GHz
DC HARMONIC BALANCE
⎛ λ op ⎞
Where θ < 90 0 ⎜⎜ 90 0 ⎟
⎟
⎝ 4 ⎠ Figure 11 ADS schematic of the reflection oscillator
with ‘ideal’ microstrip RF bias lines, capacitors and
inductors fitted.
Rearrange to get θ ie
VARIABLE FREQUENCY NEGATIVE
⎛X ⎞ RESISTANCE OSCILLATOR [2,3,4]
θ = tan −1 ⎜ L ⎟ + 90 = 51.5 + 90 = 142 degrees
⎝ Zo ⎠ In the following section the design of a varactor con-
trolled variable frequency negative resistance oscillator
(2) Calculate 0.55pF capacitor as an open-circuit stub. is given. This design example will use the same device
and RF circuitry used in the fixed frequency version.
ωC = 2π.2E 9 .0.55E −12 = 0.069 In order to pick a suitable varactor for the design we
need to decide on the tuning bandwidth. This particular
tanθ requirement is for a tuning bandwidth of 20MHz/V over
ωC = ∴ tan −1 (Xc.Zo) the tuning range of 1 to 10V. To give us a bit of margin
Zo
to cover temperature effects etc we pick a bandwidth of
~ 25MHz/V. The varactor should have minimal case
θ = 19 degrees parasitics so we opt for a SMT package device.
⎛ λ op ⎞
Where θ < 90 0 ⎜⎜ 90 0 ⎟ The equivalent circuit of a typical varactor together with
⎝ 4 ⎟⎠ the package parasitics is shown in Figure 12.
Sheet
6 of 9
where :
C = diode capacitance
V
V = applied voltage,
R
V = junction contact potential (~ 0.7V)
J
γ = Capacitance exponent
C = diode capacitance
V
From the fixed frequency oscillator we can calculate the Table 4 Showing the resulting tuning ranges for the
inductance of the input of the reflection oscillator as: varactor and coupling capacitor combinations. To
achieve ~ 25MHz/V we use a series capacitor of
⎛ 1 ⎞
2
⎛ 1 ⎞
2 0.4pF in series with the varactor. Note also that we
⎜ ⎟ ⎜ ⎟ need a total inductance of 30nH including the in-
2.π .f ⎠ 2.π .2E 9 ⎠
L=⎝ = ⎝ = 11.5nH ductance of 11.5nH looking into the reflection am-
C 0.55E -12 plifier.
By rearranging the varactor equation we can find CJ by These resonator components were added to the harmonic
using the data CV=1.0pF at 4V. balance ADS schematic as shown in Figure 13. Note the
inductor has been reduced to adjust the frequency to
This will give a value of CJ ~ 2.39pF 2GHz at the mid control voltage (ie Cvaractor = 1pF).
30
V_DC
R
SRC2
Vdc=10 V R
R1
20 m1
R=200 Ohm
R3
R=5370 Ohm
10
OscPort
Osc1
V= R DC_Feed
DC_Feed2
0 m1
R2
Z=1.1 Ohm
NumOctaves=2 R=500 Ohm
DC_Feed
DC_Feed1
harmindex=1
C
C3
Steps=10 -10 dBm(HB.vout)=15.987
FundIndex=1
C=1.6 pF MaxLoopGainStep= pb_hp_AT42085_19911003
Q1 -20
I_Probe
vout Ic 0.0 0.5 1.0 1.5 2.0 2.5 3.0
L C
L1 C5 L dBm(HB.vout)
L=11.0 nH C=0.5 pF
DC L2
C
C4
harmindex
R=
DC_Feed
DC_Feed3
L=4.0 nH
R=
C=100 pF harmindex HB.freq
DC
DC1 0 0.0000 Hz
DC_Block
DC_Block1
Term
Term1
1 2.131GHz
Num=1 2 4.262GHz
HARMONIC BALANCE Z=50 Ohm
HarmonicBalance -40
HB1
Freq[1]=2.0 GHz
Order[1]=3 -60 m2
m2 noisefreq=10.01kHz
-80 pnfm=-86.41 dBc
Figure 13 ADS schematic of the Voltage Controlled
-100
Oscillator (VCO). The varactor is represented by
C3 which is set to 0.66pF (for 10V control voltage) -120
and 1.66pF (for 10V control voltage).
-140
The simulation was run with the min & max varactor -160
capacitances resulting in the plots shown in Figure 14 1E 1E 1E 1E 1E 1E 2E
and Figure 15 respectively. 2
pnfm, dBc
3 4 5 6 7 7
30 noisefreq, Hz
-120
-140
-160
1E 1E 1E 1E 1E 1E 2E
2 3 4 5 6 7 7
pnfm, dBc noisefreq, Hz
V_DC TLOC R
SRC2 TL5 R1
Vdc=10 V R
R3
Z=20.0 Ohm
E=90
R=200 Ohm dBm(HB.vout)
R=5370 Ohm F=2 GHz
30
Ref
Ref
R
R2 TLOC
DC
R=500 Ohm TL4 25
TLIN TLIN
Z=20.0 Ohm
DC TL2 TL3
DC1
TLIN Z=80.0 Ohm Z=80.0 Ohm
E=90
F=2 GHz 20 m1
TL6 E=90
F=2 GHz
E=90
F=2 GHz
m1 harmindex=1
Z=50.0 Ohm
E=85 pb_hp_AT42085_19911003 15 dBm(HB.vout)=14.585
F=2 GHz Q1
I_Probe
C vout Ic 10
C5 vb
OscPort
C=2.5 pF C
Osc1
V=
TLOC C4 5
TL7 C=100 pF
Z=1.1 Ohm Z=50.0 Ohm
NumOctaves=2
C Steps=10
E=142 Ref 0
F=2 GHz Term
C3
C=Cv pF
FundIndex=1 Term1 0.0 0.5 1.0 1.5 2.0 2.5 3.0
MaxLoopGainStep= Num=1
Var
Eqn VAR
HARMONIC BALANCE
Z=50 Ohm
harmindex
VAR1
Cv=1.0
HarmonicBalance
harmindex HB.freq
HB1 0 0.0000 Hz
Freq[1]=2.0 GHz
Order[1]=3 1 1.994GHz
2 3.989GHz
Figure 16 Final VCO ADS schematic. The lumped ele- -40
ment inductors have been replaced with ‘ideal’ micro-
strip lines. The varactor is represented by the 1pF ca- -60 m2
pacitor (which equates to the mid-tuning voltage). In -80 m2 noisefreq=10.01kHz
operation the open-circuit stub on the resonator would pnfm=-94.83 dBc
be trimmed/extended to center the oscillating fre- -100
quency. The open-circuit stub on the bipolar base
-120
would be trimmed/extended to ensure reliable oscilla-
tion - especially over temperature. -140
CONCLUSION
This paper discussed the design of a basic fixed-
frequency reflection oscillator and a variable frequency
varactor controlled version, both using a distributed
resonator and micro-strip circuit elements. Both oscilla-
tors were designed to operate at 2GHz with the variable
frequency version designed to have a tuning bandwidth
of ~ 20MHz/V. Throughout the design process staring with
the basic oscillator (using ideal lumped elements) to the more
complex final design (using ‘ideal’ micro-strip elements),
Agilent ADS large and small simulations were given at each
stage to verify circuit performance. It was noted that substitut-
ing ‘ideal’ micro-strip elements for the lumped RF bias and
resonator circuits considerable reduced the tuning bandwidth
of the oscillator. This was overcome in this design by increas-
ing the varactor coupling capacitor. Another possibility would
be to use a larger capacitance swing varactor ie one with a
larger Gamma eg MaCom MA46H071 (Gamma 0.75) or even
the MA46H201 (Gamma 1.25).
REFERENCES
[1] Microwave Engineering, David M Pozar, Addison-
Wesley, 1993, ISBN 0-201-50418, p183