Oscillators Theory and Practice - QSL.net

Q Peaking at ~34000, and

Eagleware Gene

d Delay Peaking at 110 uS

esis Run

Crystal with Series

• 30 pF series C with parasitics sh
• Dominate factor is series C, note

with a specific series C in the cir

C, L and Parasitics

hift frequency up 340 Hz
e: crystals are specified/etched
rcuit

Colpitts Crys

Maybe
3-15 MHz
for values

shown

stal Oscillator

• Schematic shows some
practical values and one
way to accomplish bias.

• Crystal operates above
series resonance and
acts like a very high Q
inductor.

• Can replace tuning
capacitor with a varactor
circuit to implement
electronic tuning.

• Room for improvement…

3K load , 10 mW?

TTL Gate Cry

• One of many single and
• Crystal runs fundamenta
• Crystal “sees” ~30 pF for
• Negative feedback resist
• Rectangular output, harm
• Fine for digital / less critic

ystal Oscillator

dual gate implementations
al series mode
r correct frequency and trim
tors bias gates for startup
monic rich past 900 MHz
cal applications, watch EMI

Pierce + Oscillator S

• Ostensibly an AC model of a Pierc
– Get extra 180° of phase shift fr

• If you ground the collector and red
follower) oscillator

• If you ground the base, you get you
says grounded base Colpitts) desig

• If they are really all the same circu

Simplified Schematic

ce Oscillator (grounded emitter)
rom capacitors and crystal
draw you get a Colpitts (emitter
u get yet a third (Clapp? Messer
gn.
uit then performance is … similar?

Pierce Oscillat

• Not a Messer favorite but recommen
Oscillatek

• Crystal operates as an inductor.
• Output tank tuned off (?, low) and he

tor Schematic

nded by Collins, worked well for
elps suppress harmonics

75 MHz Oscillator, Som

• Gain limited to
desired overtone
• Can Test Amp

• Bypassing, regulator and b
• Positive supply implementa
• Can trim with resistor in pla

in series mode, can add se

me Redeeming Features

buffer not shown
ation probably preferred
ace of crystal, crystal operated
eries C with crystal

End of Crysta

Octob

al Oscillators

ber 09

Regenerativ

ve Oscillator

• Circuit captured for
historical interest

• Not considered
necessary for new
designs

SMALL SIGNAL ANALYSIS OF

OPEN LOOP CHARACTERISTICS

LARGE SIGNAL OSCILLATOR AN

NALYSIS USING GENESYS/HARBEC

Large Signal Oscillator Ana
with Harbec

• Shows schematic, output spectru
• Voltage plots can be used to see

alysis Using Genesis 2006

um, start-up, Q, phase, & loop gain
swings across VCO varactors

GENESYS NON-LINEAR OSC
• Shows poor start-up & blocking

(Cayenn

CILLATOR SIMULATION
g (?) of a sub-optimum design
ne)





Phase Noise an

• Voltage and current noi
supplies, and VCO cont
junction capacitances a

• PN is characterized vers
overall in degrees or rad
interest

– Higher at lower frequ
• PN can be detrimental f

– Degrades phase dem

– Reciprocal mixing rai
presence of large sig

– Introduces noise in F

nd Residual FM

ise from devices, power
trol voltages modulate
and cause PN/residual FM
ses frequency ￡ and/or
dians in a bandwidth of

uency, 1/F, other
for three reasons:
modulated Bit Error Rate
ises Rx noise floor in the
gnals
FM systems

Plot of Phase Noise vs. Off

• The good stuff: Datum PTS9600, Vec
• PN increases at 10 log BW and 20 Lo

fset Frequency in 1 Hz BW

Messer Integration
verses BW, start ?
N=1600 (8 GHz)

100K 1M 10M
HP105 (not visible)

1.6° 4.3° 12.2°

Datum 9600 10 MHz
N=800 8 GHz

0.21° 0.64° 2.0°

ctron & Wenzel 501-07127 Oscillators,
og N multiplication ratio



Reserved

Negative Impedance Osci
Configu

Sig Gen

Output

Zin

• Signal circulates into negative im
is reflected back out larger than th
– Assumes the conjugate match
looks like -45 ohms

• IF the conjugate match is perfect
– If a signal is present and close
in phase step with the input
– Injection gains in excess of 50
temperature

• A frequency determining element
• Degrees of freedom are frequenc

and saturation

illator/Reflection Amplifier
uration

Bias

Conjugate Negative
Match Impedance

mpedance from signal generator and
he input.
h is not perfect, for example if Zin

the system will oscillate (-50 ohms)
e on frequency the oscillator will get

0 dB are possible at one

t like a resonator or cavity is useful
cy and impedance shift on power up

4 GHz NEGATIVE RESISTANCE
~Ala Bendix Avionics circa 198

E OSCILLATOR SCHEMATIC
80, designer Andre Polichek

• ~400 mW output

• 200 MHz tuning

• Transistor Ft ~6
GHz, so beta is
low, base and
emitter RF current
nearly equal

• Bias chokes
present open
circuit impedance

• Emitter
capacitance
creates negative
impedance at
base

NEGATIVE RESISTANCE O
850

Eagleware

OSCILLATOR SCHEMATIC
MHz

e Genesys Example

NEGATIVE RESISTANCE OS

SCILLATOR: OPEN LOOP

Development Fixture fo

VB Vcc

..

Input board & base Device Output b
Mtg.
Block

Modulator
PS

Triple Fixture Triple
Stub DUT Stub
Tuner Tuner

or 60 Watt Pulsed Osc.

board & base Fixture:

• Provides
transistor mtg.

• Provides for
bias, DC blocks

• Allows
separation for
impedance
measurements

Sig Gen Pulse
Power
Pad Meter
r
Spectrum
analyzer

1.2 GHz Pulsed P

C1
RFC

Feedthru

1 uS

• Custom 60 watt pulsed power tr
• Microstrip substrate bonded on
• Emitter and collector resonators
• Oscillator fed a dual transistor q

stability.

Power Oscillator

+Vcc

RFC

Out

Microstrip C3
coupler C2

ransistor originally from MSC
aluminum plate for heat transfer
s intended to stabilize frequency
quad hybrid amplifier for impedance

Good Practice

1. Use a well regulated and/or post fil
2. Maximize high loaded Q for best pe
3. Try to swamp variable device paras
4. Incorporate several dB of extra loo
5. Control/minimize the gain at off freq
6. Buffered oscillators and oscillator b
7. Don’t parallel bypass capacitors, br
8. Use high stability caps everywhere

– NPO or sometimes N for resona
– Higher stability bypass caps
9. Uses sealed thermistors for temper
10. Series mode crystal oscillators are
11. Provide a good thermal and mecha
– Use mechanically/electrically st
– Ground and/or isolate the corre

for Oscillators

ltered power system
erformance
sitics with stable capacitors
op gain in feedback oscillators
quencies, consider high-pass feedback
buffering is/are highly desirable
reak up with beads
e if possible (glass, ceramic, porcelain):
ate circuits

rature comp.
more deterministic, “try series R”
anical environment, watch shock & vib.
table tuning components
ect (human) end of tuning elements

Good Practi

1. Run the oscillator off of a b
a. See that it starts and run
b. See good start up with “A
c. Retest under the real po

2. Look for noise peaks at oth
3. For production oscillators (

a. Try different date codes
b. Test over temperature (l

freeze and soldering iro
c. Monitor XCO’s weekly a

months or more, at leas

ice: Testing

bare power supply (system):
ns properly over a good range
AC and DC” power up
ower and load conditions
her frequencies for instabilities
(establish and keep all data!):
of amplifier devices
last resort: judicious quick
on, limit temperature shock)
at constant temp for several
st sample