High Frequency VCO Design and Schematics

 

Iulian Rosu, YO3DAC – VA3IUL, http://www.qsl.net/va3iul/

pdf version

 

This note will review the process by which VCO (Voltage Controlled Oscillator) designers choose their oscillator’s topology and devices based on performance requirements, components types and DC power requirements.

            Basic oscillator design specifications often require a given output power into a specified load at the design frequency. The drive level and bias current set the fundamental output current and the oscillation frequency is set by the resonator components.

Transistor selection of the transistor should consider noise, frequency, and power requirements. Based on the particular device, the design may account for parasitics of the device affecting resonator components as well as nonlinear performance specifications.

All the VCO schematics presented below were practical build using the Infineon SiGe transistor BFP420, and any of them can be re-tuned for different frequency ranges changing varicaps and LC tank values.

 

VCO Specifications

 

Phase Noise varies typically by 3dB with temperature, in the –55ºC to +85ºC range.

VCO Design Recommendations

Nonlinear Effects in VCOs
 

    Oscillator circuit nonlinearities cause low-frequency noise components to be up-converted and to appear as noise sidebands on the VCO output. Although this statement is intuitively obvious, quantifying this mechanism is much more complex. Second-order nonlinear distortion determines the degree of noise contamination of the oscillator output for instance. Therefore, second-order distortion in the oscillator should be minimized. The degree to which any oscillator accomplishes this goal can be judged based on the second harmonic output level of the oscillator.

Another useful indicator of good oscillator design is the change in oscillation frequency versus DC bias reduction.

 

Other VCO impairments including injection locking, load pulling, and power supply frequency pushing can cause serious oscillator performance degradation, particularly in phase-locked systems. If the induced impairments fall within the closed-loop bandwidth of the system, potentially chaotic spectral behavior can result. Design margins must be identified and held for each of these potential problem areas.
 

Injection Locking
   

Injection locking can be shown that when a signal of sufficient amplitude and sufficiently small frequency error is impressed on a free-running oscillator. Over time, the free-running oscillator changes its frequency to that of the impressed signal with a corresponding change in its signal phase and amplitude. Normally, injection locking is a very undesirable situation, but it has been used to advantage on occasion such as in narrowband bit synchronizers.

 

Load Pulling
   

VCO load pulling refers to the change in oscillator frequency that occurs when the oscillator load impedance is changed. If this impedance change is dynamic in nature, load pulling of the oscillator leads to direct frequency modulation of the oscillator. Obviously, if the VCO is contained within a phase-locked loop and the frequency of modulation lies within the closed-loop bandwidth, unwanted interactions can result.
One of the most serious load pulling situations that can occur in practice arises in modulators where the modulation signal causes (low-frequency) baseband frequency modulation of the load.

In this situation the load reflection coefficient, become a function of the modulation signal.

 

Frequency Pushing
   

VCO frequency pushing is the technical term applied to the oscillator frequency perturbations that result from small changes in the oscillator's supply voltage(s). These perturbations can result from a number of factors including changes in device capacitance values caused by modified reverse biased junction capacitances, changes in the oscillator self-limiting signal mechanism, and changes in the sustaining stage gain.

Oscillator frequency pushing can lead to substantial phase noise degradation because any power supply noise directly can lead to frequency modulation of the oscillator.

 

Varactor Diode Nonlinear Effects
 

One of the main nonlinear VCO elements, particularly in wideband VCOs, is the varactor diode. The potentially large voltage swing across the varactor(s) leads to departures in the frequency tuning curve from nominal and up-conversion of low-frequency noise components that contribute to VCO phase noise sidebands.
It can be shown that minimal varactor distortion occurs when the VCO tuning varactors are used in the back-to-back topology. Second-order distortion is theoretically reduced to zero when matched abrupt junction varactor diodes are used in this configuration.

 In order to lower the VCO Phase Noise, a number of design rules should be respected:

 

There is a trade-off between the Q factor of the oscillator, its size and its price. The low Q-Factor of an LC tank and its component tolerances  needs careful design for phase noise without individual readjustment of the oscillators.

Usually a larger resonator will have a higher Q (e.g. a quarter wavelength coaxial resonator).

A bipolar transistor biased at a low collector current will keep the flicker corner frequency to a minimum, typically around 6 to 15 KHz (Most semi-conductor manufacturers can provide the frequency corner (fc) of their devices as well as the 1/f characteristic.

In order to increase the power at the input of the oscillator, the current has to be increased. However, a low current consumption is critical to preserving battery life and keeping a low fc. In a practical application, the current will be set  based on output power required to drive the system (typically a mixer), and then the Phase Noise will need to be achieved through other means.

- The abrupt tuning diodes will provide a very high Q and will also operate over a very wide tuning voltage range (0 to 60 V). The abrupt tuning diode provides the best phase noise performance because of its high quality factor.

- The hyperabrupt tuning diodes, because of their linear voltage vs. capacitance characteristic, will provide a much more linear tuning characteristic than the abrupt diodes. These are the best choice for wide band tuning VCO's. An octave tuning range can be covered in less than 20 V tuning range. Their disadvantage is that they have a much lower Q and therefore provide a phase noise characteristic higher than that provided by the abrupt diodes.

CAD analysis can be used to chose the varactor diode doping profiles for linear frequency tuning even in the presence of large signals.

This is the most challenging compromise because the thermal noise from the equivalent noise resistance of the varactor works together with the tuning gain of the VCO to generate phase noise. This compromise will be the limiting factor determining the phase noise performance.

 

VCO Topologies

 

Parallel Tuned Colpitts VCO

 

There are 3 types of BJT Colpitts VCOs. Common-Collector, Common-Emitter and Common-Base.

The most used is Common-Collector configuration where the output is often taken from the collector terminal, simply acting as a buffer for the oscillator connection at the base-emitter terminals.

This is the only Colpitts arrangement in which the load is not part of the three-terminal model or the oscillator equation; though care must be taken to ensure that the collector output voltage does not significantly feedback through the base-collector junction capacitance.

As an alternative, the output of the common collector could also be taken across emitter resistance Re.

 

Series Tuned Colpitts VCO (Clapp VCO)

The series-tuned Colpitts circuit (or Clapp oscillator) works in much the same way as the parallel one.

 

Wideband Colpitts VCO

 

Hartley VCO

 

Wideband Differential Push-Push VCO

Voltage controlled push-push oscillators have been known for some period of time, particularly because they provide a proven design which has twice the output frequency capability of any single transistor in a microwave VCO.

    • The push-push VCO use two symmetric subcircuits to oscillate in odd mode (180° out of phase) to the fundamental frequency fo, when the 2nd harmonic (2fo) is coupled to the output port.
    • Hence, at the output of a push-push oscillator the fundamental signal fo and odd harmonics (3fo, 5fo,..) are canceled out, while the even harmonics (2fo, 4fo,..) are constructively combined in-phase at the output network, delivering the 2nd harmonic to the output load.
    • Any push-push oscillator must rely on generation of signals from each transistor that should be rich in 2nd harmonic components because the fundamental components will cancel out. For this reason, bipolar transistors are preferred because they have better 2nd harmonic characteristics due to inherent non-linearities.
    • As the two sub-oscillators use a common resonator and they oscillate at halve of the output frequency, higher resonator Q-factors are available.
    • In a push-push oscillator the fundamental frequency signal is terminated by a “virtual ground’. Thus, the loaded Q-factor is equal to the unloaded Q-factor of the oscillator, a fact which improves the phase noise     performances of the oscillator.
    • Another improvement in the phase noise is coming from the fact that by definition, push-push oscillators provides load termination for the 2nd harmonic signal, which otherwise would be reflected back, fluctuates each voltage of the oscillator and works as a noise power for the fundamental oscillation, which leads to increase of the phase noise.
    • Additionally, push-push oscillators are highly resistant to load pulling effects, because the sub-oscillators are terminated by a ‘virtual ground’ and only the 2nd harmonic frequency component is influenced by the oscillators load impedance. This behavior will improve the frequency stability of the oscillator.
    • To obtain strong fundamental fo rejection at the push-push output, it is very important to have a good symmetry of the circuit and of the PCB layout.
    • For temperature compensation, current mirror circuits could be added at the emitter and the base of the transistors, but this will be done at the expense of increasing the wideband phase noise.

Differential Cross-Coupled VCO

 

The cross-coupled differential transistor pair presents a negative resistance to the resonator due to positive feedback.

This negative resistance cancel the losses from the resonator enabling sustained oscillation.

Frequency variation is achieved with two varicap diodes BB135.

 

Negative Resistance VCO

The output can be taken by capacitive coupling at the emitter (low level) or at the collector (higher level, but have more spurious).

 

Franklin VCO

 

Franklin oscillator uses two transistor stages having the same common terminal (emitter for bipolar device) when the greater power gain and better isolation from the resonant circuit is possible compared with the case of a single-stage configuration.

There are two possible configurations for the resonant circuit, parallel and series. The circuit presented below uses a parallel LC resonant circuit (L1 and the varctor diode).

In the case of a parallel resonant circuit configuration, the resonant LC circuit is isolated from the input of the first stage and the output of the second stage by means of small shunt capacitances C1 and C2 having high reactances at the resonant frequency.

    In this circuit, each stage shifts phase 180° so that the total phase shift is 360° which is equivalent to zero phase shift. We may say that one stage serves as the phase inverting element in place of the RC or LC network which generally performs this function. It is, from an analytical viewpoint, immaterial which stage we choose to look upon as amplifier or phase inverter. The configuration is essentially symmetrical in this respect; both stages provide amplification and phase inversion.

 

Goral VCO

 

 

    The Goral VCO has an emitter-follower stage inserted in the feedback path of an otherwise conventional Colpitts oscillator circuit.

    Note that the emitter-follower is directly coupled to the JFET. It may be necessary to experiment with bias-determining resistances to ascertain Class-A operation from the emitter-follower. Also, the output transistor is intended to operate in its Class-A region.

 

Cascode VCO

 

To provide higher isolation of the load from the VCO resonant circuit a cascode VCO configuration, can be used.

The negative resistance oscillation conditions for common emitter transistor Q1 are provided by using the feedback inductance L1.

Vackar VCO

 

And here is the winner. If you want to build a very stable, low phase noise, and low spurious VCO, definitely Vackar VCO is the choice.

This is not a common type in the RF “professional” world; one reason could be the name of its inventor.

A Vackar VCO is a variation of the split-capacitance oscillator model. It is similar to a Colpitts or Clapp VCO in this respect. It differs in that the output level is more stable over frequency, and has a wider bandwidth when compared to a Colpitts or Clapp design.

 

    The Vackar VCO circuit incorporates a π-section tank to attain the needed 180° phase-reversal in the feedback loop.

 

References:

 

1. Alpha Industries - VCO Application notes

2. Minicircuits - VCO Application notes

3. Oscillator Basics and Low-Noise Techniques for Microwave Oscillators and VCOs - U.Rohde

4. Oscillator Design and Computer Simulation - R.Rhea

5. RF and Microwave Transistor Oscillator Design – A. Grebennikov

6. Practical Oscillator Handbook - I. Gottlieb

7. Frequency Synthesizers Design Handbook - J.A.Crawford

8. RF Design Magazine - 1997 - 2003

9. Microwave Journal - 1997 - 2008

10. Microwaves & RF - 2002 - 2006

Flag Counter