Local Oscillator and Frequency Synthesizer Circuits

Chapter 8

Local Oscillator and Frequency Synthesizer Circuits

The local oscillator is used in superheterodyne receivers to convert the incoming RF signal to the IF. Similarly, in product detectors, local oscillators are used to convert the IF signal to audio. They come in three flavors: L-C variable frequency oscillators, crystal oscillators, and crystal frequency synthesizers. In this chapter we will look at all three types. The local oscillator must be at least spectrally pure or the mixing will suffer. If the receiver is of variable frequency, then it also must be agile (able to switch frequencies in as little as microseconds) and the increments of the frequency change must be small. The frequency resolution usually is 1–100 Hz below 30 MHz, with a few as low as 0.001 Hz. Generally, at VHF/UHF frequencies the resolution can be 1000 Hz.

L-C VARIABLE FREQUENCY OSCILLATORS Variable frequency oscillators (VFOs) are radio frequency signal generators that can be

continuously tuned using an inductor connected to either an air variable capacitor, mica “trimmer” capacitor, or a voltage-tuned variable capacitance diode (varactor). VFOs differ from crystal oscillators in that the frequency can be varied in the VFOs, while in the crystal oscillators it is either fixed or variable over only a tiny region. VFO circuits can be used as signal generators in test equipment, to control transmitters, as the local oscillator in either superheterodyne or direct conversion receiver projects, or in any other applications where a continuously variable source of RF energy is needed.

RF OSCILLATOR BASICS Both VFOs and crystal oscillators are part of a class of circuits called feedback oscillators. Figure 8.1 shows the basic configuration of this type of circuit; it consists of an amplifier with open-loop gain Avol and a feedback network (which usually is frequency selective) with a “gain” of β.

109

110

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

GAIN

VI

S

VF

A1 AVOL

FEEDBACK NETWORK

b

Fig. 8.1 Feedback oscillator block diagram.

BARKHAUSEN’S CRITERIA If two conditions, called Barkhausen’s criteria, are met, then the circuit will oscillate: (1) The loop-gain is unity or greater and (2) the feedback signal arriving back at the input is phase shifted 360º (which is the same as 0º). For most practical circuits, with the 180º provided by an inverting amplifier, an additional 180º of phase shift must be provided by the feedback network. By its nature, an inductorcapacitor (L-C) tuned circuit can provide this 180º phase shift at only one frequency.

BASIC CATEGORIES OF VFO CIRCUIT We consider three general categories of feedback RF oscillator: Armstrong oscillators, Hartley oscillators, and Colpitts oscillators (also, a subset of the Colpitts oscillators are the Clapp oscillators). These basic configurations are shown in Figure 8.2. The circuit in Figure 8.2A is the Armstrong oscillator. It is identified by the feedback link positioned as a secondary winding on the tuning coil. The circuit in Figure 8.2B is the Hartley oscillator. It is identified by the resonant feedback network that contains a tapped inductor (which effectively forms an inductive voltage divider).

FREQUENCY DETERMINING COMPONENTS

The Colpitts and Clapp circuits (Figure 8.2C) are identified by the feedback network that contains a tapped capacitance voltage divider. In both cases, the feedback voltage divider is part of the resonant L-C tuning network.

POWER SUPPLIES FOR VFO CIRCUITS It always is a good idea to use only regulated DC power supplies with VFO (or any oscillator) circuits. In fact, most experts agree that a single regulator serving the oscillator is best, because it is not affected by load changes in other circuits on the same side of the regulator. The reason for using only regulated DC is that most oscillators shift frequency a slight amount when the power supply voltages change. Although not all of the oscillator circuits in this chapter show the regulator, it is a good idea to use one anyway.

SOME BASIC CIRCUIT CONFIGURATIONS The basic circuits just discussed can be configured in any of several different ways. In this section we use one of three active de-

Local Oscillator and Frequency Synthesizer Circuits

111

A1

A1 C1A L1

C1B

C1

L2

A C

L1

A1

L1

B C1

vices as the amplifier portion: junction field effect transistor (JFET), metal oxide field effect transistor (MOSFET), or the NE-602/NE612 RF balanced mixer integrated circuit. The basic JFET is the MPF-102 device, while the MOSFET is the 40673 device. Input-Side Circuit Configurations Figure 8.3 shows two input configurations, one each for MPF-102 and 40673. Figure 8.3A shows the circuit for the JFET device. It consists of a gate resistor of 100 kΩ to ground and a diode (1N914, 1N4148, or equivalent). In many cases, you need a capacitor in the gate circuit (C1), especially if a DC source or ground is directly in the circuit. To prevent loading the tuned circuit, it is customary to make the coupling capacitor small compared to the tuning capacitor: Typically values of 2–10 pF are used for HF and MW VFO circuits. The same circuit can be used on the MOSFET, but a DC bias circuit is needed for

Fig. 8.2 Oscillators: (A) Armstrong; (B) Hartley; (C) Colpitts.

the second gate, G2. The bias network shown in Figure 8.3B (R2/R3) is set to bias G2 to about one third the V+ voltage. A bypass/decoupling capacitor (C2) is used to set G2 to a low impedance for RF, while keeping it at the bias voltage for DC. The diode in the input circuit perplexes some people when they first see it in oscillator circuits. The function of the diode is to clean the signal and make it closer to a low-harmonic sine wave (all non-sine waves or distorted sine waves have harmonic content; by definition, a “pure” sine wave has no harmonics). Figure 8.4A shows the sine wave output signal when the diode is connected. Note that the waveform is a reasonably good sine wave. The waveform in Figure 8.4B is a distorted sine wave and has a higher amplitude than in Figure 8.4A. The circuit can be either parallel tuned (Figure 8.5A), in the case of Colpitts or Hartley oscillators, or series tuned (Figure

112

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

+12 VDC

C2 0.01 µF

R3 100K

R2 33K

A

C1 100 pF

D G

TUNED CIRCUIT

Q1 MPF-102 (or equiv.)

B

C1 100 pF

D G2 G1

Q1 40673 (or equiv.)

S

S D1 1N4148

R1 100K

TUNED CIRCUIT

D1 1N4148

R1 100K

Fig. 8.3 Oscillator circuit with a diode to forward bias the transistor: (A) JFET; (B) MOSFET.

A

B

Fig. 8.4 Oscillator waveform: (A) with diode; (B) without diode.

8.5B) in the case of Clapp oscillators. In the Hartley oscillators, the inductor (L1) is tapped. In any L-C resonant circuit, resonance is that point where the inductive reactance (XL) and capacitive reactance (XC) are equal. Because these elements cancel out, the impedance of such a circuit is resistive. Generally, the C/L ratio should be high, so it is common practice to select a relatively small inductance and match it with a higher capacitance. Many of the os-

cillators in this section are designed for the middle of the 1–10 MHz band and use inductance values on the order of 3.3–7 µH. These inductance values are relatively easy to obtain when using either “solenoid” (cylindrical) or “toroidal” coil forms. Both fixed value and variable inductors can be used for these circuits. The capacitance can be made up from more than one capacitor, generally the best practice. The change of frequency is propor-

Local Oscillator and Frequency Synthesizer Circuits

113

C4

L1

C2 TRIM

C1 MAIN

C4

C3

L1

FHz

=

1 2p

C2 TRIM

C1 MAIN

L1 ( C1 + C 2 + C 3 )

ASSUMES C4 << (C1+C2+C3)

C3

B

A

Fig. 8.5 Circuits: (A) parallel tuned; (B) series tuned.

tional to the square root of the ratio of the capacitance change: F max = C max F min C min

(8.1)

That is why a variable capacitor for the AM broadcast band consists of a 25–365 pF air variable capacitor shunted by a 25 pF (or so) trimmer capacitor (usually set to around 15 pF). The 3.08:1 ratio of maximum to minimum capacitance is more than sufficient to cover the 3.02:1 ratio of the maximum and minimum frequencies of the AM broadcast band. Selecting values of L and C is a somewhat tedious and iterative affair. One is advised to sit down with a calculator and make a few trials. Part of the problem occurs because both fixed and variable capacitors come in standard values that may or may not be exactly what is needed. Juggle the inductance (which is easy to wind to a custom value) to provide the desired frequency change with the available variable capacitors. It is common practice to make the total of the fixed capacitors plus the maximum values of the variable (main tuning and trimmer) capacitors somewhat larger than the to-

tal required to resonate at the very lowest frequency in the desired range. When the trimmer is set to a value less than maximum, the total capacitance will be close to the desired value. Hartley JFET VFO The Hartley oscillator, you may recall, is identified by a feedback path that includes a tapped inductor; the inductor also is part of the resonant tuning circuit of the oscillator. Figure 8.6 shows a simple Hartley oscillator based on the MPF-102 JFET. The output signal is taken through a small-value capacitor (to limit loading) connected to the emitter of the transistor. The frequency of oscillation is set by the combined effect of L1, C1, and C2: FHz =

1 2π ( L1)(C 1 + C 2)

(8.2)

To resonate at 5 MHz with a 5 µH inductor, a total capacitance of about 200 pF is required. Because of stray capacitances and errors in the values of actual capacitors, it is common practice to use more total capacitance than needed and use variable

114

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

+12 VDC C6 0.1 µF

D2 9.1V 400-mW

R1 180 1-W C5 0.01 µF

C3 10 pF

L1 C2 TRIM

C1 MAIN

Q1 MPF-102

R2 100K

D1 1N4148

C4 0.001 µF OUTPUT

Fig. 8.6 Hartley oscillator circuit.

+12 VDC C6 0.1 µF

C7 10 µF

+

R4 220

C5 0.01 µF

Q1 MPF-102 C3 10 pF

R2 100K

D1 1N4148

OUT B C4 0.01 µF

R3 150

C2 0.01 µF

C9 80 pF TRIMMER

C1 0.001 µF

8 t.

14 t.

L1

VT 0-12V

5 t.

R1 10K

D2 NTE-618 (etc.)

Fig. 8.7 Voltage-tuned Hartley oscillator.

C8 0.001 µF

OUT A

Local Oscillator and Frequency Synthesizer Circuits

capacitors to trim. For example, we could use a 140 pF variable capacitor for the main tuning and an 80 pF trimmer to set the maximum to the required value. Figure 8.7 shows a 5 MHz Hartley VFO circuit based on the MPF-102 JFET. It is very similar to the previous circuit in basic concept, but there are some differences. The most significant difference is the use of a variable capacitance diode (varactor) instead of the main tuning capacitor. The diode shown here is a 14–440 pF varactor used to tune the AM broadcast band in radio receivers. Another significant difference is that the output signal is taken from a secondary winding on the tuning inductor. This winding consists of fewer turns than the lower portion of the main tuning inductor. Care must be taken to not load down this circuit by connecting a varying load resistance across the output winding. The circuit of Figure 8.7 can be made to sweep the entire frequency range by applying a sawtooth waveform that rises from 0 V to +12 V. This type of circuit makes it relatively easy to build a sweep generator circuit or a swept-tuned radio receiver. In general, the sweep rate should be around 40 Hz if the detector has a narrowband filter. Other sweep frequencies are usable as well, but care must be taken not to “ring” any following resonant circuits or filters by a too-fast sweep rate. The tuning properties of a varactor L-C circuit are nonlinear because of the nature of varactor diodes. A graph of tuning voltage versus frequency is recommended and that only the linear portion of the curve be used for sweep purposes. The circuit in Figure 8.8A is capable of producing as much as several volts (that is volts). The actual tuning range depends on the particular components used. The heart of the oscillator is an MPF-102 JFET (Q1). Two different tuning schemes are provided. For a limited range, the main tuning is provided by a 365 pF AM broadcast band tuning capacitor and the tuning range is 5000–5500 kHz. In that case, the varactor diode (D2) is disconnected and not used. The alternate scheme deletes C1 and uses either a 365 pF

115

variable or the varactor circuit (shown) at point A. This version has a rather wider tuning range for the same capacitance change. In the previous circuit, the tuning range was reduced by the capacitor divider action of C4 and C5. The circuit of Figure 8.8A can be modified by adding or subtracting capacitors from the tuning network. As shown, C1, C2, C3, C4, and C5 are all part of the tuning circuit. The 126 pF fixed capacitor (needed for 5–5.5 MHz operation) is made from parallel 82 and 47 pF capacitors. A version of the circuit using the NTE-618 or ECG-618, 440 pF diode was built without the mica trimmer capacitor (C2). It produced a voltage (Vt ) versus frequency characteristic shown in Figure 8.8B (this curve is for a circuit with C2 removed). The varactor is comfortable over a range of 0 to +12 V (although my DC power supply goes down to only +1.26 V, which is why the offset at 2.2 MHz). The tuning range is 2.2–4.1 MHz, but this can be lowered by increasing any of the capacitors (e.g., C3) or by either eliminating or reducing the capacitors (C1–C5). The main inductor (L1) consists of 35 turns of #28 enameled wire on an Amidon T-50-2 (RED) toroidal core. The tap is placed at eight turns from the ground end. The tap is formed by winding two separate but contiguous windings: one of 8 turns and one of 27 turns. These two windings are connected together electrically at the point where they come together. The wire ends of the two coils are soldered and used as a single wire to connect to the source (S) of the JFET oscillator. The drain (D) of the JFET is kept at a ground potential for RF signals by the bypass capacitor, C8. The output signal is taken from the source (S) terminal through a 10 pF capacitor. This signal is fed to Gate 1 (G1) of the 40673 dual-gate MOSFET used as an output buffer amplifier stage. This circuit produces a 44 dB gain and is responsible for making the signal voltage so large. The output transformer (T1) consists of an Amidon Associates FT-5043 core wound with 20 turns of #28 enameled wire for the primary winding and 6 turns of the same wire for the secondary winding.

116

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

The output signal of Figure 8.8A is huge as RF oscillators go, so it may have to be attenuated in some cases. This job can be done by connecting an attenuator resistor pad in series with the output signal or reducing the DC bias voltage on the MOSFET Gate 2. This latter job can be done by reducing the value of R4 until the desired output

signal is achieved (do not reduce it below about 1 KΩ). Clapp VFO Circuit The Colpitts and Clapp oscillators are very similar to each other in that both depend on a capacitor voltage divider (C4 and C5 in +12 VDC

C9 0.1 µF U1 78L05 R6 100

R4 68K R2 100

C8 0.01 µF

C14 0.01 µF

T1 6 t.

C6 3 pF

C2 34 pF

C3 129 pF

C4 100 pF

R1 100K

D1 1N4148

C7 10 pF

F.B .

RF OUTPUT

Q2 40673

R3 100K

C5 82 pF

C1 365 pF

R7 100

C11 0.01 µF

0.01 µF A D2 440 pF 10K

A

VT

V

TUNING VOLTAGE (V T)

12

1.26

B

F 2.2

FREQUENCY (MHz)

Fig. 8.8 Hartley VFO: (A) circuit; (B) tuning voltage curve.

+

R5 100K

C10 0.01 µF

20 t.

Q1 MPF-102

C12 0.1 µF

4.2

C13 10 µF

Local Oscillator and Frequency Synthesizer Circuits

Figure 8.9) for feedback. The difference in the two oscillator circuits is that a Colpitts oscillator uses a parallel resonant L-C circuit, while a Clapp oscillator uses a series resonant L-C circuit. The circuit in Figure 8.9 is a Clapp oscillator because of the series tuned L-C network. This circuit can be used from 0.5 to 7 MHz, even higher if C4 and C5 are reduced to about 100 pF. The output circuit from this circuit is taken from the source (S) of the JFET oscillator transistor. The output circuit includes an RF choke (RFC1) that builds the output amplitude. Bias voltage for the JFET is provided by R2 and the source-drain current flowing through it. NE-602 or NE-612 VFO Circuits The NE-602AN and NE-612AN devices are double-balanced modulators and oscillator integrated circuits. Normally, they are used as the RF front ends of radio receivers, but if the DBM is unbalanced by placing a 10 KΩ resistor from the RF input (pin 1) to ground, it will function as an oscillator that produces about 500 mV output signal.

117

Figure 8.10 shows an NE-602AN/NE612AN Colpitts oscillator circuit. Three capacitors (C1, C2, and C3) are used in this circuit, rather than two, because of a need for DC blocking. These capacitors should be equal to each other and have a value on the order of 2400 pF/FMHz. The inductor should have an approximate value of 7 µH/FMHz. The tuning capacitor, C4, should have a value that will resonate with the selected inductor: C4 =

1

(8.3)

4π f 2 L1 2

For example, a 5000 kHz (5 MHz) oscillator should have network capacitors of 2400 pF/5 MHz, or 480 pF (use 470 pF standardvalue capacitors). The inductor should be 1.4 µH (17 turns on an Amidon T-50-2 (RED) core). To resonate with the 1.4 µH inductor requires 723 pF. But 470 pF/2, or 236 pF, already are in the circuit because of the series network C2/C3; hence, a variable capacitor (C4) of 723 pF − 236 pF, or 487 pF. An NE-602AN Hartley oscillator is shown in Figure 8.11. This circuit is identified by the tapped coil in the L-C network. The

+9 VDC

C8 0.1 µF Q1 MPF-102

C9 5 pF

R1 100K

C1 25 pF

Fig. 8.9 Series-tuned Clapp oscillator.

D1 1N4148

C4 0.001 µF

C5 0.001 µF

C2 10 pF

C7 33 pF

RFC1 1 mH

C3 10 pF R2 100

C6 0.01 µF

RF OUT

118

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

+5 VDC

R1 100

8

4 OUT 1 5

IC1 NE-602

C7 0.01 µF

OUT 2

1 R2 10K

7

3

2

6 C1

C5 0.01 µF

C2

C3

L1

C4

Fig. 8.10 NE-602AN/NE-612AN Colpitts oscillator.

+5 VDC

R1 100

8

4 OUT 1 5

IC1 NE-602

C6 0.01 µF

OUT 2

1 R2 10K

7

3

2

6

C4 0.01 µF

C1

C2 0.01 µF

C3 0.01 µF

L1

Fig. 8.11 NE-602AN Hartley oscillator.

value of the inductor is about 10 µH/FMHz, and is tapped from one fourth to one third the way from the ground end. The capacitor needs to resonate at the desired frequency. For our 5 MHz example, an inductor of 2 µH

is used, which means 20 turns of wire on a T-50-2 (RED) core. A voltage-tuned Clapp NE-602AN oscillator is shown in Figure 8.12. This circuit uses a varactor diode to set the operating frequency.

Local Oscillator and Frequency Synthesizer Circuits

+5 VDC

R1 100

8

119

4 OUT 1

C5 0.01 µF

5

IC1 NE-602

OUT 2

1 R2 10K

7

3

2

6

C4 0.01 µF

C1 100 pF

C3 0.01 µF

C2 100 pF

L1 R3 10K VT D1

Fig. 8.12 Voltage-tuned NE-602AN oscillator.

With the 100 pF capacitors shown, this circuit has oscillated from about 6 MHz to about 15 MHz, using an NTE-614 (33 pF) diode.

CRYSTAL OSCILLATORS Radio frequency oscillators can be built using a number of different types of frequency selective resonator. Common types include inductor-capacitor (L-C) networks and quartz crystal resonators. The crystal resonator has, by far, the best accuracy and stability, and it makes single-frequency receivers possible. Piezoelectric Crystals Certain naturally occurring and human-made materials exhibit the property of piezoelectricity: Rochelle salts, quartz, and tourmaline are examples. Rochelle salts crystals are not used for RF oscillators, although at one time they were used extensively for phonograph pick-up cartridges. Tourmaline crystals can be used for some RF applications but are not

often used due to high cost. Tourmaline is considered a semiprecious stone, so tourmaline crystals are more likely to wind up as gemstones in jewelry than in radio circuits. That leaves quartz as the preferred material for radio crystals. Figure 8.13 shows a typical natural quartz crystal. Actual crystals rarely have all the planes and facets shown. The three optical axes (X, Y, and Z) in the crystal are used to establish the geometry and locations of various cuts. The actual crystal segments used in RF circuits are sliced out of the main crystal. Some slices are taken along the optical axes, so are called Y-cut, X-cut, and Z-cut slabs. Others are taken from various sections and are given letter designations such as BT, BC, FT, AT, and so forth. Piezoelectricity All materials contain electrons and protons, but in most materials their alignment is random. This produces a net electrical potential in any one direction of zero. But, in crystalline

120

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

Z

GT

AT CUT

JT

HIGH FREQUENCY OSCILLATORS

ZERO TEMPERATURE COEFFICIENT

BT LOW FREQUENCY OSCILLATORS

Y HARMONIC OSCILLATORS

X

BC

ET

FT

CT

DT

AC

Y

X

Y CUT

NT

FILTERS

X CUT

MT

Fig. 8.13 Crystal structure.

materials, the atoms are lined up, so they can form electrical potentials. Piezoelectricity refers to the generation of electrical potentials due to mechanical deformation of the crystal. Figure 8.14 shows the piezoelectric effect. A zero-center voltmeter is connected across a crystal slab. In Figure 8.14A, the slab is at rest, so the potential across the surfaces is zero. But in Figure 8.14B, the crystal slab is deformed in the upward direction, and a positive potential is seen across the slab. When the crystal slab is deformed in the opposite direction, a negative voltage is noted. If the crystal is mechanically “pinged” once, it will vibrate back and forth, producing an oscillating potential across its terminals, at its resonant frequency. Due to losses, the oscillation will die out in short order. But,

Z if the crystal is repetitively pinged, then it will generate a sustained oscillation on its resonant frequency. It is not terribly practical to stand there with a tiny little hammer pinging the crystal all the while the oscillator is running, however. Fortunately, piezoelectricity also works in the reverse mode: If an electrical potential is applied across the slab it will deform. Therefore, if we amplify the output of the crystal and then feed back some of the amplified output to electrically “re-ping” the crystal, it will sustain oscillation on its resonant frequency. EQUIVALENT CIRCUIT Figure 8.15A shows the equivalent R-L-C circuit of a crystal resonator, and Figure 8.15B shows the impedance vs. frequency plot for

Local Oscillator and Frequency Synthesizer Circuits

121

A 0

B 0

C 0

Fig. 8.14 Crystals generate voltage when deflected.

the crystal. The equivalent circuit has four basic components: series inductance (Ls ), series resistance (Rs ), series capacitance (Cs ), and parallel capacitance (Cp ). Because there are two capacitances, there are two resonances: series and parallel. The series resonance point is where the impedance curve crosses the zero line, while parallel resonance occurs a bit higher on the curve. CRYSTAL PACKAGING Over the years a number of different packages have been used for crystals. Even today there are different styles. Figure 8.16A shows a representation of the largest class of packages. It is a hermetically sealed small metal

package, in various sizes. The actual quartz crystal slab is mounted on support struts inside the package (Figure 8.16B), which are in turn mounted to either a wire header or pins. Some crystals use pins for the electrical connections, typically mounted in sockets. The pin type of package can be soldered directly to a printed circuit board, but care must be taken to prevent fracturing the crystal with heat. Not all pins are easily soldered, although it can help if the pins are scraped to reveal fresh metal before soldering. Normally, however, if the crystal is soldered into the circuit, a wire-lead package is used. Some crystals may short circuit if installed on a printed circuit board with either

122

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

A

XL

CP

INDUCTIVE REACTANCE

LS

CS

RS

PARALLEL RESONANCE RANGE ANTIRESONANCE ESR

CAPACITIVE REACTANCE

FREQUENCY

B

FS

XC

Fig. 8.15 Crystal: (A) equivalent circuit; (B) impedance diagram.

through- (via-) holes or a ground plane on the top side of the board. In those cases, the usual practice is to insert a thin insulator between the PCB and the crystal (Figure 8.16C). Temperature Performance According to temperature performance, there are three basic categories of crystal oscillator: room temperature crystal oscillators (RTXO), temperature-compensated crystal oscillators (TCXO), and oven-controlled crystal oscillators (OCXO). We look at each of these groups. ROOM TEMPERATURE CRYSTAL OSCILLATORS The RTXO takes no special precautions about frequency drift. But, with proper selection of crystal cut and reasonable attention to construction, stability on the order of 2.5 parts per million (ppm), 2.5 × 10–6, over the

temperature range 0–50°C is possible. The RTXO is used only on economy model counters used for noncritical applications. TEMPERATURE-COMPENSATED CRYSTAL OSCILLATORS The TCXO circuit also works over the 0–50°C temperature range but is designed for much better stability. The temperature coefficients of certain components of the TCXO are designed to counter the drift of the crystal, so the overall stability is improved to 0.5 ppm (5 × 10–7). The cost of TCXOs has decreased markedly over the years to the point where relatively low-cost upgrades to economy receivers give them a rather respectable stability specification. OVEN-CONTROLLED CRYSTAL OSCILLATORS The best stability is achieved from the OCXO time base. These oscillators place the resonating crystal inside a heated oven that

Local Oscillator and Frequency Synthesizer Circuits

123

QUARTZ CRYSTAL METAL CASE

A

ELECTRODE

SUPPOR T STRUT

SUPPOR T STRUT INSULATOR

B

PINS

CRYSTAL

INSULATOR

C PCB

Fig. 8.16 Crystal package: (A) external view; (B) internal view; (C) use of insulator on double-sided printed wiring board.

keeps its operating temperature constant, usually near 70 or 80°C. Two forms of crystal oven are used in OCXO designs: on/off and proportional control. The on/off type is similar to the simple furnace control in houses. It has a snap action that turns the oven heater on when the internal chamber temperature drops below a certain minimum point and off when it rises

to a certain maximum point. The proportional control type operates the heating circuit continuously and supplies an amount of heating proportional to the actual temperature difference between the chamber and the set point. The on/off form of oven is capable of 0.1 ppm (10–7). OCXOs that use a proportional control oven can reach a stability of 0.0002 ppm

124

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

(2 × 10–10) with a 20-min warm-up and 0.0001 ppm (1.4 × 10–10) after 24 hours. It is common practice to design the counter to leave the OCTX turned on even when the counter is off. Some portable frequency counters, such as those used in two-way radio servicing, have a battery back-up to keep the OCXO turned on while the counter is in transit. This variation is referred to as the temperature stability of the counter time base. We also must consider short-term stability and long-term stability (aging). Short-Term Stability The short-term stability is the random frequency and phase variation due to noise that occurs in any oscillator circuit. It sometimes is called time domain stability or fractional frequency deviation. In practice, the shortterm stability has to be a type of rms (root mean square) value averaged over 1 sec. The short-term stability measure is given as σ(∆f/f )(t). Typical values of short-term stability for the different forms of clock oscillator are listed in Table 8.1. Long-Term Stability The long-term stability of the time base clock oscillator is due largely to crystal aging. The nature of the crystal, the quality of the crystal, and the plane from which the particular resonator was cut from the original quartz crystal are determining factors in defining aging. This figure is usually given in terms of Table 8.1 Short-Term Stability for the Different Forms of Clock Oscillator Root Mean Square

Parts per Million

RTXO

2 × 10–9 rms

0.002 ppm

TCXO

1 × 10–9 rms

0.001 ppm

OCXO (on/off)

5 × 10

0.0005 ppm

OCXO (proportional control)

1 × 10–11 rms

Oscillator

–10

rms

0.00001 ppm

Table 8.2

Long-Term Stability for the Different Forms of Clock Oscillator

Oscillator

Units per Month

Parts per Million

RTXO

3 × 10–7/month

0.3 ppm

TCXO

1 × 10 /month

0.1 ppm

OCXO (on/off)

1 × 10–7/month

0.1 ppm

OCXO (proportional control)

–7

1.5 × 10–8/month

0.015 ppm

5 × 10–10/day

0.0005 ppm

frequency units per month, as shown in Table 8.2. Oscillators The amplifier used in an oscillator circuit can be any of many different devices. In some circuits it is a common-emitter bipolar transistor (NPN or PNP devices). In others, it is JFET or MOSFET. In older equipment, it was a vacuum tube. In modern circuits, the active device probably is either an integrated circuit operational amplifier or some other form of linear IC amplifier. The amplifier most frequently is an inverting type, so the output is out of phase with the input by 180º. As a result, to obtain the required 360º phase shift, an additional phase shift of 180º must be provided in the feedback network at the frequency of oscillation only. If the network is designed to produce this phase shift at only one frequency, then the oscillator produces a sine wave output on that frequency. In a crystal oscillator, amplified noise in the circuit at start-up initiates the crystal oscillation, but the feedback voltage is used to continuously re-ping the crystal to keep it oscillating. Colpitts Crystal Oscillator Circuit Figure 8.17 shows a basic Colpitts oscillator circuit. The active element is an NPN bipolar transistor, although FET and IC versions will be shown in circuits to be discussed later.

Local Oscillator and Frequency Synthesizer Circuits

quire C1 = C2, while in other recommendations C2 = 3 × C1 or 4 × C1.

V+

R1 220K

C4 0.01 µF

Miller Oscillators

Q1

C1

C3 100 pF

Y1

OUTPUT C2

125

R2 1000

Fig. 8.17 NPN Colpitts crystal oscillator.

The transistor’s DC bias is derived from resistor R1 connected between V+ and the transistor base terminal. The crystal resonator is connected from the base to ground on this common emitter circuit. This circuit is usable from about 1–18 MHz or so. Because it is a Colpitts oscillator, the circuit uses a capacitive voltage divider (C1 and C2) for the feedback network. When this circuit initially is turned on, current begins to flow from collector to emitter. When this current first flows, the crystal is electrically pinged and begins to oscillate. Following initial start-up, a sample of the signal at the emitter is fed to C1/C2 and from there to the base. Because the emitter AC signal voltage is in phase with the base signal voltage, the Barkhausen requirements are met. Some experimentation will yield the optimum values of C1, C2, and R2 to ensure proper starting and running. The general rule for R1 is to use the lowest value that will permit sure starting when the circuit is powered up. Two approaches to finding values for C1 and C2 are presented in the literature, but both assume that the total capacitive reactance of the two capacitors in series is 300 Ω. In one scheme, the initial trial values re-

Miller oscillators are analogous to the tuned input/tuned output variable frequency oscillator, because they have a crystal at the input of the active device and an L-C tuned circuit at the output. Figure 8.18 shows a basic Miller circuit built with a JFET. Any common RF device can be used for Q1 (e.g., the MPF-102). DC bias is provided by R2, which places the source terminal at a potential above ground due to the channel current flowing in Q1. The source must be kept at ground potential for AC, so a bypass capacitor (C4) is provided. The reactance of this capacitor must be less than one tenth the value of R2 at the lowest intended frequency of operation. The output circuit of the oscillator is tuned by a parallel resonant L-C tank circuit, L1/C1. The tuned circuit must be adjusted to the resonant frequency of the oscillator, although best performance usually occurs at a frequency slightly removed from the crystal V+ C3 0.01 µF

C1

L1

Q1 C2 0.001 µF

Y1

R1 100K

R2 220

Fig. 8.18 JFET Miller crystal oscillator.

C4 0.01 µF

126

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

+12 VDC C4 0.01 µF

R4 47K

R3 100K

C7 OUTPUT

T1 C5

C6

RS 10 to 47 W D G2

Q1 40673

G1

C2 2200 pF C1 50 pF

Y1

R1 47K

D1 1N4148 R2 470

S

C3 330

Fig. 8.19 MOSFET Miller oscillator.

frequency. If you monitor the output signal level while adjusting either C1 or L1, you will note a distinct difference between the high side and the low side of the crystal frequency. Best operation usually occurs at the low side. Whichever is selected, however, care must be taken that the oscillator will start reliably when cold started. Output signal can be taken either from capacitor C2, as shown, or through a link coupling winding on L1. The Miller oscillator circuit of Figure 8.18 has the advantage of being simple to build but suffers from some problems. One problem is that the feedback is highly variable from one transistor to the next because it is created by the gate-drain capacitance of Q1. Also outputlevel variations have been noted, as well as frequency pulling, under output load impedance variations, which are not good attributes for an oscillator. Further, JFETs of the same type number and different crystals of the same type number from the same manufacturer can

show a large difference in starting ability. I also noted problems with this circuit when either the JFET or crystal ages. I have seen JFET oscillators that worked well and then failed. When the JFET was replaced, it started working again. What surprised me was that the JFET tested as good. An improved Miller oscillator is shown in Figure 8.19. This circuit uses a dual-gate MOSFET transistor, such as the 40673 device, as the active element. It is a fundamental mode oscillator that uses the parallel resonant frequency of the crystal. The crystal circuit is connected to Gate 1, while Gate 2 is biased to a DC level. This circuit can provide a stability of 15–20 ppm if AT-cut or BT-cut crystals are used. A problem that might be seen with this circuit is parasitic oscillation at VHF frequencies. The MOSFETs used typically have substantial gain at VHF, so could oscillate at any frequency where Barkhausen’s criteria are

127

Local Oscillator and Frequency Synthesizer Circuits

met. There are two approaches to solving this problem. One approach is to insert a ferrite bead on the lead of Gate 1 of the MOSFET. The ferrite bead acts like a VHF/UHF RF choke. The second approach is to insert a snubber resistor (RS in Figure 8.19) between the crystal and Gate 1 of the MOSFET. Usually, some value between 10 and 47 Ω will provide the necessary protection. Use the highest value that permits sure starting of the oscillator. One interesting aspect of the Miller oscillator of Figure 8.19 is that it can be used as a frequency multiplier (not to be confused with an overtone oscillator), if the tuned network in the drain circuit of Q1 is tuned to an integer multiple of the crystal frequency.

V+

C3

R1 C2

Y1

OUTPUT

Q1

C1

Fig. 8.20 Pierce oscillator.

Pierce Oscillators The Pierce oscillator is characterized by the crystal being connected between the output and input of the active device. Figure 8.20 shows the basic Pierce crystal oscillator circuit using a bipolar NPN transistor (e.g., 2N2222 or 2N5179). The crystal is connected directly from the collector to the base of Q1. Output is taken through capacitor C2 con-

nected to the collector. This circuit is used extensively in low-cost receiver circuits but is not recommended. An improved Pierce oscillator is shown in Figure 8.21. This circuit includes a capacitor (C1) for pulling the crystal a small amount to precisely tune the frequency. This circuit is designed for frequencies between 10 and 20

+5 to +12 VDC C5 0.01 µF

R2 1K

C4 100 pF OUTPUT

C1 47 pF

Y1 15 MHz

R1 56K

Q1 C2 180 pF

Fig. 8.21 Pierce oscillator for frequencies between 10 and 20 MHz.

C3 39 pF

128

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

V+

C6 0.1 µF

R4 2.2K R3 3.3K C1 60 pF

C5 0.1 µF R5 120K

OUTPUT

Y1

Fig. 8.22 Pierce oscillator for frequencies in the 50–500 kHz region.

C3 3900 pF

C2 100 pF

MHz with the capacitance values shown. If the output is lightly loaded (keep C4 small), then the oscillator will provide a reasonable output stability at a level of near 0 dBm. Figure 8.22 is a variation on the theme that will work in the 50–500 kHz region. This circuit is very similar to the earlier circuit except for increased capacitance values to account for the lower frequency. In both circuits, ordinary NPN devices such as the 2N2222 can be used successfully.

C4 3900 pF

R1 10K

R2 100

C4 0.01 µF V+

C1*

Q1 2N5770

Butler Oscillators The Butler oscillator looks superficially like the Colpitts in some manifestations (Figure 8.23). The difference is that the crystal is connected between the tap on the feedback network and the emitter of the transistor. This particular circuit is a series mode oscillator. The value of R1 should be whatever value between 100 and 1000 Ω will result in reliable oscillation and starting while minimizing crystal dissipation. A table of capacitance values for feedback network C1/C2 is provided. For the 3–10 MHz range, use C1 = 47 pF and C2 = 390 pF; for 10–20 MHz, select C1 = 22 pF and C2 = 220 pF.

OUTPUT

L1

R3 15K

C3 0.01 µF

Y1

R2 5.6K

R1 330

C2*

FREQ (MHz)

C1 (pF)

C1 (pF)

3 - 10

47

390

10 - 20

22

220

Fig. 8.23 Butler NPN transistor oscillator.

129

Local Oscillator and Frequency Synthesizer Circuits

The collector circuit is tuned by the combination of C1 and L1. This circuit may well oscillate with the crystal shorted, and care must be taken to ensure that the “free” oscillation and the crystal oscillation frequencies are the same. The crystal should take over oscillation when it is in the circuit. The Butler oscillator of Figure 8.23 is capable of 10–20 ppm stability if a buffer amplifier with good isolation is provided at the output. Otherwise, some frequency pulling with load variations might be noted. The output signal is taken from a coupling winding over L1. This winding typically is only a few turns of wire on one end of L1. Alternatively, a tap on L1 might be provided and connected to a low-value capacitor. This approach might change some resonances unless care is taken. Another alternative output scheme is to connect a small value capacitor to the collector of Q1. Keep the value low to reduce loading and also the effects of the output capacitor on the resonance of L1/C1. A somewhat more complex Butler oscillator is shown in Figure 8.24. This circuit is sometimes called an aperiodic oscillator circuit. It uses two additional transistors to provide buffering and also as part of the feedback

circuit. This circuit operates in the frequency range from about 300 kHz to 10 MHz, although some care must be taken in the selection of the transistor. Many low-frequency crystals exhibit a lower equivalent series resistance in one of the higher frequency modes of oscillation than in the fundamental mode, so you might find this circuit oscillating at some frequency in the medium wave or HF region, rather than the LF. The key to preventing this problem is to use a transistor with a lower gain-bandwidth product (e.g., 2N3565 or equivalent). At this point it might be wise to point an aspect of “universal” replacement lines of transistors. Because crystal oscillators may operate in an unwanted overtone mode (i.e., at a higher frequency) or due to stray L-C components they may parasitically oscillate on a VHF or UHF frequency, you will want to keep the gain-bandwidth (GBW) product of the active device low. But many replacement lines use a single high-frequency transistor with similar gain, collector current, and power dissipation ratings as a “one size fits all” replacement for transistors with lower GBW products. I have seen this happen with service replacements on older equipment. The original V+

C4 0.1 µF C7 0.01 µF R8 12K

R6 1.5K R2 27K

C5 0.1 µF

R4 22K

C6 0.001 µF OUTPUT R7 12K

C1 50 pF Y1

C3 0.01 µF

R1 10K

R3 6.8K

C2 15 pF

R5 1.5K

Fig. 8.24 Butler oscillator with an aperiodic oscillator circuit.

R9 3.3K

130

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

component may not be available, so a universal, service shop replacement line device is selected. Then that parasitic oscillations and other problems are discovered because the new replacement has a GBW of 200 MHz, where the old device was a 50 MHz transistor. This problem can show up especially severely in RF amplifiers and low-frequency oscillators where L-C components naturally exist or any circuit where the stray and distributed L-C elements provide the required phase shift on some frequency above the unity GBW point. The circuit of Figure 8.24 produces a sine wave output but not without relatively strong harmonic output. The second and third harmonics are particularly evident. However, if harmonics are desired (as when the oscillator is used in a frequency multiplier circuit), then strong harmonics up to 30

MHz can be generated from a 100 kHz crystal if R5 is reduced to about 1000 Ω. The output of this oscillator is taken through an emitter follower buffer stage. This circuit can be used as a general buffer for a number of oscillator circuits. It generally is a good practice to use a buffer amplifier with any oscillator to reduce loading and smooth out load impedance variations. Another variation on the Butler theme is shown in Figure 8.25. This circuit is similar to Figure 8.24 but a bit less sensitive to frequency pulling due to DC power supply voltage variations. However, it is good engineering practice to use a separate voltage regulator for all oscillator circuits to prevent such variation. The availability of low-cost, three-terminal integrated circuit voltage regulators makes prevention quite easy.

+12 VDC

C6 0.01 µF

R1 3.9K Q2 R2 470

Q3

Q1

C5 0.001 µF

C1 50 pF

OUTPUT

Y1

C3 0.1 µF

R3 1K

C2 33 pF

R7 470

R5 1K

R6 1K

C4 0.1 µF

R4 27K

Fig. 8.25 Butler oscillator that is a bit less sensitive to frequency pulling.

131

Local Oscillator and Frequency Synthesizer Circuits

An improved Butler oscillator is shown in Figure 8.26. Again, this circuit is based on Figure 8.24. Both circuits can be used at frequencies from LF up to the mid-HF region (about 12–15 MHz) if appropriate values of R3 and R5 are used. The improvement of Figure 8.27 over Figure 8.24 is the limiting diodes (D1 and D2) provided between the two oscillator transistors (Q1 and Q2). These diodes can be ordinary 1N4148 small-signal diodes. The circuit of Figure 8.26 is more stable than Figure 8.24 because crystal dissipation is limited and it has more sure cold starting. The Butler oscillators are series mode circuits but, because of the series capacitors, can use parallel mode crystals. For a strictly series mode circuit, eliminate the capacitors in series with the crystal and replace it with a short circuit.

gate capacitances of Q1. This circuit can be used with parallel-mode crystals from about 3 to 20 MHz with proper values of C1 and C2 (see table provided in Figure 8.27). Frequency trimming of the oscillator can be done by shunting a small value trimmer capacitor across the crystal. The trimmer also can be placed in series with the crystal. If the oscillator tends to oscillate parasitically in the VHF region, try using the snubber resistor method (R4 in Figure 8.27). This could occur because the JFET used at Q1 has sufficient gain at VHF to permit Barkhausen’s criteria to operate at some frequency where strays and distributed L-C elements produce the correct phase shift. A value between 10 and 47 Ω usually eliminates the problem. Alternatively, a small ferrite bead can be slipped over the gate terminal of Q1 to act as a small-value VHF/UHF RF choke. Figure 8.28 is very similar to Figure 8.27, except for two features. First, the active device is an n-channel MOSFET rather than a JFET. Any of the single-gate devices (e.g., 3N128) can be used. One must, however, be mindful of the possibility of electrostatic damage when the MOSFET is used.

Colpitts Oscillators The Colpitts oscillator is characterized by a feedback network consisting of a tapped capacitive voltage divider. In Figure 8.27, the feedback is provided by C1 and C2, although the situation is somewhat modified by the

R10 100 V+

C4 0.1 µF C7 0.01 µF C8 0.22 µF

R2 27K

R8 12K

R6 1.5K

C5 0.1 µF

R4 22K D2

Q3 C6 0.001 µF

D1

OUTPUT

Q2 R7 12K

C1 50 pF

Q1

R9 3.3K

Y1

C3 0.01 µF

R1 10K

R3 6.8K

C2 15 pF

R5 1.5K

Fig. 8.26 Butler oscillator with limiting diodes between the two oscillator transistors.

132

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

V+

R3 150 C4

R4 (See Text)

Q1 MPF-102

C3 0.001 µF

C1* C5 50 pF

Y1

R1 47K

OUTPUT C2*

Fig. 8.27 JFET Colpitts oscillator.

R2 1K

FREQ (MHz)

C1 (pF)

C2 (pF)

3 - 10

27

68

10 - 20

10

27

V+

R3 150 C4

Q1 3N128

C5 50 pF

D1 1N4148 Y1

C3 0.001 µF

C1* R1 1 MEG

OUTPUT C2*

Fig. 8.28 MOSFET Colpitts oscillator.

R2 1K

FREQ (MHz)

C1 (pF)

C2 (pF)

3 - 10

22

180

10 - 20

10

82

Local Oscillator and Frequency Synthesizer Circuits

The other difference is that a 1N4148 small-signal diode is used shunting the gatesource path to provide a small amount of automatic gain control. When the signal appearing across the crystal and feedback network is sufficiently large, the diode rectifies the signal and produces a DC bias on the gate that counters the source bias provided by R2. This diode helps smooth out amplitude variations, especially when more than one crystal is switched in and out of the circuit. Another variation on the Colpitts theme is the impedance inverting oscillator circuit of Figure 8.29. It will provide stability of 10 ppm over a wide temperature range (0–60°C) if a wise selection of components is made (C1–C3 and L1 are particularly troublesome). The circuit will also remain within ±0.001% over a DC power supply variation of 2:1 (provided

the crystal dissipation is not exceeded). Harmonic output of the circuit typically is low. The oscillating frequency is set by adjusting inductor L1. The turn counts shown in the table in Figure 8.29 presume a 6.5-mm slug-tuned coil form designed for use in the frequency range 3–20 MHz. Some experimentation is needed, depending on the particular form used. The idea is to set the resonant frequency of the coil and C1–C3 combined to something near the crystal frequency. Sometimes, you want to add a tuned circuit to the output circuit of oscillators. The harmonics of the oscillator are suppressed when this is done. But, in this case, a transistor equivalent of the old-fashioned TGTP oscillator will result, because of the action of the output tuned circuit and the L1/C1–C3 combination. Do not do it.

V+ C6 0.01 µF

C4 0.01 µF R3 15K

C1*

Q1 2N5770 L1*

C5 47 pF

C2* R2 6.8K

R4 560

OUTPUT C3*

Y1

R1*

FREQ (MHz)

C1 (pF)

C2 (pF)

C3 (pF)

R1

L1

2 - 4 MHz

60 turns

4 - 6 MHz

40 turns

6 - 10 MHz

25 turns

3 - 10

1000

270

270

1.5K

10 - 15

100

220

220

680

15 turns

15 - 20

100

100

100

680

10 turns

Fig. 8.29 Tunable NPN Colpitts crystal oscillator.

133

134

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

Overtone Oscillators Thus far only the fundamental oscillating mode has been discussed. But crystals oscillate at more than one frequency. The oscillations of a crystal slab are in the form of bulk acoustic waves and can occur at any wave frequency that produces an odd halfwavelength of the crystal’s physical dimensions (e.g., λ/2, 3λ/2, 5λ/2, 7λ/2, 9λ/2, where the fundamental mode is λ/2). Note that these frequencies are not harmonics of the fundamental mode but actually valid oscillation modes for the crystal slab. The frequencies fall close to but not directly on some of the harmonics of the fundamental mode (which probably accounts for the confusion). The overtone frequency will be marked on the crystal rather than the fundamental mode (it is rare to find fundamental mode crystals above 20 MHz or so, because their thinness makes them more likely to fracture at low values of power dissipation). The problem to solve in an overtone oscillator is encouraging oscillation on the correct

overtone while squelching oscillations at the fundamental mode and undesired overtones. Crystal manufacturers can help with correct methods, but the oscillator designer still bears responsibility for solving this problem. Figure 8.30 shows a third-overtone Butler oscillator that will operate at frequencies between 15 and 65 MHz. The inductor (L1) is set to resonate close to the crystal frequency and is used in part to ensure overtone mode oscillation. If moderate DC supply voltages are used (e.g., 9–12 V in most cases), the harmonic content is low (−40 dB) and stability is at least as good as a similar fundamental mode Butler oscillator. Figure 8.31 is a third-overtone impedance inverting Colpitts-style oscillator that will operate over the 15–65 MHz range. As in similar circuits, inductor L1 is tuned to the overtone and resonated with C1 (combined with the capacitances of C2 and C3). Values for C1 through C3 and winding instructions for a 6.5-mm low-band VHF coil former are shown in the table. Note the resistor across crystal Y1. This resistor tends to snub out oscillations +12 VDC

C6 0.01 µF RFC1 56 µH

R1 15K

C5 0.001 µF

L1

OUTPUT

Y1 C7 0.001 µF

R2 4.7K

R3 220

Fig. 8.30 Third-overtone Butler oscillator.

C1 30 pF

C2 22 pF

C3 180 pF

C4 100 pF

Local Oscillator and Frequency Synthesizer Circuits

135

V+ C5 0.001 µF

L1

R3 15K

C1

Q1 2N5770 C2 R1 560

C4

R2 5.6K

Y1

OUTPUT C3

R4 470

FREQ (MHz)

C1 (pF)

C2 (pF)

C3 (pF)

C4 (pF)

L1 (0.25-inch form)

15 - 25

100

100

68

33

12 t, #30, CW

25 - 55

100

68

47

33

8 t, #30, CW

50 - 65

68

33

15

22

6 t, #22, CW

Fig. 8.31 Third-overtone Colpitts oscillator.

in modes other than the overtone, including the fundamental mode. Care must be taken to not make L1 too large, otherwise it will resonate at a lower frequency (with C1–C3), forming an oscillator on a frequency not related to either the crystal’s fundamental mode or overtones. The oscillator may perfectly well act as a seriestuned Clapp oscillator. Operation of the circuit of Figure 8.31 to 110-MHz, with fifthor seventh-overtone crystals, can be accomplished by modifying this circuit to the form shown in Figure 8.32.

TACTICS TO IMPROVE OSCILLATOR ACCURACY AND STABILITY An ideal sine wave RF oscillator produces nice, clean harmonic and noise-free output on a stable frequency. But real RF oscillators

tend to have certain problems that deteriorate from the quality of their output signals. Load impedance variation and DC power supply variation can cause the oscillator frequency to shift. Temperature changes also cause frequency change problems. Noise that modulates the oscillator, generated externally or internally, produces phase noise sidebands around the oscillator signal. An oscillator that changes frequency without any help from the operator is said to drift. Frequency stability generally refers to freedom from frequency changes over a relatively short period of time (e.g., few seconds to dozens of minutes). This problem is different from aging, which is frequency change over relatively long periods of time (i.e., Hz/yr) caused by the aging of the components (some electronic components tend to change value with long use). Temperature changes are a large contributor to this problem in two

136

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

+12 VDC C6 0.001 µF RFC1 22 µH R1 10K

L1*

C5 1000 pF OUTPUT

L2* C1 Q1 2N5770

R4*

Y1

R2 2.7K

C4 1000 pF

R3 220

C3 22 pF

C2

FREQ (MHz)

C1 (pF)

C2 (pF)

C3 (pF)

L1

L2

65 - 85

15

150

100

7 t., #24, 3/16 in. CW

10 t. #34 over 10-ohm 1/4-W

85 - 110

10

100

68

4 t., #24, 3/16 in. 1WD

10 t. #34 over 10-ohm 1/4-W

CW = CLOSE WOUND 1WD = SPACED 1 WIRE DIAMETER

Fig. 8.32 Modification of crystal to operate at 65 to 110 MHz, with fifth- or seventh-overtone crystals.

forms: warm-up drift (i.e., drift in first 15 min) and ambient temperature change drift. If certain guidelines are followed, then it is possible to build a very stable oscillator. For the most part, the comments that follow apply to both crystal oscillators and L-C tuned oscillators (e.g., variable frequency oscillators), although in some cases one or the other is indicated by the text. Temperature Temperature variation has a tremendous effect on oscillator stability. Avoid locating the oscillator circuit near any source of heat within the equipment it serves. In other words, keep it away from power transistors

or IC devices, voltage regulators, rectifiers, lamps, or other sources of heat. THERMAL ISOLATION One approach is to thermally isolate the oscillator circuits. Figure 8.33 shows one such method. The oscillator is built inside a metal shielded cabinet, as usual, but has styrofoam insulation applied to the sides. One type of poster board has a backing of styrofoam to give it substance enough for selfsupport. It is easy to cut using the hobby or razor knives. Cut the pieces to size, then glue them to the metal surface of the shielded cabinet, using contact cement or some other form of cement that will adhere to metal and paper.

Local Oscillator and Frequency Synthesizer Circuits

VARIABLE CAPACITOR

MAIN INDUCTOR

137

PRINTED CIRCUIT BOARD

RF CONNECTOR

Fig. 8.33 Thermal treatment of a VFO.

STYROFOAM PANEL INSULATION

CRYSTAL OVENS It is common practice to use a constanttemperature oven (Figure 8.34) for crystal oscillators, when very good temperature stability is required. The oven keeps the crystal at a constant temperature of 75 or 80°C. There are two basic oven forms: snap action and proportional. The snap action oven uses a thermal switch that turns on and off at preset temperatures, much like the thermometer in a centrally heated house. The proportional oven uses a tem-

CIRCUITRY & TEMP SENSOR

CRYSTAL

Fig. 8.34 Crystal ovens.

METAL SHIELDED BOX

perature sensor and circuitry to provide heating as needed to keep the temperature stable. A number of integrated circuits are available that serve as proportional controllers. Both socket-mounted and printedcircuit ovens are available (often from crystal manufacturers). SELF-HEATING Operate the oscillator at as low a power level as can be tolerated to prevent self-heating of the active device and associated frequency-

PCB CRYSTAL OVEN

CRYSTAL

XTAL SOCKET

138

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

determining components. It generally is agreed that a power level on the order of 1–10 mW is sufficient. If higher power is needed, then a buffer amplifier can be used. The buffer amplifier also isolates the oscillator from load variations. Other Criteria USE LOW FREQUENCIES In general, L-C-controlled VFOs should not be operated at frequencies above about 12 MHz. For higher frequencies, it is better to use a lower-frequency VFO and heterodyne it against a crystal oscillator to produce the higher frequency. For example, one common combination uses a 5–5.5 MHz main VFO for all HF bands in SSB transceivers. FEEDBACK LEVEL Use only as much feedback in the oscillator as is needed to ensure that the oscillator starts quickly when turned on, stays operational, and does not “pull” in frequency when the load impedance changes. In some cases, a small-value capacitor is located between the L-C resonant circuit and the gate or base of the active device. The small-value capacitor is used to prevent drift by lightly loading the tuned circuit. The most common means for doing this job is to use a 3-12 pF NP0 disk ceramic capacitor. Adjust its value to the minimum level that ensures good start-

ing and freedom from frequency changes under varying load conditions. OUTPUT ISOLATION A buffer amplifier, even if it is a unity gain emitter follower, is highly recommended. It permits the oscillator signal power to build up, if needed, without loading the oscillator. The principal use of the buffer is to isolate the oscillator from variations in the output load conditions. Figures 8.35 and 8.36 show typical buffer amplifier circuits. The circuit in Figure 8.35 is a standard emitter follower (i.e., common collector) circuit. It produces near unity voltage gain, but some power gain. The buffer amplifier shown in Figure 8.36 is a feedback amplifier using a 4:1 BALUN-style toroidal core transformer in the collector circuit. This circuit exhibits an input impedance near 50 Ω, so it should be used only if the oscillator circuit can drive 50 Ω. In some cases, a higher impedance circuit is needed. DC POWER SUPPLY Power supply voltage variations have a tendency to frequency modulate the oscillator signal. Because dynamic circuit conditions often result in a momentary transient droop in the supply voltage and line voltage variations can cause both transient drops and peaks, it is a good idea to use a voltageregulated DC power supply on the oscillator. V+

C7 0.01 µF R8 12K C5 0.1 µF INPUT

C6 0.001 µF OUTPUT R7 12K

Fig. 8.35 Standard postamplifier circuit.

R9 3.3K

Local Oscillator and Frequency Synthesizer Circuits

C5 0.1 µF

139

+12 VDC

R6 150

C4 0.1 µF

R7 68

C6 0.1 µF OUTPUT

R2 3.3K

C3 0.1 µF

T1

R1 560 C1 0.1 µF

Q1 2N5179

INPUT

R4 10 R3 1K

Fig. 8.36 Postamplifier circuit with 50 Ω input/output impedances.

It also is a very good idea to use a voltage regulator to serve the oscillator alone, even if another voltage regulator is used to regulate the voltage applied to other circuits (see Figure 8.37). Although this double-regulation approach may have been a cost burden, it is reasonable given the cost of three-terminal IC voltage regulators and the advantage gained. For most low-powered oscillators, a simple low-power L-series (e.g., 78L06) three-terminal integrated circuit voltage regulator is sufficient (see U2 in Figure 8.37). The L-series devices provide up to 100 mA of current at the specified voltage, enough for most oscillator circuits. Capacitors on both the input and output sides of the voltage regulator (C1 and C2 in Figure 8.37) add further protection from noise and transients. The value of these capacitors is selected according to the amount

R5 100

C2 0.1 µF

of current drawn. The idea is to have a local supply of stored current to temporarily handle sudden demand changes, allowing time for a transient to pass or the regulator to “catch up” with the changed situation. VIBRATION ISOLATION The frequency setting components of the oscillator can affect the stability performance, too. The inductor should be mounted to prevent vibration. While this requirement means different things to different styles of coil, it nonetheless is important. Some people prefer to mount coils rigidly, while others will put them on a vibration isolating shock absorber material. I have seen a receiver that required a very low SSB noise sideband local oscillator be unable to meet specifications until some rubber shock absorbing material was placed underneath the oscillator printed

140

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

C5 0.1 µF

OTHER CIRCUITS

VIN

+15 TO +18 V

C4 100 µF

+

U1 78L12 GND

VOUT

C1 2.2 µF

VIN

+

U2 78L08 GND

VOUT

C2 1 µF

+

C3 0.1 µF

OSCILLATOR CIRCUIT

0V

Fig. 8.37 Power supply scheme using a voltage regulator.

wiring board. Any time vibration affects crystals, inductors, or capacitors, high vibrationinduced noise on the oscillator output signal is possible. COIL CORE SELECTION Air core coils generally are considered superior to those with either ferrite or powdered iron cores, because the magnetic properties of the cores are affected by temperature variation. Of those coils that do use cores, the slug tuned are said to be best, because they can be operated with only a small amount of the tuning core actually inside the windings of the coil, reducing the vulnerability to temperature effects. Still, toroidal cores have a certain endearing charm and can be used wherever the ambient temperature is relatively constant. The type-SF material is said to be the best in this regard, and it is easily available. Or use type-6 material. For example, you could wind a T-506 core and expect relatively good frequency stability. COIL-CORE PROCESSING One source recommends tightly winding the coil wire onto the toroidal core, then heat an-

nealing the assembly. This means placing it in boiling water for several minutes, then removing and allowing it to cool in ambient room air while it sits on an insulated pad. I have not tried this, but the source reported remarkable freedom from inductor-caused thermal drift. For most applications, especially where the temperature is relatively stable, the coil with a magnetic core can be wound from enameled wire (#20–#32 AWG usually is specified), but for best stability Litz wire is recommended. Although a bit hard to get in small quantities, this wire offers superior performance over relatively wide changes in temperature. Be aware that this nickel-based wire is difficult to solder properly, so be prepared for a bit of frustration. AIR CORE COILS For air core coils, use #18 SWG or larger bare solid wire, wound on a low-temperature coefficient of expansion formers. Figure 8.38 shows how the coil can be mounted in a project. Stand-off insulators provided adequate clearance for the coil and hold it to the chassis. The mount shown in Figure 8.38 has shock absorbers to prevent movement.

Local Oscillator and Frequency Synthesizer Circuits

141

selected to create a counterdrift that cancels out the natural drift of the circuit. The main tuning air variable capacitor should be an old-fashioned double bearing (i.e., a bearing surface on each end plate). For best results, it should be made with either brass or iron (not aluminum) stator and rotor plates. The capacitor should be rugged. TEMPCO CIRCUIT Temperature-compensated crystal oscillators can be built by using the temperature coefficient of some of the capacitors in the circuit to cancel drift in the opposite direction. The circuit in Figure 8.39 was used in an radio transceiver at one time. It is a standard Colpitts crystal oscillator, but with the addition of a temperature coefficient cancellation circuit (C8A, C8B, C9, and C10). The key to this circuit is C8, which is a differential variable capacitor; that is, it contains two sections connected reciprocally so that one is increasing capacitance as the other is decreasing capacitance as the shaft turns. In most cases, C9 and C10 are of equal value, but one is NP0 and the other N750 or N1500. If C8 is differential and C9 = C10, then the net capacitance across crystal Y1 is constant with changes in C8 shaft rotation. But the net temperature coefficient does change. The idea is to crank in the amount of temperature coefficient that exactly cancels the drift in the circuit’s other components.

Fig. 8.38 Air coil mounting.

CAPACITOR SELECTION The trimmer capacitors used in the oscillator circuit should be air dielectric types, rather than ceramic or mica dielectric trimmers, because of the lower temperature coefficient. The small fixed capacitors used in the oscillator should be either NP0 disk ceramics (i.e., zero temperature coefficient), silvered mica, or polystyrene types. Some people dislike the silvered mica types because they tend to be a bit quirky with respect to the temperature coefficient. Even samples from the same batch can have widely differing temperature coefficients on either side of 0. Sometimes, fixed capacitors will come with other than a zero temperature coefficient in an oscillator frequency-determining circuit. These are done to make temperaturecompensated oscillators. The temperature coefficients of certain critical capacitors are

U1 78L05

R2 10K

C4 0.01 µF

+9 to +12 VDC

C5 1 µF

C6 10 pF Q1

C8A

C8B

C1 Y1

C9 NPO

C10 N750

C7 50 pF

C3 100 pF OUTPUT TO BUFFER

R1 10K C2

Fig. 8.39 Temperature compensation of a crystal oscillator.

R3 470

142

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

R3 10K

R2 22

R1

C1 0.01 µF L1

V+

U1 MVS460/ ATK33B

C3 0.001 µF

+

C2 4.7 µF D1 VARACTOR

Fig. 8.40 Temperature-compensating IC.

The problem with this circuit is that differential capacitors are quite expensive. It may be useful, however, to connect the circuit as shown and find the correct setting. Once the setting is determined, disconnect C8–C10 without losing the setting and measure the capacitance of the two sections. With this information you can calculate the capacitances required for each temperature coefficient and replace the network with appropriately selected fixed capacitors. In general, when selecting temperature compensating capacitance, keep in mind the following relationship for the temperature coefficient of frequency (TCF):  T × C 1 TC 2 × C 1  − TCL + C 1 +  C Total C Total  (8.4)  TCF = 2

where TCF is the temperature coefficient of frequency; TCL is the temperature coefficient of the inductor; TC1 is the temperature coefficient of C1; TC2 is the temperature coefficient of C2; CTotal is the sum of all capacitances in parallel with L1, including C1, C2, and all strays and other capacitors. This equation assumes that two capacitors, C1 and C2, are shunted across an inductor in a VFO circuit.

VARACTORS Voltage variable capacitance diodes (varactors) often are used as a replacement for the main tuning capacitor. If this is done, then it becomes critical to control the environmental temperature of the oscillator. Temperature variations seem to result in changes in diode pin junction capacitance, and that contributes much to thermal drift. Varactor temperature coefficients of 450 ppm are not unusual. A good way to stabilize these circuits is to supply a voltage regulator that has a temperature coefficient opposite in direction to the varactor drift. By matching these two, it is possible to cancel the drift of the varactor. Figure 8.40 shows a basic circuit that will accomplish this job. Diode D1, in series with DC blocking capacitor C1, is in parallel with the inductor (L1). These components are connected into an oscillator circuit (not shown for sake of simplicity). Normally, resistor R3 (10–100 KΩ, typically) is used to isolate the diode from the tuning voltage source and is connected to the wiper of a potentiometer that varies the voltage. In this circuit, however, an MVS-460 (USA type number) or ZTK33B (European type number) varicap voltage stabilizer is present. This device is similar to a zener diode but has the required −2.3 mV/°C temperature coefficient. It operates at a current of 5 mA (0.005 A) at a supply voltage of

Local Oscillator and Frequency Synthesizer Circuits

34–40 V DC. The value of resistor R1 is found from R1 =

(V + ) − 33 Volts 0.005 Amps

(8.5)

For example, when V+ is 40 V, then the resistor should be 1400 Ω (use 1.5 KΩ).

FREQUENCY SYNTHESIZERS In modern radio receivers, the local oscillator usually is a frequency synthesizer. In general, the synthesizer outputs stable phase-coherent signals derived from a master oscillator. The master oscillator may be a high-precision crystal oscillator or an atomic standard (such as a rubidium gas cell or cesium atomic beam). The characteristics that must be specified for the frequency synthesizer include frequency range, frequency resolution, maximum frequency error, settling time, frequency indication method, reference frequency, harmonic distortion, SSB phase noise, discrete spurious frequencies (“spurs”), and wideband noise. Figure 8.41 shows a single-loop, phaselocked loop (PLL) frequency synthesizer. The signal is generated in a voltage-controlled oscillator (VCO) circuit. The output of the VCO is directed to a divide-by-N counter circuit, as well as being output. The output of the divideby-N counter is applied to one input of a

143

phase detector circuit. The other input of the phase detector receives the reference oscillator circuit. This oscillator may be a crystal oscillator with a frequency divider output, to produce the reference signal. The output of the phase detector is an error signal proportional to Fn − FR, where Fn is the VCO signal and FR is the reference signal. This signal is processed in a low-pass filter, and often a DC amplifier, before being applied to the control voltage input of the VCO. Thus, the VCO frequency is pulled to the reference oscillator frequency. Consider an example of a receiver designed to operate on 162.55 MHz. If the voltage controlled oscillator works between 155 and 165 MHz, a frequency division ratio in the divide-by-N counter of 16,255 will reduce the operating frequency to 1 kHz. Some signal sources produce a single output frequency (or a discrete number of fixed output frequencies). These receivers may be used for channelized receiver systems. Other signal sources produce output over a very wide range of frequencies. Output Signal Quality It would be nice if all signal sources were ideal; that is, the output frequency and output level were noiseless and perfectly calibrated. This never occurs, although the differences in these specifications are between high- and low-quality receivers.

REFERENCE OSCILLATOR FOUT

VOLTAGE CONTROLLED OSCILLATOR

DIVIDE-BY-N COUNTER

DC AMPLIFIER

Fig. 8.41 PLL synthesizer block diagram.

LOW PASS FILTER

PHASE DETECTOR

144

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

Frequency The important considerations regarding frequency are the range, resolution, accuracy, and (in certain receivers applications) the switching speed.

RESOLUTION The resolution is the statement of the smallest increment of frequency that can be set. On analog receivers have no counter, the resolution is poor. The resolution may (but not certainly) be improved by adding a digital frequency counter to measure the output frequency. On modern synthesizers, it is possible to set frequency with extremely good resolution. ACCURACY Accuracy refers to how nearly the actual output frequency matches the set frequency. It is a function of the set frequency (and how closely it can be set), FSet; long-term aging (τaging); and the time since the last calibration (τcal). Mathematically, Accuracy = ±FSet × τAging × τCal

(8.6)

For example, the receiver is set to 480 MHz and has an aging rate of 0.155 ppm/yr. It has been six months (0.5 yr) since the last calibration. Accuracy = ±FSet × τAging × τCal = ± (480-MHz) × (0.155 ppm/yr) × (0.5 yr) = ± 37.2 Hz.

There also may be some random variation in the output frequency. Figure 8.42 shows the uncertainty band around the set frequency. The actual output frequency, Fo, will be FSet ± Accuracy. The general practice is to calibrate a receiver on six-month or annual schedules, depending on the use.

AMPLITUDE

RANGE The frequency range is a specification that tells the specific frequencies covered. In some cases, only one frequency, or some small number of discrete frequencies, is covered. In other cases, one or more bands of frequencies are provided.

UNCERTAINTY

FSET

F

FREQUENCY

Fig. 8.42 Band of amplitude vs. frequency with uncertainty.

SWITCHING SPEED (SETTLING TIME) The settling time is the length of time, usually in milliseconds or microseconds, required for a synthesized signal source to move to a new frequency when digitally commanded to change. It is calculated as the length of time for the error of the frequency or output level commanded by the change to come into specification range. Output Level The output level can be expressed in voltage, power, or dBm notation. All are equivalent, although one or the other will be preferred in most cases. The most common method of describing the output level is in dBm. As with frequency, some factors affect the accuracy of the actual output vs. the set output. Spectral Purity The output signal is not always nice and clean. Although the purity of the output sig-

Local Oscillator and Frequency Synthesizer Circuits

nal is a distinguishing factor that differentiates lower- and higher-quality generators, all produce signals other than the one desired. Figure 8.43A shows a typical spectrum output. This display is what might be seen on a

spectrum analyzer. The main signal is a CW sine wave, so ideally we would expect only a single spike, with a height proportional to the output level. But, a lot of other signals are present.

AMPLITUDE

MAIN SIGNAL

~ 30 dB ~ -45 dB ~ -65 dB HARMONIC SUBHARMONIC

A

PHASE NOISE

nF/2

SPUR

F

FOther

nF

FREQUENCY

RESIDUAL FM IS INTEGRATED PHASE NOISE OVER SPECIFIED BANDWIDTH

FO

B -3000

-300

145

+300

+3000

Fig. 8.43 Spectral purity: (A) spurs and phase noise on an LO signal; (B) residual FM.

146

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

First, note that the main signal is spread out by phase noise. This noise is random variation around the main frequency. When integrated over a specified bandwidth, such as 300–3000 Hz, the phase noise is called residual FM (Figure 8.43B). Second, harmonics are present. If the main signal has a frequency of F, the harmonics have frequencies of nF, where n is an integer. For example, the second harmonic is 2F, and the third harmonic is 3F. In many cases, the 3F harmonic is stronger than the 2F harmonic, although in general higher harmonics are weaker than lower harmonics. There sometimes also are subharmonics. These are integer quotients of the main signal. Again, if the F is the main signal frequency, nF/2 represents the subharmonics. Typically, unless something interferes with the output signal, subharmonics are not as prominent. One thing that does make subharmonics prominent, however, is the use of frequency multiplier or divider stages (which is the case in many modern generators). Finally, miscellaneous spurious signals (spurs) are found on some generators. These might be due to power supply ripple modulating the output signal, parasitic oscillations, digital noise from counter- or phase-locked loop circuits, and other sources. Harmonics and spurs usually are measured in terms of decibels below the carrier (dBc), where the carrier is the amplitude of the main output signal. In general, the fewer are the unwanted components, the better the signal source. Phase noise warrants some special consideration. It usually is measured in terms of dBc/Hz; that is, decibels below the carrier per hertz of bandwidth. This noise is concentrated around the main signal frequency and normally graphed on a log-log scale to permit both close-in and further-out noise components to be compressed on one graph. Architectures A synthesizer architecture is shown in Figure 8.44. This modern signal generator is

capable of producing very accurate, highquality signals. Three main sections compose the signal source: reference section, frequency synthesizer, and output section. In Figure 8.44, each section is broken down into further components for the sake of easy analysis. REFERENCE SECTION The reference section is at the very core of the signal generation process. It is an accurate, stable fixed-frequency source such as a crystal oscillator. The frequency of the reference section must be precisely adjustable over a small range so it can be compared to a higher-order standard, such as a cesium beam oscillator or WWVB comparator receiver, for calibration. Because it controls the frequency synthesizer, the stability of the reference section determines the overall stability of the signal generator. The stability of ordinary crystal oscillators is reasonably good for many purposes but not for use in the reference section of a signal source. For that purpose, either temperature-compensated crystal oscillators or oven-controlled crystal oscillators are used. The TCXO typically exhibits crystal aging of better then ±2 ppm/yr and a temperature aging of ±1 ppm/yr. The OCXO is capable of 0.1 ppm/yr for crystal aging and 0.01 ppm/yr for temperature aging. The crystal oscillator usually is operated at a frequency such as 5 MHz, but lower frequencies often are needed. To generate these lower frequencies, a divide-by-N digital counter is provided. This circuit divides the output frequency by some integer, N, to produce a much lower reference frequency. Many signal generators provide REF. OUT and EXT. REF. IN capability (these connections often are located on the rear panel of signal generators, so they may be overlooked). The REF. OUT connector provides the reference signal to other instruments, or it can be used for calibration. The EXT. REF. IN allows the use of an external reference source in place of the internal reference section. This sometimes is done to lock up two signal generators that must work together or

1/2

Local Oscillator and Frequency Synthesizer Circuits

147

DIVIDE-BY-N

REF IN

f

DC CONTROL

INTEGRATOR

REFERENCE SECTION

VOLTAGE CONTROLLED OSCILLATOR

FREQUENCY SYNTHESIZER

EXT. REF. IN

OUTPUT SECTION

REF. OUT

VARIABLE OUTPUT ATTENUATOR

REFERENCE OSCILLATOR

/2

DIVIDE-BY-N

ALC MODULATOR

OUTPUT POWER AMPLIFIER

SAMPLER

ALC RECTIFIER ALC DRIVER

Fig. 8.44 Frequency synthesizer block diagram.

to substitute a much more accurate reference such as a cesium beam clock. FREQUENCY SYNTHESIZER The actual signal is produced in the frequency synthesizer section. It is generated by a voltage controlled oscillator, whose output is compared to the reference signal. A voltage variable capacitance diode (varactor) can be used for the VCO, as can surface acoustical wave oscillators (used at higher frequencies and in the microwave bands). The frequency of the VCO is set by a DC control voltage applied to its tuning input line. This control voltage is generated by integrating the output of a phase detector or phase comparator that receives the reference frequency and a divide-by-N version of the

VCO frequency as inputs. When the two frequencies are equal, the output of the phase detector is 0, so the VCO tuning voltage is at some quiescent value. But, if the frequency of the VCO drifts, the phase detector output becomes nonzero. The integrator output ramps up in the direction that will cancel the frequency drift of the VCO. The VCO frequency is continuously held in check by corrections from the integrated output of the phase detector. This type of circuit is a phase-locked loop. Suppose, for example, a signal source has a reference of 5 MHz and is divided by 20 to produce a 250 kHz reference. If the frequency synthesizer divide-by-N stage is set for, say, N = 511, then the VCO output frequency will be 0.25 MHz × 511 = 127.75

148

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

MHz. Band switching, operating frequency, and frequency resolution are controlled by manipulating the reference frequency divideby-N and VCO divide-by-N settings. In some cases, the frequency is entered by keypad, and this tells the signal generator how to set these values. Alternatively, “tunable” signal generators may have a digital encoder shaft connected to a front panel control. The noise component of the output signal is composed of thermal noise and phase noise. Of these two, the phase noise tends to dominate the performance of the signal source. Both the reference oscillator and VCO phase noise contribute to the overall output phase noise. There is a 20 log (N) degradation of the phase noise performance of the signal source because of the divide-by-N nature of the PLL. Fortunately, the bandwidth of the PLL tends to limit the contribution of the VCO phase noise to the overall phase noise performance. OUTPUT SECTION The output section performs three basic functions: It boosts power output to a specified maximum level, provides precision control over the actual output level, and keeps the output level constant as the frequency is changed. The power amplifier is a wideband amplifier that produces an output level of some value in excess of the required maximum level (e.g., +13 dBm). A calibrated precision attenuator can be used to set the actual output level to any lower value required (e.g., −136 to +13 dBm). The automatic level control (ALC) modulator essentially is an amplitude modulator controlled by a DC voltage developed by rectifying and filtering a sample of the RF output level. The ALC driver compares the actual output level with a preset value and adjusts the control signal to the ALC modulator in a direction that cancels the error. Sweep generators (sweepers) are used to produce a signal that changes frequency over a specified range. They are used in receivers that must sweep through a band of frequencies (e.g., spectrum analyzers or elec-

tronic warfare receivers). Although the sweeper nominally resembles an FM generator, there are differences. For one thing, the sweep range usually is quite a bit larger than the deviation of an FM signal generator. Also, the sweeper tends to change frequency from one limit to the other, then snaps back to the first limit. There are three basic forms of sweep: linear (ramp) sweep, stepped sweep, and list sweep. All these forms use a voltage-controlled oscillator frequency synthesizer, but the difference is in the waveform used for sweeping the output signal. Figure 8.45A shows linear sawtooth waveform sweep. At time T1, the ramp applied to the VCO is at 0 and begins ramping up. The frequency begins to move upwards from the 0 V value in a linear manner until time T2, at which point it snaps back to the 0 V value. The stepped sweep (Figure 8.45B) uses a series of discrete voltage steps to change the frequency of the VCO. This method does not produce an output on every frequency in the output voltage range from T1 to T2 but only at specific values determined by the steps. The steps are produced in a circuit such as shown in Figure 8.46. A digital-to-analog converter (DAC) produces an output voltage that is proportional to the binary number applied to its inputs. The maximum output voltage is set by an internal or external DC reference voltage and the applied binary number. In this case, the DAC input is driven by a binary counter, which in turn is incremented by a digital clock (square wave generator). The maximum number of steps that can be accommodated using any given binary counter is 2N, where N is the bit length of the counter. For the 16-bit counter shown in Figure 8.46, therefore, a total of 216 = 65,536 different levels (including 0) can be created. The maximum output voltage is less than the reference voltage by the 1 LSB (least significant bit) voltage, or 1/2N. If a 16-bit counter is connected to a DAC with a 10.00 V reference source, then the 1 LSB step is 0.00015 V and the maximum output voltage is 10.00 − 0.00015 V = 9.99985 V.

149

Local Oscillator and Frequency Synthesizer Circuits

VOLTAGE

V

T T1

A

T2 TIME

VOLTAGE

V

B

T0

T T1

T2

Fig. 8.45 Sweep waveforms: (A) sawtooth; (B) stepped from a DAC circuit.

CLOCK

DC REFERENCE

Fig. 8.46

16-BIT BINARY COUNTER

DIGITAL-TO-ANALOG CONVERTER

DAC circuit and binary counter to generate a stepped waveform.

OUTPUT

150

THE TECHNICIAN’S RADIO RECEIVER HANDBOOK

DIVIDE-BY-N

REF IN

f

DC CONTROL

VOLTAGE CONTROLLED OSCILLATOR

INTEGRATOR

SWEEP WAVEFORM GENERATOR

Fig. 8.47 Open loop synthesizer circuit.

BINARY COUNTER

CLOCK

DIVIDE-BY-N

REF IN

f

DC CONTROL

INTEGRATOR

VOLTAGE CONTROLLED OSCILLATOR

Fig. 8.48 Closed loop synthesizer circuit.

List sweepers also have a DAC to control the VCO but drive it with a series of values stored in a digital memory similar to those used in computers. This permits arbitrary waveforms to be created. Figure 8.47 shows one way sweepers are built: open loop. The frequency synthesizer loop is broken and a linear sawtooth, stepped

waveform, or list generated waveform is applied to its DC control voltage input. Another approach is shown in Figure 8.48. This is a closed loop approach. The frequency synthesizer loop is not open; rather, a binary counter or list memory is used to drive the divide-by-N counter in the synthesizer PLL feedback loop.