VFO Stability and Drift
A variable frequency oscillator (VFO) gives you something a crystal oscillator cannot: the ability to tune continuously across a frequency range rather than snapping to a fixed frequency. But this flexibility comes at a cost — LC VFOs drift. When you set a home-built signal source or receiver to 7.100 MHz and come back ten minutes later to find it on 7.095 MHz, that is VFO drift at work. Understanding why VFOs drift — and the techniques used to minimize it — is essential for building or repairing any free-running oscillator, and it explains why almost all modern equipment uses crystal-referenced synthesizers instead.
What Is a VFO and Where Is It Used?
A variable frequency oscillator is any oscillator whose output frequency can be adjusted, typically over a range of at least several hundred kilohertz. The most common mechanism is a variable capacitor in the LC tank circuit — rotating the capacitor shaft changes the capacitance, which changes the resonant frequency. This is how radios, signal generators and test oscillators were tuned for most of the twentieth century: one dial controlled a VFO that set the frequency.
In a superheterodyne receiver, the VFO — usually called the local oscillator (LO) — mixes with the incoming signal to produce the intermediate frequency (IF). If the incoming signal is at 7.2 MHz and the IF is 9 MHz, the LO must be at 7.2 + 9 = 16.2 MHz (or 9 − 7.2 = 1.8 MHz for low-side injection). Tuning the receiver from 7.0 to 7.35 MHz means tuning the LO from 16.0 to 16.35 MHz. Any drift in the LO frequency appears directly as drift in the received frequency — the signal you are listening to drifts across the IF passband and eventually disappears out of the edge.
In a signal generator or transmitter, the VFO determines the output frequency directly. A drifting signal generator makes measurements wrong; a drifting transmitter can wander out of its permitted channel and interfere with other users.
A practical VFO design incorporating the key stability elements: regulated supply, temperature-compensating capacitor, high-Q toroidal inductor, and an isolation buffer amplifier to prevent the load from pulling the frequency.
View LargerThe Causes of Frequency Drift
The resonant frequency of an LC tank is f = 1/(2π√LC). Any change in L or C produces a change in f. In a real circuit, several factors cause L and C to change over time:
1. Temperature change (dominant cause): All capacitors and inductors have a temperature coefficient — their value changes with temperature. When the circuit warms up after switch-on, or when the ambient temperature changes, L and C both change, shifting the frequency. Temperature is responsible for 80–95% of VFO drift in most practical circuits.
2. Supply voltage variation: The transistor in the oscillator has internal capacitances (collector-base junction capacitance, Ccb, and emitter-base junction capacitance, Ceb) that are part of the effective tank capacitance. These junction capacitances depend on the voltage across the junction. If the supply voltage changes, the junction voltages change, changing Ccb and Ceb, which shifts the frequency.
3. Mechanical vibration: Physical vibration causes the tank coil to flex, changing its inductance. Variable capacitors can rattle if not properly constructed. This is microphonics — addressed in detail in the next lesson.
4. Load variation: When the load impedance connected to the oscillator output changes — because a following stage is switched on and off, or different equipment is connected — the change reflects back through the buffer amplifier into the tank, slightly changing the effective Q and resonant frequency. This is called frequency pulling.
5. Component aging: Over many months and years, capacitor dielectrics settle slightly, inductor cores relax magnetically, and resistors drift in value. These are small effects compared to temperature, but they contribute to long-term drift.
Temperature Effects on Capacitors
Not all capacitors are equal in temperature stability. The dielectric material determines how much the capacitance changes with temperature, characterised by the temperature coefficient of capacitance (TCC). TCC is measured in ppm/°C (parts per million per degree Celsius).
| Capacitor Type | Dielectric Code | TCC (ppm/°C) | VFO Suitability |
|---|---|---|---|
| Silver mica | — | 0 to +70 | Excellent — low TCC, very stable |
| NP0 / C0G ceramic | NP0 or C0G | 0 ± 30 | Excellent — essentially zero drift |
| N750 ceramic | N750 | −750 ± 120 | Good for compensation (intentional negative drift) |
| N1500 ceramic | N1500 | −1500 ± 250 | Used for drift compensation |
| X7R ceramic | X7R | ±1500 or worse, nonlinear | Not acceptable in VFO tanks |
| Y5V ceramic | Y5V | ±30,000 (can vary 80% over temperature) | Completely unsuitable — disastrous drift |
| Polystyrene film | — | −150 ± 30 | Very good — predictable small negative TCC |
| Polyester (Mylar) | — | +400 to +1000 | Poor — use only if no better option available |
| Air variable | — | ±20 (mainly mechanical expansion) | Excellent — best choice for main tuning cap |
The practical message is simple: use only NP0/C0G ceramics or silver-mica capacitors in VFO tanks. Any other ceramic type — X7R, Y5V, Z5U — will cause the VFO to drift unacceptably with temperature (and with voltage). The main tuning capacitor should be a high-quality air-variable capacitor (such as one salvaged from an old broadcast receiver), which has a very small temperature coefficient because its capacitance is set by air-spaced metal plates.
Temperature Effects on Inductors
Inductors also drift with temperature, but the mechanisms are more complex:
Copper wire resistance: The resistance of copper increases about 0.4%/°C with temperature. This changes the Q of the inductor, which changes the effective reactance slightly — not a major frequency effect, but it does affect loop gain and can cause the oscillator to vary slightly in amplitude.
Dimensional changes: The wire, former, and core all expand slightly when heated. For an air-wound coil on a plastic former, the former typically expands more than the wire, spreading the turns apart and reducing the inductance (inductance decreases as turns move further apart). For a toroidal core, the core's permeability changes with temperature.
Core permeability changes: Ferrite and iron-powder core materials change their permeability (and therefore the inductance wound on them) with temperature. Iron-powder cores (such as the T50-6 red and T50-2 red cores used in toroidal inductors) have a positive temperature coefficient of permeability — inductance increases with temperature. Ferrite cores (fair-rite 77 and similar) have varying behaviour. Air-wound coils have zero core temperature coefficient but are bulkier.
The temperature coefficient of inductance for a toroidal inductor wound on a T50-6 core (iron powder, mix 6) is approximately +35 ppm/°C. For a T50-2 (mix 2, red): approximately +95 ppm/°C. This positive temperature coefficient of inductance can sometimes be exploited to partially cancel the negative temperature coefficient of a well-chosen capacitor.
Supply Voltage and Load Effects
The transistor's collector-base capacitance (Ccb) varies with the reverse bias voltage across the junction. A typical small-signal NPN transistor like a 2N3904 might have Ccb of 4 pF at 5 V collector-base voltage, decreasing to 3 pF at 10 V. In a 7 MHz oscillator with a 400 pF tank capacitance, this 1 pF change represents a 0.25% change in capacitance, which changes the frequency by 0.125% — a shift of about 8.75 kHz at 7 MHz. This is not negligible.
The cure is a well-regulated supply voltage. A three-terminal IC voltage regulator (such as the 78L08 for 8 V) keeps the supply to within ±0.1 V across a wide range of input voltage and load current. Combined with good bypass capacitors close to the oscillator circuit, this eliminates supply-induced drift almost completely.
A 7 MHz Colpitts VFO has a total tank capacitance of 350 pF. The 2N3904 transistor contributes Ccb = 4 pF at nominal 8 V supply. The supply droops from 8 V to 7 V under load. At 7 V, Ccb rises to approximately 4.3 pF.
Change in effective tank capacitance: ΔC = 0.3 pF
Fractional change: ΔC/C = 0.3/350 = 0.00086 = 0.086%
Fractional frequency change: Δf/f = −(1/2) × ΔC/C = −0.043%
Frequency shift at 7 MHz: 7,000,000 × 0.00043 = 3,010 Hz ≈ 3 kHz
A 3 kHz frequency shift from a 1 V supply droop is significant: it is the whole width of a voice channel, and enormous for a narrow-band data signal or a precision measurement. Using a regulator limits the supply droop to typically ±0.05 V, reducing this effect by 20:1 — down to about 150 Hz.
Load pulling: Any reactive load connected to the oscillator output partially appears at the tank, changing the effective resonant frequency. A buffer amplifier with high input impedance (>>1 kΩ) and good reverse isolation (30 dB or better) prevents this. The buffer must be powered from the same regulated supply as the oscillator, and its input should be only loosely coupled to the tank (via a small coupling capacitor of 1–5 pF).
Construction Techniques for Stable VFOs
Good VFO construction is as much about mechanical engineering as electronics. Here are the key techniques used by experienced home builders:
Mechanical rigidity: Mount all tank components on a rigid metal chassis or board — not on a flexible PCB that will flex when the box is moved. The variable capacitor must be solidly mounted with no wobble in the shaft bearing. Any mechanical play allows the capacitance to change with vibration and touch.
Thermal mass: Large, heavy components with high heat capacity warm and cool slowly. Using a heavy die-cast aluminum enclosure slows thermal transients, reducing drift rate during warm-up. Some builders deliberately add thermal mass to the inductor assembly.
Thermal isolation of the oscillator: Mount the oscillator section inside a separate inner shield can within the main enclosure. This inner can isolates the oscillator from heat sources (power transistors, transformers) and from convection currents caused by ventilation holes. The separate can also provides RF shielding.
Spacing components for airflow: Do not pack components tightly together. Allow air to circulate so temperature gradients within the oscillator are small. Components at different temperatures cause differential drift.
Temperature compensation: If the oscillator drifts consistently in one direction with temperature — typical behaviour — add a small-value temperature-coefficient capacitor in parallel with a tank capacitor to cancel the drift. For example, if the VFO drifts low with increasing temperature (indicating the effective capacitance is increasing with temperature), add a small N750 capacitor (negative temperature coefficient of −750 ppm/°C) to produce an equal and opposite frequency shift.
Typical VFO warm-up drift curve. The initial rapid drift during the first 5 minutes occurs as the transistor, inductor, and capacitors warm to operating temperature. Drift slows and stabilises after 15–20 minutes. Temperature compensation reduces the total drift but does not eliminate the warm-up transient.
View LargerWarm-Up Drift and Long-Term Stability
VFO drift behaviour typically follows a characteristic pattern over time:
Initial rapid drift (0–5 minutes): The transistor's junction capacitances and the tank capacitors heat up quickly. This produces rapid frequency change — often 5–30 kHz in the first five minutes for an uncompensated circuit. The transistor warms fastest because it dissipates power internally.
Intermediate settling (5–20 minutes): The inductor, which has higher thermal mass, slowly reaches thermal equilibrium. The frequency drift slows but continues. The rate might be 1–5 kHz per minute.
Stable plateau (after 20 minutes): All components reach a stable temperature determined by the balance of internal power dissipation and heat flow to the environment. The frequency is now stable to within the noise and microphonic contribution. A well-designed VFO should be within 500 Hz of its final frequency after 30 minutes of warm-up.
The practical lesson is to allow VFO-based equipment to warm up for at least 15–30 minutes before critical frequency work. This was standard practice in the era of discrete-component radios and test equipment, and remains relevant for home-built oscillators and older instruments today.
VFOs Today
Virtually all commercial radios, signal generators and wireless devices made since the 1980s use PLL synthesizers or DDS instead of free-running LC VFOs. These synthesized frequency sources inherit their stability from a precision crystal reference — they are essentially as stable as the reference crystal from the moment of switch-on, with no warm-up drift to speak of.
However, LC VFOs are still built and used. Hobbyists build simple receivers, signal sources and educational projects around them for their simplicity and the satisfaction of direct frequency control, and the voltage-controlled oscillator at the heart of every PLL is itself an LC (or ring) oscillator whose drift the loop must correct. Restorers of vintage radios and test equipment meet LC VFOs every day.
Understanding VFO drift is also valuable for diagnosing problems in older equipment that may have drifting oscillators, aging component values or failed regulators. An instrument that drifts several kilohertz per hour may have a failing supply voltage regulator, a capacitor that has drifted from its original value, or a bad solder joint whose resistance changes with temperature.
⚖ Experiment: Measuring VFO Drift with a Frequency Counter
Measure the warm-up drift of any LC oscillator — a home-built VFO, or the oscillator in an older receiver or signal generator — using a frequency counter (ideally one with a TCXO or GPS-referenced timebase).
- Any LC oscillator circuit (a home-built VFO, or an older receiver or signal generator with an accessible oscillator)
- Frequency counter (even an inexpensive digital counter with a TCXO reference)
- Pencil and paper or spreadsheet to record readings
- Timer (phone stopwatch)
- Allow the frequency counter to warm up for 10 minutes before starting (it has its own oscillator that must stabilise).
- Start with the VFO cold (off for at least 30 minutes).
- Power on the VFO and simultaneously start the timer.
- Record the frequency reading every 60 seconds for 30 minutes.
- Plot frequency vs time on a graph.
- Note the direction of drift (up or down), the rate of drift in the first 5 minutes, and the time at which drift reduces to less than 100 Hz per minute.
A typical LC oscillator will show rapid drift in the first 3–5 minutes (direction depends on which capacitor temperature coefficient dominates), then slower drift out to 15–25 minutes, then a near-stable plateau. The total drift from cold start to stable might be 2–20 kHz at a few megahertz. After identifying the drift direction (up or down with temperature), you can add a compensating capacitor — try a small N750 or N1500 capacitor in parallel with one of the tank capacitors if drift is downward, or a positive-TC capacitor if drift is upward. Repeat the experiment to see if the drift is reduced. This is the standard technique used in VFO construction.
Frequently Asked Questions
Why does my VFO drift in opposite directions at different times?
This happens when the total drift has contributions with opposing temperature coefficients — typically the transistor's junction capacitance (which decreases with temperature, driving frequency up) and the tank capacitor's positive TCC (which increases with temperature, driving frequency down). As the circuit heats up, the transistor stabilises first (it reaches temperature faster), then the inductor and capacitors settle later. The net drift changes direction as each component reaches equilibrium. The solution is to identify the dominant drift component and use a targeted compensating capacitor rather than guessing randomly.
Will operating my VFO in a warm room cause it to drift more?
Only if the ambient temperature changes significantly during operation. A stable room temperature (±2°C) matters less than the temperature rise of the components themselves due to internal power dissipation. A VFO running at 8 V with a 20 mA transistor dissipates 160 mW internally — enough to heat the transistor junction by 15–25°C above ambient regardless of the room temperature. The main thing is to allow the circuit to reach its own steady thermal state before making critical measurements.
Is there any way to make an LC VFO as stable as a synthesizer?
Not without adding a phase-locked loop — in which case you essentially have a synthesizer. A free-running LC VFO cannot match the stability of a crystal-referenced synthesizer. However, you can get very good results — typically ±100 Hz at a few megahertz — by using a well-regulated supply, NP0/C0G capacitors, a high-Q air-wound or toroidal inductor, careful temperature compensation, and a rigid mechanical enclosure. For listening to broadcasts or for general bench use, ±500 Hz stability is perfectly acceptable. For narrow-band data signals or precision measurement, a synthesized frequency source is far more practical.
Test Your Knowledge
Answer the questions below to check your understanding. Every answer can be found in the lesson above.