• Missing RF Phase-Shift Feedback: For a transistor to act as a reactance modulator for FM transmitter, it must inject a current into the oscillator tank that is $90^\circ$ out of phase with the tank voltage. This requires an RC or CR phase-shifting network connected from the oscillator tank (Q2) back to the base of the modulator transistor (Q1). In your circuit, the base of Q1 has no RF connection to the oscillator tank; it only receives the audio input ($V_{in}$). Without this RF feedback, Q1 cannot simulate a variable capacitor or inductor.
• Incorrect Role of $C_1$: Capacitor $C_1$ is connected directly between the collector and base of Q1. This creates a Miller effect (increasing the input capacitance to audio) rather than acting as the required RF phase-shifter from the tank circuit.
• Massive Impedance Mismatch ($R_1 = 220\text{ k}\Omega$): The input resistor $R_1$ is far too large. Combined with the $10\text{ k}\Omega$ and $27\text{ k}\Omega$ biasing resistors, the audio modulating signal is heavily attenuated before it ever reaches the base of Q1.

• The Waveform Graph (Top Right): You can see the low-frequency audio signal (the large green sine wave) and the high-frequency carrier (the blue band) simply superimposed on top of each other. This is linear mixing/additive leakage, not frequency modulation. The carrier frequency remains completely constant and is not stretching or compressing with the audio wave.
• The FFT Graph (Bottom Right): The FFT plot is completely blank (flat line at zero). This usually happens in SPICE simulators if:
  • The simulation time step ($\Delta t$) is too large to resolve the high-frequency carrier.
  • The FFT window parameters (start time/end time) were set to a region before the oscillator stabilized, or the sampling rate is mismatched.

• Oscillator Section (Q2): This is configured as a Colpitts-style oscillator using $L_1$, $C_5$, and $C_6$. The base bias network ($R_4 = 10\text{ k}\Omega$, $R_5 = 4.7\text{ k}\Omega$) correctly provides about 1.60V to the base, which is standard for a 5V supply line.
• Modulator Section (Q1): Q1 has a radio-frequency choke ($L_2 = 100\,\mu\text{H}$) at its collector to block RF from the power supply rail. However, because its collector is decoupled from the tank via a simple $10\text{ nF}$ capacitor ($C_3$) without a loop back to its own base, it is behaving like a standard audio amplifier rather than a reactance modulator.

1. Remove $C_1$ from between the collector and base of Q1.
2. Add a Phase-Shift Network: Connect a small capacitor (e.g., $10\text{ pF}$ to $47\text{ pF}$) from the oscillator tank node to the base of Q1. Then, ensure there is a resistor (or the internal resistance of the base network) from the base of Q1 to ground. This creates the $90^\circ$ phase shift needed to make Q1 "look" like a capacitor to the tank circuit.
3. Reduce $R_1$: Drop $R_1$ from $220\text{ k}\Omega$ down to around $1\text{ k}\Omega$ to $10\text{ k}\Omega$ so your modulating audio signal actually has enough amplitude to shift the bias of Q1.

Circuit Topology

This is a two‑transistor "indirect FM" generator:

• Q1 (2N3904) — the reactance modulator. Audio Vm is AC‑coupled through R1 (220k)/C2 (100nF) into Q1's base bias node (R3=27k from Vcc, R2=10k to ground, emitter degeneration R6=470Ω bypassed by C4=100nF). C1 (1nF) runs from that base node to Q1's collector — this is the classic 90°‑phase‑shift feedback element that makes Q1's collector look like a variable reactance whose value tracks the audio signal. L2 (1mH) is Q1's collector RF choke to Vcc. The resulting variable reactance is AC‑coupled via C3 (10nF) onto Q2's collector node.
• Q2 (2N3904) — a common‑base Colpitts oscillator. Base is bootstrapped to AC ground by CC1 (10nF) and biased by R4 (10k)/R5 (4.7k). Collector load is RC (2.2k); the tank (L1=1mH with C5=C6=4.7nF forming the capacitive divider) hangs off the collector through CC2 (10nF). The divider's center tap feeds back to the emitter through CB (1µF) with RE (560Ω) setting emitter bias — a textbook Colpitts feedback loop, and the output is tapped right at the tank.

So the intent is clear: Q1's audio‑dependent reactance loads Q2's tank and pulls its resonant frequency up and down = FM.

The actual problem: the tank isn't resonant anywhere near RF

Compute the oscillator's tank frequency from the values on the schematic:

• C5, C6 in series: Ceq = 4.7nF/2 = 2.35 nF
• f = 1/(2π√(L1·Ceq)) = 1/(2π√(1mH × 2.35nF)) ≈ 103.8 kHz

That's not FM broadcast (88–108 MHz) — it's off by roughly a factor of 1000. The values used (L1 = 1 mH, C5/C6 = 4.7 nF) belong in an audio/low‑IF LC filter, not a VHF tank. For a real ~100 MHz Colpitts you'd want something like L1 ≈ 0.1–0.2 µH with C5/C6 ≈ 15–30 pF each — inductance and capacitance both need to drop by ~3–4 orders of magnitude.

This single scaling error explains everything you're seeing:

• Scope: the dashed cyan trace (CH A, audio Vm) is a normal low‑frequency sine. The solid-looking blue band (CH B, output) isn't a clean RF carrier — it's an oscillation running around ~100 kHz that's too fast for the timebase you're viewing it at, so it renders as a filled block instead of resolved cycles. It also sits close enough to the audio rate that the "carrier" and "modulation" aren't well separated — you don't get the ~1000:1 carrier‑to‑audio ratio real FM needs.
• FFT: all the energy is bunched right at the left edge (near 0 Hz) with nothing across the rest of the spectrum. If the FFT's frequency axis is scaled for an expected 88–108 MHz signal, a real oscillation at ~104 kHz is indistinguishable from DC on that scale — which is exactly the flat trace with one spike at the origin that you're getting. There's no carrier peak and no FM sidebands because there's no RF carrier to have sidebands.

Secondary issues (worth fixing once the frequency is right)

• L2/C1 form their own resonance. L2 (1mH) with C1 (1nF) resonate at ≈159 kHz — close to Q2's tank frequency. L2 is meant to be a simple non‑resonant RF choke for Q1's collector; instead you have two similar‑frequency LC networks that will pull/interact rather than giving one clean carrier.
• Reactance‑modulator phase‑shift ratio. The "reactance" behavior of Q1 depends on Xc1 (from C1) being much larger or smaller than the base‑bias resistance it works against (R3‖R2 ≈ 7.3k). At the current (wrong) frequency, Xc1 ≈ 1.5k — same order of magnitude as 7.3k, not a clean 90° condition, so Q1 would present a mixed resistive/reactive load (amplitude modulation and damping) rather than a pure reactance, even before the tank issue is fixed.
• Coupling point. C3 injects Q1's reactance at Q2's collector rather than directly across the L1/C5/C6 tank. It still works because CC2's impedance is low relative to RC, but injecting directly across the tank nodes gives stronger, more predictable frequency deviation once you're actually tuned to RF.

Bottom line: the topology (reactance modulator driving a Colpitts oscillator for FM transmitter design) is correct in concept, but the tank component values are sized for a ~100 kHz circuit, not a ~100 MHz one. The scope and FFT are correctly showing you a low‑frequency oscillation near DC — there is no missing "FM" behavior, there's simply no RF carrier being generated. Rescale L1/C5/C6 (and ideally L2, C1) into the µH/pF range for the target FM frequency and you should see a real carrier appear in the FFT with audio‑rate sidebands moving as Vm changes.

I can identify several major issues.

1. The LC values are far too large for an FM circuit

The most important values I see are approximately:

For an FM transmitter around, say, 100 MHz, the tank components would normally be in the approximate range of:

• L: tens to hundreds of nH

• C: a few pF to perhaps a few tens of pF

—not mH and µF.

2. Your output LC network resonates at only about 3.3 kHz

This is a very important clue.

You have:

• L1 = 1 mH

• C5 = 4.7 µF

• C6 = 4.7 µF

C5 and C6 appear effectively in series across L1, giving approximately:

Therefore,

gives approximately:

That's audio frequency, not FM RF.

So your output tank is not capable of producing a 88–108 MHz FM carrier.

For example, if you wanted approximately 100 MHz with a 100 nH inductor:

which gives approximately:

That's the scale of components you need.

3. There is no actual variable-reactance element

This is probably the biggest conceptual problem.

For frequency modulation, a reactance-modulated FM oscillator needs something whose effective reactance changes with the modulation voltage.

For example:

Audio signal
     │
     ▼
┌──────────────┐
│ Varactor     │
│ diode        │
└──────┬───────┘
       │ changes C
       ▼
 RF LC resonator
       │
       ▼
 FM carrier

Your schematic contains:

• Q1 = 2N3904

• Q2 = 2N3904

• L1

• L2

• capacitors

but I don't see a varactor diode or equivalent transistor reactance network.

Consequently, there is nothing that makes the LC resonant frequency move according to the audio signal.

A fixed LC circuit gives approximately:

For FM you need:

where:

Your circuit doesn't appear to provide that $C(t)$.

4. Q2 is not configured as an RF oscillator

This is another serious issue.

Look at Q2.

Its base is biased through:

• R4 = 10 kΩ

• R5 = 4.7 kΩ

but CC1 = 10 µF is connected from the base node to ground.

At RF frequencies, 10 µF is effectively an extremely low impedance.

So the base is essentially AC-grounded.

More importantly, the signal from Q1 through C3 appears to be connected to the collector side of Q2, rather than providing the appropriate RF feedback to its base.

An oscillator requires positive feedback.

Something conceptually like:

             ┌─────────────┐
             │             │
             ▼             │
       ┌──────────┐        │
       │ transistor│───────┘
       └────┬─────┘   feedback
            │
            ▼
          LC tank

Your Q2 stage doesn't provide an obvious RF feedback path capable of sustaining oscillation.

Therefore, Q2 is not going to spontaneously generate a stable FM carrier just because L1/C5/C6 are connected at its output.

5. C5/C6 = 4.7 µF completely destroys the intended RF tank

This deserves special attention.

For an RF oscillator, capacitors such as:

are enormously larger than what you'd normally use in a 100-MHz resonator.

At 100 MHz, the reactance of 4.7 µF would theoretically be:

So at RF, it behaves almost like a short circuit.

Your C5/C6 therefore do not behave as useful RF tuning capacitors.

6. The coupling capacitors are also wrong for RF

You have:

• C3 = 10 µF

• CC1 = 10 µF

• CC2 = 10 µF

These are commonly useful for audio-frequency coupling, but not as the primary coupling components in a 100-MHz circuit.

For an FM RF circuit, you would typically see values such as:

• 1 pF

• 2.2 pF

• 5 pF

• 10 pF

• 22 pF

• 47 pF

• 100 pF

depending on the circuit.

7. Your waveform is therefore not the expected FM waveform

The upper graph appears to show a dense/high-frequency-looking signal, but it does not show the characteristic FM behavior clearly.

Remember that FM doesn't necessarily look dramatically different in the time domain.

For:

the amplitude should remain approximately constant.

What changes is the instantaneous frequency.

So for a low-frequency modulating sine wave, you should see:

low frequency       high frequency       low frequency

  /\    /\             /\/\/\/\/\/\         /\    /\
 /  \  /  \           /\/\/\/\/\/\/\       /  \  /  \

The carrier cycles become closer together and farther apart while their amplitude remains approximately constant.

Your graph doesn't demonstrate that clearly.

8. The FFT should show an FM carrier and sidebands

This is another way to diagnose the problem.

Suppose you use:

and:

For FM, the spectrum should contain:

and sidebands at:

etc.

So conceptually:

Amplitude

             │
             │       │
             │   │   │   │
             │   │   │   │
─────────────┴───┴───┴───┴────────── frequency
             fc- fm fc fc+fm

With larger modulation index, more sidebands become significant.

Your FFT appears essentially flat/empty, which strongly suggests that there isn't a meaningful RF oscillation at the output—or the FFT/probe/time-window settings aren't appropriate.

9. There is another major issue: your simulation timestep/sample rate

If you're trying to simulate a 100-MHz FM signal in Proteus, you need to be careful with the simulation settings.

For example:

means:

You need several simulation points within each 10-ns carrier cycle.

If your transient simulation timestep is, for example:

you cannot see a 100-MHz carrier correctly.

The simulator would effectively skip over the RF waveform.

For an RF simulation, use an appropriately small timestep, potentially in the sub-nanosecond to few-nanosecond range, depending on the frequency and what you're trying to resolve.

10. There may also be a unit-entry problem in Proteus

This is actually something I would check first.

If your intention was something like:

• L1 = 100 nH

• C5 = 4.7 pF

• C6 = 4.7 pF

but you entered:

• L1 = 1 mH

• C5 = 4.7 µF

• C6 = 4.7 µF

then the resulting circuit is off by many orders of magnitude.

Likewise, perhaps you intended:

C5 = 4.7 pF
C6 = 4.7 pF
CC2 = 10 pF
CC1 = 10 pF
C3 = 10 pF

rather than µF.

The circuit should look more like this conceptually

For a proper reactance-modulated FM oscillator:

                    +VCC
                      │
                      │
                  RF oscillator
                      │
                 ┌────┴────┐
                 │         │
                 │    LC tank
                 │         │
                 └────┬────┘
                      │
                    RF OUT
                      │
                     ...
                      
              Variable capacitance
                      │
                ┌─────┴─────┐
Audio ─────────►│  Varactor │
                │   diode   │
                └───────────┘
                      ▲
                      │
                 reverse bias

The audio voltage changes the varactor capacitance:

which changes:

and therefore produces FM.

What I would change in your particular circuit

If your objective is something like 100 MHz FM, I would not try to fix this by merely changing one component.

I'd redesign the RF section.

Replace the output tank

Instead of:

L1 = 1 mH
C5 = 4.7 µF
C6 = 4.7 µF

use something approximately in the range of:

L1 ≈ 100 nH
C ≈ 20–30 pF

for a 100-MHz region, then provide a varactor diode for modulation/tuning.

Replace RF coupling capacitors

Instead of:

10 µF

use values in the pF range where appropriate.

Add the actual reactance modulator

For example:

                 +Vbias
                   │
                   R
                   │
Audio ── C ────────┤
                   │
                 Varactor
                   │
                   ├──── LC oscillator
                   │
                  GND

The exact topology matters, so I wouldn't simply insert a varactor randomly.

Most important diagnosis

I'd rank the problems like this:

So the absence of FM is expected from the circuit as currently drawn. The fundamental problem is that it is neither a correctly tuned RF oscillator nor a complete reactance-modulated FM oscillator.

If you give me the intended carrier frequency (e.g. 10.7 MHz, 88–108 MHz, or 100 MHz) and the input modulation frequency/amplitude, I can redesign this exact two-transistor circuit with specific L, C, resistor and varactor values that you can enter directly into Proteus.

5. DeepSeek AI Circuit Analysis