I had designed a Differential Amplifier AM Modulator previously and found that I should make some correction to the circuit as an improvement. The differential amplifier AM (Amplitude Modulation) is also called "Long-Tailed Pair" modulator. Here I will write about the flaws in the previous differential amplifier based AM modulator for AM transmitter and the correctness I have done.
The circuit is shown below.
The differential amplifier-based AM modulator is much better and linear than a diode-based AM modulator. In the above circuit, the transistor Q3 acts as a voltage-controlled current source that "feeds" the differential pair (Q1 and Q2). This circuit is actually a Gilbert mixer that multiplies the carrier ($V_c$) and the modulating audio ($V_m$) and results in much lower harmonic distortion. Also, by using BAT1 and BAT2 (±3.7V) gives the circuit good "headroom," which helps keep the transistors in their linear region.
Problems & Solutions
While the circuit topology is ok, there are a few errors in the wiring and component values that might cause it to fail.
The Output Tank (C1): I had set $C1$ set to $0.1\mu F$.
The Problem: At any RF frequency (like 1MHz), a $0.1\mu F$ capacitor has almost zero impedance. It will short your entire output signal directly to ground. If the inductor $L1$ is $25.36\mu H$ and we want to it working in the AM band (~1MHz), $C1$ should be changed to around $1000pF$ ($1nF$). The LC then resonant at 1MHz(AM band)
The Fix: However, I changed both the values of L1 and C1. I opted for L1=100uH and C1=250pF which is standard values used in actual radios.
The Math:
Transistor Biasing (Q3): At the carrier input there is the diode (D1) and a resistor (R3) connected at the base of $Q3$, but the base is also connected directly to the $V_c$ input.
The Fix: We need a coupling capacitor (e.g., 100nF) between your $V_c$ source and the base of $Q3$. Without it, the DC offset from the signal generator will fight with the biasing of $D1/R3$. Similarly, I added coupling capacitor C3 of 10uF at the audio signal input for not disturbing the biasing circuit that follows it.
- Diode(D1) Orientation: In the original circuit, the diode D1 was oriented downwards (Cathode to Ground, Anode to Base). I changed the orientation (Anode to Ground, Cathode to Base). The diode acts as a DC voltage reference that sets the "idle" current for the entire circuit by clamping the base of the tail transistor ($Q3$). In the original orientation, it held the base at $+0.7\text{V}$, forcing excessive current that pushed the signal against the supply rail and caused top-side clipping. By reversing the diode to clamp the base at $-0.7\text{V}$, you reduced the total current and shifted the signal’s operating point into the center of your $7.4\text{V}$ window, providing equal headroom for the waveform to swing both up and down symmetrically.
- Emitter Resistor: I changed the emitter resistor (previously labelled R5 now R1) value from 10k to 470ohm. R1 sets the "gain" of your modulator. By reducing it, you shifted the circuit from a "stiff," low-current mode to a high-sensitivity, high-gain mode, making it much more efficient at translating your carrier signal into a strong AM wave.
Load Balancing: $R2$ and $R6$ are $4.7k\Omega$. This is a bit high for high-frequency RF if you have any parasitic capacitance. Dropping them to $1k\Omega$ will give you a "sharper" response at the cost of a little gain.
Transistors: I replaced the 2N5550 BJT transistor with 2N3904 for better RF speed. When choosing a transistor for RF (Radio Frequency), the most important specification is the Transition Frequency ($f_T$). This is the frequency at which the transistor's gain drops to 1. To get good performance, you want an $f_T$ that is much higher than your operating frequency (1 MHz to 4 MHz).
Practical Values: In the circuit I have used exact values but for practical build, I replaced them with real world actually available 220uH inductor and capacitor (100) values.
