A regenerative circuit is an amplifier circuit that employs positive feedback (also known as regeneration or reaction). Some of the output of the amplifying device is applied back to its input to add to the input signal, increasing the amplification. One example is the Schmitt trigger (which is also known as a regenerative comparator), but the most common use of the term is in RF amplifiers, and especially regenerative receivers, to greatly increase the gain of a single amplifier stage.

The regenerative receiver was invented in 1912 and patented in 1914 by American electrical engineer Edwin Armstrong when he was an undergraduate at Columbia University. It was widely used between 1915 and World War II. Advantages of regenerative receivers include increased sensitivity with modest hardware requirements, and increased selectivity because the Q of the tuned circuit will be increased when the amplifying vacuum tube or transistor has its feedback loop around the tuned circuit (via a "tickler" winding or a tapping on the coil) because it introduces some negative resistance.

Due partly to its tendency to radiate interference when oscillating, by the 1930s the regenerative receiver was largely superseded by other TRF receiver designs (for example "reflex" receivers) and especially by another Armstrong invention - superheterodyne receivers and is largely considered obsolete. Regeneration (now called positive feedback) is still widely used in other areas of electronics, such as in oscillators, active filters, and bootstrapped amplifiers.

A receiver circuit that used larger amounts of regeneration in a more complicated way to achieve even higher amplification, the superregenerative receiver, was also invented by Armstrong in 1922. It was never widely used in general commercial receivers, but due to its small parts count it was used in specialized applications. One widespread use during WWII was IFF transceivers, where single tuned circuit completed the entire electronics system. It is still used in a few specialized low data rate applications, such as garage door openers, wireless networking devices, walkie-talkies and toys.

The gain of any amplifying device, such as a vacuum tube, transistor, or op amp, can be increased by feeding some of the energy from its output back into its input in phase with the original input signal. This is called positive feedback or regeneration. Because of the large amplification possible with regeneration, regenerative receivers often use only a single amplifying element (tube or transistor). In a regenerative receiver the output of the tube or transistor is connected back to its own input through a tuned circuit (LC circuit). The tuned circuit allows positive feedback only at its resonant frequency. In regenerative receivers using only one active device, the same tuned circuit is coupled to the antenna and also serves to select the radio frequency to be received, usually by means of variable capacitance. In the regenerative circuit discussed here, the active device also functions as a detector; this circuit is also known as a regenerative detector. A regeneration control is usually provided for adjusting the amount of feedback (the loop gain). It is desirable for the circuit design to provide regeneration control that can gradually increase feedback to the point of oscillation and that provides control of the oscillation from small to larger amplitude and back to no oscillation without jumps of amplitude or hysteresis in control.

Two important attributes of a radio receiver are sensitivity and selectivity. The regenerative detector provides sensitivity and selectivity due to voltage amplification and the characteristics of a resonant circuit consisting of inductance and capacitance. The regenerative voltage amplification ${\displaystyle u_{\mathrm {o} }}$ is ${\displaystyle u_{\mathrm {o} }=u/(1-ua)}$ where ${\displaystyle u}$ is the non-regenerative amplification and ${\displaystyle a}$ is the portion of the output signal fed back to the L2 C2 circuit. As ${\displaystyle 1-ua}$ becomes smaller the amplification increases. The ${\displaystyle Q}$ of the tuned circuit (L2 C2) without regeneration is ${\displaystyle Q=X_{\mathrm {L} }/R}$ where ${\displaystyle X_{\mathrm {L} }}$ is the reactance of the coil and ${\displaystyle R}$ represents the total dissipative loss of the tuned circuit. The positive feedback compensates the energy loss caused by ${\displaystyle R}$, so it may be viewed as introducing a negative resistance ${\displaystyle R_{\mathrm {r} }}$ to the tuned circuit. The ${\displaystyle Q}$ of the tuned circuit with regeneration is ${\displaystyle Q_{\mathrm {reg} }=X_{\mathrm {L} }/(R-|R_{\mathrm {r} }|)}$. The regeneration increases the ${\displaystyle Q}$. Oscillation begins when ${\displaystyle |R_{\mathrm {r} }|=R}$.

Regeneration can increase the detection gain of a detector by a factor of 1,700 or more. This is quite an improvement, especially for the low-gain vacuum tubes of the 1920s and early 1930s. The type 36 screen-grid tube (obsolete since the mid-1930s) had a non-regenerative detection gain (audio frequency plate voltage divided by radio frequency input voltage) of only 9.2 at 7.2 MHz, but in a regenerative detector, had detection gain as high as 7,900 at critical regeneration (non-oscillating) and as high as 15,800 with regeneration just above critical. The "... non-oscillating regenerative amplification is limited by the stability of the circuit elements, tube [or device] characteristics and [stability of] supply voltages which determine the maximum value of regeneration obtainable without self-oscillation". Intrinsically, there is little or no difference in the gain and stability available from vacuum tubes, JFETs, MOSFETs or bipolar junction transistors (BJTs).

A major improvement in stability and a small improvement in available gain for reception of CW radiotelegraphy is provided by the use of a separate oscillator, known as a heterodyne oscillator or beat oscillator. Providing the oscillation separately from the detector allows the regenerative detector to be set for maximum gain and selectivity - which is always in the non-oscillating condition. Interaction between the detector and the beat oscillator can be minimized by operating the beat oscillator at half of the receiver operating frequency, using the second harmonic of the beat oscillator in the detector.