Classical vacuum-tube mixers in the amateur radio literature and in vintage amateur radio equipment are commonly grid- or cathode driven with a tube-based LC or crystal oscillator. This is fitting because such circuits commonly require several to tens of volts of local-oscillator (LO) drive at a high impedance. Driving a classical vacuum-tube mixer with a solid-state LC or crystal oscillator that operates at a low power-supply level is difficult.

Figure 1 shows a mixer circuit that solves this problem. Its core is a 6SH7 sharp-cutoff pentode with its control grid resonant at the incoming-signal frequency and its plate circuit resonant at the desired intermediate frequency. LO energy is applied to the gate of a 2N7000 MOSFET connected as a switch between circuit common and the ground returns of the pentode's cathode bias resistor and cathode bypass capacitor.

Figure 1—Circuit of the hybrid pentode-MOSFET regenerative mixer. Only an LO peak voltage high enough to turn on the 2N7000—per its specifications, 0.8 to 3 V, typically 2.1 V—need be applied to its gate; if necessary, dc bias can be applied to the 2N7000 as shown in Figure 2. Although this drawing shows the pentode screen fed by means of a voltage-dropping resistance, powering the screen via a voltage divider or voltage-regulated supply is recommended to establish a fixed cutoff voltage for the tube.

Figure 2—Options for biasing the 2N7000 when the available LO peak voltage is insufficient to turn on the 2N7000. I use Option A in a modified Allied A-2516 receiver and in the modified first mixer of a Heathkit HW-16 transceiver.

Figure 2 shows options for biasing the 2N7000 in situations where the available LO voltage is too low to turn on the 2N7000. (This is the case in my modified Allied A-2516 receiver, in which the local oscillator—now operating at 9 V instead of 18 V as in a stock receiver—is amplified only by a BJT follower before application to associated circuitry.) Adjusting the bias level by ear while receiving signals in a busy band segment is sufficient for non-critical use.

The 270-Ω resistor in the 6SH7 grid acts to offset the 6SH7's negative resistance to disallow oscillation. It may or may not be necessary—or a value other than 270 Ω may be more optimal—in a given application.

The switched-amplifier mixer can be used in RF-to-IF and IF-to-AF applications. That is, it works as a mixer for RF and also as a so-called product detector for rendering RF signals as audio. I've had success in using the Figure 1 circuit to turn an IF amplifier tube into a detector (6EW6 IF amplifier in a modified Heathkit HW-16 transceiver) and also in using a high-transconductance beam power tube (8106) as the detector in a modified Allied A-2516 receiver. When using the circuit as an RF-to-AF mixer, it's important to minimize beat frequency oscillator (BFO) pickup by the switched-amplifier grid, gate or base; otherwise, that not-caused-by-ac-line-induction species of audio hum known all too well to direct-conversion-receiver experimenters may result.

Lower-transconductance tubes are arguably better as switched-cathode mixers than higher-transconductance types for the purpose of minimizing regeneration-based troubles. The 6G6G (octal 6AK6), 12A5 (a power pentode), and 12A6 (a beam power tube) have been used in the circuit with good results.

Whence the Regeneration?

Why the Figure 1 circuit is regenerative may not be immediately apparent. But when we realize that an RF pentode tube has a grid-to-cathode capacitance on the order of 10 pF and that a 2N7000 FET has a drain-to-source capacitance of 10 to 25 pF, we see that we've set up a Colpitts capacitive voltage divider. This isn't speculative; I confirmed on the bench that the circuit can oscillate when the Q of the tuned-circuit inductor is sufficiently high and the parallel capacitance shunting the Colpitts divider is sufficiently low. (A coil of a few microhenries will suffice.)

On confirming that regeneration in the Figure 1 circuit occurs by means of Colpitts-divider-based feedback, I realized that I had encountered this effect (other than in oscillators) before: as the cause of instability in pentode mixers used by Goodman in his band-imaging receiver designs, and as the mechanism underlying output enhancement in an ill-explained "regenerative" frequency-multiplication scheme for transmitters (Figure 3).

Figure 3—Now that we know that regeneration in the Figure 1 circuit comes from the operation of Colpitts-capacitive-divider-based feedback, we can more fully describe the mechanism underlying output enhancement in this frequency-multiplier circuit from the 1947 Radio Handbook. I can't be the only reader of such treatments to have been frustrated by the ascription of feedback to just "[an] undersized cathode bypass capacitor and large cathode resistor"!

Replacing the Pentode with Cascoded JFETs

The 2N7000 MOSFET works equally well when switching the solid-state analog of a vacuum tube cathode. Figure 4 shows how junction field-effect transistors (JFETs) can be used to implement an all-solid-state version of the mixer circuit.

Figure 4—The 2N7000 switch works equally well with the two-cascoded-JFET, synthetic-tetrode mixer described and evaluated by Campbell, Hayward and Larkin in Experimental Methods in RF Design. Here the switching signal is provided by a BJT Pierce crystal oscillator, with the 2N7000's gate-to-source capacitance serving as part of the capacitive voltage divider on which the Pierce depends. For the output transformer core I used a toroid of -43 ferrite material; per EMRFD, -61 ferrite material would provide higher gain. In the implementation of this mixer now in use as the RF-to-IF frequency changer in my modified Allied A-2516 receiver, the output tuned circuit consists of the 1415-kHz BFO transformer from a 3- to 6-MHz "Command" receiver, modified to tune to 1843 kHz, with the 3.3-kilohm loading resistor omitted. (And what's a Regenerodyne? Glad you asked!)