This is a recovered file. The images in this post may be out of order, and there may be duplicates.
Balanced autodyne mixer - The RadioBoard Forums
Post by qrp-gaijin » Sat Oct 07, 2017 7:21
This is my attempt at making a solid state version of vladn's balanced autodyne mixer described in viewtopic.php?f=3&p=76279#p76279 .
Original tube circuit
First, let's look at vladn's original circuit:

The image is a little fuzzy, but the idea is:
Connect two tubes essentially in parallel.
Make a Vackar oscillator, connecting the hot end of the tank inductor to the anodes, and the cold end of the tank inductor to the grids.
Inject the RF signal from the tank into both cathodes in a balanced fashion. This is achieved by a bifilar winding, with the winding for the first tube being phased oppositely to the winding for the second tube.
Extract the IF signal from both anodes in a balanced fashion. This is again achieved by a bifilar winding, with the winding for the first tube being phased oppositely to the winding for the second tube.
The bifilar windings, with opposite phases for each tube, serve to cancel out the LO signal from appearing at the IF port or the RF port. Any LO voltage in one tube's anode (or cathode) winding is exactly canceled out by the oppositely-phased signal from the other tube's corresponding winding.
I note that vladn's post was from November 2011, and used a Vackar oscillator to attempt to even out the LO amplitude while tuning across the band. However, one month later, vladn posted his hybrid-feedback circuits, which offer better equalisation of the LO amplitude (viewtopic.php?f=3&t=3714).
Translating it to solid state
Based on the above understanding, I attempted to convert the circuit to solid state, while allowing operation off of low voltage (1.2 volts). The below circuit is the result. I used hybrid feedback (Armstrong and Vackar) instead of only Vackar feedback. I simplified the biasing for the transistors, using the low-voltage "short circuit biasing" technique where base and collector have the same bias voltage. The rest of the circuit is similar in spirit to vladn's circuit: RF injected into the emitters in balanced fashion and IF taken from the collectors in balanced fashion. I eliminated vladn's C4/C5 capacitive voltage divider, intended to reduce the tank loading by the tube on the hot end of the tank. In my circuit, the collector is connected directly to the hot end of the tank. Reduced loading could be achieved by connecting the collector to a properly-phased and separate link winding on top of L7, to only partially couple the collector into the LO tank. Probably a tap on L7 would also work.
Below, when tuning capacitor C3 is tuned to 200 pF, the LO amplitude (green) is about 120 mV.
Below, when tuning capacitor C3 is tuned to 20 pF, the LO amplitude (green) is again about 120 mV. The LO signal remains a clean sine wave (with no squegging) over the entire tuning range, and has quite well-equalised amplitude. Amplitude equalisation over the tuning range was done by adjusting L8 and C2. (EDIT: Note the simulation is

somewhat incomplete because it only accounts for serial losses in L7; a more complete simulation would require including both serial and parallel losses in L7, which will affect the required values of L8 and C2 to equalise the regeneration level over the tuning range.) Thanks to the amplitude equalisation, the peak emitter current also stays equalised as the LO is tuned, at about 140 microamps with the current circuit parameters. The equalised LO amplitude and emitter current make for more consistent behaviour of the autodyne mixer across its tuning range.
If I have some time I may try simulating the conversion gain of the mixer.
RF-to-LO isolation
The LO signal at the IF port (blue line) is very small, being canceled out by the balanced, oppositely-phased windings in the collector.
IF to RF reverse isolation?
This balanced arrangement achieves some isolation of the LO from the RF and IF ports. However, another interesting question is how good the isolation is between the IF port and the RF port -- specifically, the reverse isolation between the IF tank and the RF tank.
I am considering allowing the RF tank to be freely tuned via C1 (non-ganged), and regenerating the the IF tank L6/C6. Previous experiments revealed that poor RF-IF isolation when using an autodyne mixer with a a regenerated IF tank can lead to undesired effects like the RF tuning affecting the regeneration level and/or the tuning of the regenerative stage. However, isolation -- or at least, reverse isolation -- between the IF tank and the RF tank should prevent the IF tank from "seeing" any impedance changes caused by retuning of the RF tank.
A simplistic argument could hint at good IF-RF reverse isolation. The transistors should be operating in common-base mode as far as the RF signal is concerned -- the RF signal is injected into the emitters, and the bases of the transistors have a low impedance to ground at the signal's RF frequency (through C2). Therefore the IF output at the collector could be isolated from the impedance changes at the RF input thanks to the common-base transistors. This idea would need to be tested either in a circuit simulator or in a hardware prototype.
Making a receiver
Instead of a regenerated IF tank, a crystal or mechanical filter could be used as vladn suggests here: . Follow the IF filter with an MK484/TA7642 and you'd have an ultra-simple superhet.



















