Another VHF Super-Regenerative Receiver

Senin, 07 Juni 2010
Similar to my other VHF Super-Regenerative Receiver this design borrows heavily from Charles Kitchen N1TEV's various published designs. The audio stage is my own design, and I spent a lot of time tuning the component values for the FM broadcast band.

The three inductors are mounted orthogonally with respect to each other to minimise cross coupling. It isn't perfect, they are all in each other's near field and are not co-incident (obviously!) so there is some interaction, but it is greatly reduced compared to mounting them parallel, right alongside of each other. It is important to minimise local oscillator leakage back out the antenna, but as the entire board isn't screened the effort may not actually be worthwhile.
The low impedance current source supply for the regenerative stage improves performance over the usual 5 k pot regeneration control. It also reduces the current consumption of the entire circuit by about 2 mA, the regenerative stage pulling less than 3 mA max. The RF isolation stage by comparison pulls 3-4 mA. The circuit oscillates nicely down to 5 or 6 V which is much better than the previous receiver which had problems achieving superregeneration with a 9 V supply.

With a little bit of careful adjustment of the band-spread trimmer and the main oscillator coil I was able to achieve 87-109 MHz coverage. It spreads nicely across the 180 degrees sweep of the tuning capacitor which is designed for linear tuning in commercial AM radios. The main tuning coil is seven turns of 1mm bare copper wire from a length of mains cable. The inside diameter is about 10mm, I wound it on my hobby knife's Aluminum rod handle.
The other two coils are wound on soda straw using 0.4mm wire. The thick transparent soda straws from Subway, which have an outside diameter of about 8mm, make excellent coil formers for light winding wire. The RFC for the buffer amplifier drain circuit is 30 turns, the coil in the source of the detector is 25 turns. Neither are especially critical, but the detector one may need some experimentation to achieve the best super-regeneration quench. It is important however, that their self-resonant frequency is well above the frequency of operation.
The audio output stage is fairly crude. The design could, in theory at least, suffer thermal run away. Adding a few tens of Ohm resistance in the emitter circuits would prevent this. I did not find it necessary however and the amp seems to work just fine with minimal cross over distortion for such a simple design. The diodes gently bias a few mA of standing current, ie class AB operation, but there is no bootstrapping so positive going amplitude distortion might be a problem at high output powers. All that said, driving 32 Ohm headphones the sound quality seemed quite good. Good enough compared to the quality of the audio recovered by slope detection of FM by such a simple detector.
The volume control pot could be placed as the collector load of the audio preamp, saving a capacitor, but I wanted to have one side of it earthed so I could solder it down to the board like the regeneration pot and the tuning capacitor. The same could be achieved by using a PNP device in the preamp instead, but I didn't think of that at the time. :-)
The biggest problem with using this receiver to listen to FM broadcast radio is beating of the quench frequency with the higher frequency stereo signal components. Quite frankly is sounds horrible and makes it hard to listen to stereo music broadcasts for extended periods. I can't see an easy way to fix this. Increasing the quench frequency so it is above the L-R sidebands and any SCA subcarriers reduces the selectivity and gain of the detector significantly, and there are still mixing problems: The superregeneration process is very non-linear and the IMD is pretty bad, aliasing the sidebands into the audible range.
Perhaps locking the quench at 38 kHz with a separate crystal oscillator would help? Any slight frequency difference beating should be below audio frequencies. The quench filtering could be better. This is a common problem with all superregenerative designs I've found on the net. The single pole RC filter works 'well enough', especially when the output is an LM386 and for human consumption. However the output stage I use works to many hundreds of kHz so the inductance and mechanical response of the mylar speakers is relied upon to deal with the remaining leakage quench signal. Injecting the output of this circuit into a sound card or a tape recorder could cause problems with beating against the sample rate or the bias oscillator.