Regenerative FM Broadcast Receiver

This radio comes from Everyday Practical Electronics magazine, January 2006, in the Ingenuity Unlimited (Readers Circuits) section. Francis Hall, from Meinerzhagen, Germany calls it his "Euro Set". I made only minor changes to the circuit, the diagram below is as I built it, see the article for as Francis specified it.

FM Regenerative Receiver Circuit Diagram

I must admit, at first I was sceptical looking at the circuit diagram, a single 1.5 V cell supply? But after a quick check of the audio stage bias with my calculator I saw that Mr Hall had designed it well, so I dug out the solderless breadboard and made up the audio strip. It "blurt" tested OK, and was surprisingly quiet and well-behaved (I used BC548B devices).

The front-end was more problematic, it was obvious that a real RF transistor would be required. The article specified a BF199, but I had none in stock, so I built a point-to-point hack of the circuit with a piece of IC socket to allow experimentation with whatever transistors I had in the junk box. Initial experiments were frustrating, I began to believe it would never oscillate, BC54x's definately don't cut it, neither did a PN3563, or a whole stack of 2SC's I tried. Eventually I found an ancient BF173, on plugging that in I saw a new peak on the spectrum analyzer - it was oscillating at last!

Now that I had a transistor that would oscillate, I rebuilt the front-end around a large tuning gang with an internal 2:1 reduction drive. (I got this unit from KW Tubes on eBay several years ago.). I substituted a 5K pot for the 10K one specified (all I had), and superglued it to the end of the tuning gang, giving me a device with a knob on each end. This physical arrangement is actually quite nice to use.

Using the FM Regenerative Receiver

I retained the transistor socket, so other devices could be tried in the future. One thing became obvious during experimentation, it *requires* a large L:C ratio. Too much capacitance and it will not oscillate at 1.5V. The biggest challenge seems to be to get it oscillating, then you can adjust the components to the frequency band of interest. My frequency determining components, the coil, tuning gang, and series caps and quite different from the article - yours probably will be too. I used 5 turns of tinned 0.71 mm wire on a soda straw for the coil. I am unsure of the capacitance on the gang, I didn't measure it. The bandspread cap is a 56 pF in series with a 33 pF = near 20.7 pF. Mine tunes about 85 MHz to 122 MHz. I could achieve better bandspread, but with the reduction drive the radio is fairly easy to use as-is. Covering part of the airband is a nice feature too.

RF Stage Photo

The audio amp was rebuilt on a piece of punch-board, this time using NOS Philips BC547s which aren't quite as good as the devices used in the breadboard test, but which worked fine. Some 5-minute epoxy was used to glue the board and headphone socket to the frame of the tuning gang. A slide switch was added as an "on" switch, and a pair of magnetic battery connectors (Jaycar) were used to make connection to the AA cell powering the device.

Audio Amplifier Picture

The completed device pulls about 2.5 mA, giving it a very long battery life. The regenerative stage and the first two audio pre-amps draw 450 uA.

Battery Attachment Picture

The Good

The radio is an excellent performer. It easily picks up all the local channels with more than enough volume. With the regeneration turned up it is narrow enough to distort the received signal purely because its bandwidth is too narrow for the signal's deviation.

Its audio quality is surprisingly good. Unlike the super-regenerative receivers I've built in the past there is no quench to mix the stereo sub-carrier and pilot into the baseband. It can also have an extremely high-Q, unlike the super-regenerative receivers which have a minimum bandwith of four times the quench. It is perhaps slightly less sensitive, but much, much nicer to use; far easier to tune for best audio quality.

To tune, you may use the RF stage gently oscillating to locate the station of interest. Tuning across a station will sound like listening to FM on a SSB receiver as most of the signal's energy will be outside the detector slope. Once you've got a channel near-by, you back-off the regeneration until it stops oscillating and the Q is at an optimal point, then shift the frequency to put the slope at the best position for low distortion. This requires some dexterity, manipulating the tune and regeneration controls at the same time to place the signal in the bandpass and adjust its width for an undistorted audio output. It takes more words to describe than actually do, you will quickly become proficient.

The Bad

Because of the regenerative stage topology, neither side of the tuning cap can be referenced to ground. This means the entire bulk of the receiver is RF hot, which makes it suffer badly from hand-capacitance effects. The regeneration control spindle is coupled to this capacitively (via its case) because I glued it directly to the gang frame - a physical design problem, not a limitation of the circuit as the former is, but annoying none the less. I would like to redesign the regenerative stage to use a common-base oscillator which should make the radio much more stable and pleasent to use.

There is no volume control. With strong signals the RF stage can overdrive the audio stage. At the very least this is painfully loud, but it also pushes the relatively simple amps into their +ve-going distortion and even into full clipping on very strong local signals. (The local ABC Classic FM - 92.9 MHz is a good example, it is a gigantic signal at my QTH and simply can't be tuned in properly at any regeneration level). Turning down the regeneration as a volume control is very non-optimal, it rapidly reduces the Q of the tuned circuit, widening the bandwidth and destroying the selectivity, however this does work for powerful stations in otherwise relatively quiet parts of the band (2WS FM - 101.7 MHz for example).

The Ugly

This radio is quite possibly the ugliest thing I've ever built. It is fairly small, and fits in a Decor 250 ml plastic tub for transport, but its still an ugly hack.

Radio in Box

Usage Experence

The lack of AGC means mobile use can be almost impossible. Unlike lower wave bands where the longer wave lengths mean multipathing isn't a huge issue, at VHF multipath related fading is severe. Over water or in the city the fades can be so extreme and rapid it is very difficult to tune the radio, or listen to the program.

Stationary the device is quite pleasant to use, despite the body-capacitance issue. I've been considering lowering the gain of the 1st audio stage or putting in a volume pot, the radio is significantly more sensitive than is required at times, and often keeping it sufficiently selective in the crowded 50 kW+ upper-end of the FM band means the audio is simply too loud or distorted.

The radio suffers mildly from being unshielded. Some interference from mobile phones in close proximity is experienced, but putting your finger on the 1st audio stage biasing resistors shunts most of it. The problem is quite minor however, no where near as severe as with the recent MW Receiver with gets hammered by any phone within 20 metres and TV transmitters in Sydney Harbour.

Future Ideas

My next attempt at a similar radio will probably be a hybrid including a regenerative stage at a fixed IF in the HF region, with a front-end converter based on a varicap tuned VFO feeding a simple cascode mixer. I'll carefully design an 80-115 MHz filter and an LNA to feed it from signals taken from the headphone lead and build the entire circuit inside a shielded box to avoid mobile phone noise. I may be able to implement an AGC at the LNA stage, or a diode attenuator in the front-end to control the total system gain, however this will degrade the noise figure a bit - on the FM broadcast band this shouldn't be a huge issue as the signals are generally quite strong in the metro area where I'll be mostly using the unit. That said, a non-linear element in the front-end might produce intermod issues from the same very powerful signals in-band, so varying the gain of the LNA might be the better approach.