Hi-Fi

- - - LPF DC Operating Point ---
V(supply): 6 V
V(out): 2.0419 V
V(d): 5.35688 V
Ic(Q5): -0.645428 mA
Id(J1): 1.04209 mA

The audio filter circuit that I designed in did not use the notch but is an S-K (Sallen-Key) 3 pole design. It could be tweaked a bit for the European station spacing, but is probably okay without any design change as it is. The filter compromise reduces noise at the expense of bandwidth. Most cheap radios have inadequate filtering, erring on the side of too much in the case of AM tuners (about 3KHz. roll off) to too little in the case of cheap radios. I erred on the side of (perhaps) too little, in order to include more bandwidth so that the audio upper spectrum would be present if broadcast. I'm relying on the sharper cutoff of a 3-pole Chebyshev filter to help more with noise while allowing "flat" (has some peaking) response to 6.6KHz. Listening revealed that I could get a little more by allowing 10KHz. bandwidth, but the noise was much more noticeable. The chosen bandwidth works with the 9KHz. European. channel spacing as well.
Note that I used a BF244B FET in the circuit, but a 2N4416 for the simulation. The two devices have very close to the same characteristics. Also, the filter works just about as well
with some standard components for C4 and C2. Here are the new values:
C4 old=15nF, new=10nF
C2 old=39nF, new=33nF
The resulting filter response looks very much the same, but the filter gains some bandwidth. These parts are easier to get, but 5% tolerance would be advised for C2, C3, and C4.
I later discovered that the JFET-BJT follower circuit works better when the drain resistor (R14 in the simulation, R18 on the full schematic) is increased to 1K ohm. I made this change when I found that one of my circuits had more distortion than expected. The FETs have a lot of variation, and increasing the resistor value makes the circuit more tolerant over the full range of possible BF244B FET characteristics.

I have to say that this tuner gives me "near FM" quality on AM. The few distractions come from static which AM is famous for, and can't be eliminated, but other than this, it sounds
great.




Here is a Hi-Fi AM superhet receiver using the linear detector discussed above. The whole thing fits on a small PC board pictured right. The linear detector is on the part of the PC board in the foreground, and the TO220 package to the left of it is an LM317 voltage regulator for the 6V power rail.
The front end circuit is shown below. It uses the NXP / Philips NE602 double-balanced mixer chip; or, as here, the SA612, which is a lower cost version.

C5 and C6 were NOT fitted in the final build. The NE602 / SA612 has differential input and output, but I wanted the option of running single ended in case the ground would help. The filter formed by the 21.5 ohm resistor and 10uH choke keeps feedback from occurring through the power supply rail, which will cause the receiver to oscillate if those components are not present. C17 is part of that filter too. The oscillator coil was used "inverted" from what is normal. The tap is closest to the "hot" end of the coil, since the SA612 oscillator doesn't need much "step-up" from the transformer. In fact, the resistor is to reduce the drive somewhat, to keep the oscillator amplitude from being too much.
To avoid stability problems, the "loopstick" antenna should preferably be shielded from the rest of the circuit. Placing the circuit in a metal box, with the loop antenna outside is optimum. My implementation had the loop antenna attached to the PC board, which allows some feedback through the receiver of the 455KHz. signal and hence some "tweet".
The IF amplifier is shown on the complete Circuit Schematic (GIF image, click to open it in a separate browser window) and is discussed in more detail later. Note that the Toko IF transformers have an internal resonating capacitor, which is not shown on the diagrams. The detector and audio output part of the circuit is shown below.

Q7, Q5, Q6, and Q8 comprise the detector circuit. Q9 and Q10 form a unity gain buffer, and 3-pole active low pass filter that cuts off at 7KHz. (12 dB down at 10KHz.). Q9 is a BF244B
N channel JFET. The IF amplifier has AGC driven by the detector output at the top of R16.