I decided to build this TRF receiver as a stepping stone to designing and building my own MW/FM/SW receiver. This circuit was selected for its easy-to-understand design (no inter-stage feedback, unlike the regen). We just have a nice straightforward flow; Tuner → RF Amp → Detector → AF Amp → Power Amp. I believe this was the state-of-the-art in radio receiver design in the 1920's.
Some of the parts aren't exactly as specified in the linked web page above. For example, the transistors I ordered (BF245C) I thought were the only n-channel JFETs sold by Maplin that are not surface mount (they're in a standard TO-92 package). I've since realised Maplin do actually sell the 2N3819 - I didn't find them first because of the messed-up categories on their web site. I was worried the transistor characteristics would be too different to the 2N3819, but they were fine, and they're half the price.
This was the first time I built anything on a breadboard. Although my first attempt worked, it was messy and prone to shorting:
Believe it or not, there was some organisation there. Of the four breadboard columns, the first on the left was mostly taken up with the RF amplifier, with the demodulator at the bottom (starting with the diode and the large ceramic capacitor). The signal was patched up to the top of the second column. The entire second column was taken up by the AF pre-amp. Originally I had the trimmer pot on the input of the AF pre-amp, but that had migrated to the power amp by the time the photo was taken. The third and fourth columns were taken up by the power amp. The power amp was laid out more neatly, but that was more down to the need to straddle the LM386 IC across two columns, as well as the low component count. I added an LED power indicator which shared the fourth column with the power amp.
I couldn't get the thing working at all Friday evening, no doubt due to one or more short circuits or popped-out patch wires I hadn't noticed. I decided to rebuild, learning from the mistakes of my first attempt. I think this is a lot neater:
This time I tried to keep most of the patch wires flat, and I gave the components more room. I've moved most of the external patches to the bottom part of the breadboard, which keeps it all nice and neat, and out of the way. I'm pretty happy with it now, although I'm sure I'll get better at breadboarding over time.
I'm having real problems getting the power amp to work properly - it's always either too quiet or oscillating - no middle ground. That's why I moved the trimmer pot over there, but it didn't help. I'm stuck using active speakers until I can get that working. The rebuild actually had a mistake in the power amp - the LM386's bypass capacitor was hooked up to pin 8 (gain) but it should have been pin 7 (bypass). Strangely that made it stop oscillating, and the radio station could be heard in the speaker, albeit not very loudly. Unfortunately that's not really a solution. I guess I'll buy some more trimmer pots when next month's budget comes around, and I can then control attentation and gain on the power amp - hopefully that will solve my power amp problems.
I used Maplin's standard MW/LW coil in place of my home-made MW coil. To be honest it didn't make a great deal of difference. I guess we don't get a very good signal in this room without an antenna. It turned out I'd accidentally wired up the LW coil instead of the MW coil. This happened due to my not fully understanding how inductance affects resonant frequency; I mis-read the LW coil's inductance as 4.1μH, when it's actually 4.1mH, and so I expected the coil with more windings to be the MW coil. Well now I know - higher inductance means lower resonant frequency. The same goes for capacitance; f = 1/(2 * pi * sqrt(LC)). As it happens, I got a better signal on 198 kHz LW (BBC R4) than 693 kHz MW (BBC R5). In fact, I got a better MW signal on my home made coil than the standard ferrite-core one. I guess that's down to the much bigger surface area making it a better antenna. The standard coil would no doubt be a whole other story if it was wired up to a good MW antenna. I'm not receiving much with no antenna, although I did pick up some French radio while I was inadvertently on LW (probably either Europe 1 or France Inter); At some point I'll get round to building a tuned loop aerial, which should help a lot.
I'll either be building a wooden stand or buying a plastic enclosure for the MW/LW coil. It's just too awkward without a stand or box - not easy to solder, and the wires keep snapping. My aim is to have a coil "module" with a banana-plug socket for each of the 8 wires, so I can plug it in to whatever receiver I'm playing with. I should really do the same for the variable capacitor, with home-made LW and MW markings (at least whenever I identify a station), and I guess calculated pF markings. I still haven't got round to getting a knob for that variable capacitor - My human-skin-capacitance is getting annoying. I'll also sort out an enclosure for my kitchen-roll coil, which can be tapped where I spliced the wires together, hopefully allowing me to tune higher frequencies (SW bands?).
Update 2009-02-08 20:00: I've built a basic straight-wire antenna using the remainder of the broken telephone extension I used to build my coil, and connected it to the outer coil. This has improved the TRF's reception a lot, to the point where I can probably identify a few stations tonight, and make some sort of markings for my variable capacitor (if I can find a knob). The aerial wire is threaded through 4 window blind brackets above the two windows in the bedroom, which gives it pretty good height as indoor aerials go, and nearly the full length of the room. I'm thinking about putting a longer straight-wire aerial in the loft, at which point a crystal set may be doable.
Now that I actually have an aerial, there's a bigger voltage on the demodulator output - enough to go straight into the power amplifier, bypassing the AF pre-amp stage. This appears to have solved the oscillation problem - I'm thinking the 100k trimmer pot was not enough to attenuate the signal coming from the AF pre-amp into the power amp. Another improvement was to simply turn the 0.5W speaker over so the volume is increased by turning the entire shoe box enclosure into a makeshift speaker cabinet.