Nominally, the JMD is complete for 80 meters. But I know better. I just do.
The BFO output seems low, with no readily apparent fault. And there’s that mixer tube substitution. But to test my hypothesis, I fired up the signal generator and tried to get a signal from the antenna terminal to the speaker.
I did get a signal. But it was completely underwhelming. I could just barely hear a -30 dBm signal in the speaker. Having built other receivers before, I found this not terribly surprising. You have to remember, this is a junk box radio, built from junk box parts, to a 50 year old design, that may not really have been well-tested.
I did find one reference to the radio working, but remember that I also confirmed a typo on the tube heaters, and ARRL Handbooks of the past are not known for being error-free. So the adventure continues.
This is the time where ya really have to think about gain distribution in the receiver, and start worrying closely about signal levels.
I retested the audio amp with my 6 dollar eBay wonder audio signal generator. The scope said the lowest level the generator could generate was about 200 millivolts. 200 millivolts did produce ear shattering volume in the headphones, and drove a monitor speaker to more than comfortable room volume. It seems unlikely that the audio amp is the issue.
The IF amp was retested, and again showed an overall gain of about 25 dB, 15 dB net of the filter. This number seems OK, a second stage with another 25 dB may be a good way to go, but I would not expect this to be a big fail in the overall scheme of things.
But here’s the thing. Based on the audio amp voltage gain, I think I have about 50 dB of gain there. The IF amp is at a net 15 dB, so that’s about 65 dB. A signal injected post IF filter has about 70 dB gain following it. This is pretty good. And while I don’t know how much gain this mixer has, 20 to 30 dB of gain seems to be common for vacuum tube mixers. That’s 100 dB of gain, this is getting close to what a radio should do.
But I don’t get much volume at all, with a -40 dbm signal audible but weak, and that’s injecting the signal post IF filter.
What haven’t I tested? Well, the product detector is about the only link there, and I haven’t tested the mixer other than in the assembled radio.
I don’t think weak BFO drive to the product detector should result in low output – distorted output on stronger signals for sure – but not low output. I looked back at the old Miser’s Dream original that used a 12AX7 crystal oscillator, and thought about converting the BFO to a crystal oscillator to see what kind of output I would get. Rummaging around in the junkbox produced an old Kenwood (maybe a TS511?) oscillator board – another flea market find – that had three crystals on it that were for a 3395 kHz filter. No idea why the Kenwood filter used the same center frequency as Heathkit but an interesting, serendipitous find.
So I tried to fire up the oscillator board, but it was clearly dead. Makes sense for a couple dollar flea market treasure, but all three crystals tested good. At this point it’s convert the oscillator to a crystal controlled BFO and try to eliminate a problem, or…..WAIT! Let’s leave well enough alone and build a 2N3904 oscillator with one of the crystals and see what happens.
This is what happened.
The circuit board was from an old Progressive Receiver project, meant to be a single band oscillator for a converter. A bit of redesigning yielded a stand alone oscillator that should work as a BFO. After an initially disappointing test where the oscillator produced only -10 dB of signal, I discovered that the emitter bias resistor I grabbed was 2700 ohms, rather than 270 ohms. Changing the resistor gave me over 2 volts of BFO signal, and each crystal worked, and could be pulled almost 3 kHz. So this oscillator and the crystal could give me all the BFO I needed with separate USB, LSB and CW freqs.
I’m not keen on just throwing in different solutions on a vintage project, but this at least would let me test the product detector with plenty of BFO signal. So I wired it up, and, Tada! not a *(&%^ bit of difference in the product detector performance.
So what’s next? Well, I had used 1N34 diodes in the detector, because that’s what I had, even though 1N67s were spec’d in the Handbook. A quick look on eBay when I started the project suggested 1N67s weren’t readily available, hence the swap. They are both germanium, and I saw one article using this product detector with 1N34s and claiming it worked ok.
I started looking for additional diode cross reference articles without much luck, but then tripped across a thread discussing that many “1N34” diodes are actually Schottky silicon diodes and not germanium like in the old days. It could be that this is the problem. Several searches of an increasing random nature eventually turned up a supply of vintage 1N67 diodes at a CB repair shop. I also found guaranteed vintage germanium 1N60 diodes available on eBay.
So at this point, I’ve spent an additional 20 bucks chasing down vintage germanium diodes to try in the product detector before I go further. Don’t fix it if it ain’t broke. Those diodes should be here by Tuesday April 21, or Project Day 37. I won’t change anything til then.
If the diodes don’t solve the sensitivity issue, then I’ll try a vacuum tube product detector circuit I’ve had good luck with in the past. If that doesn’t work, it’s back to head scratching, but at least I see a couple paths out of the rabbit hole. Here’s hoping one works.
No point trouble shooting an entire radio when you know there is an unsolved problem.