Well it’s been a week. I’ll keep the status report brief as the frustration has been pretty intense.
The diode situation
As mentioned last time, I feared that my recent (non-vintage) 1N34’s might be lowering the detector sensitivity. After finding the 1N67A’s at a CB shop, I did the swap, and there was a noticeable but non-critical difference in performance. Things sounded a bit cleaner, and there was marginally more detector output.
But not a game changer. This is good news if you want to build this circuit with 1N34’s.
The IF Amp and the non-tuning coil
You may remember the IF amp was suspect because one tuned circuit would not peak. This turned out to be a cold solder joint on an IF transformer lead, discovered by reflowing the joint. It didn’t increase the gain much, but at least I know that the amp is wired right.
The receiver now barely receives a -90 dBm signal.
The Q-Multiplier and non-obtanium RF coils
So the good news, of sorts, was that as it sits (or sat, see below) there was enough gain to listen to the receiver and copy 80 meter cw and SSB signals. Bandpass sounded good. VFO and BFO were stable enough for satisfactory novice operation, which was the goal.
But this was with just a single tuned front end. I have said from the beginning that the front end may cause problems because the coils deemed critical by the ARRL Lab were not available. I built the RF Q-multiplier circuit, which worked fine – you could see the signals increase as you increased the Q – but despite trying several different configurations of home-wound coils, I could never get the Q-multiplied signal strength more than a few dB above that of my single tuned circuit.
A bit of research suggested that a properly operating RF Q multiplier ought to perk the front end up by a good 10 dB. I concluded that lack of the critical coils was not going to get me there. I’ll just go ahead with a double tuned circuit for image rejection, perhaps with an RF amp tube in between the circuits for the higher bands. But that’s a problem for a later day.
The Front End Trap
While puzzling out the apparent lack of sufficient gain I tested the front end trap that the Lab specified as necessary. If you look back at the schematic, you’ll see that a 3395 khz trap is needed because of the somewhat marginal front end. The Lab did note a reduction in 80 meter sensitivity from the trap being so near 80 meters, but said that the receiver was, in essence, good enough.
I ran some tests to see what effect the 3395 trap had on 80 meters, and was unpleasantly surprised. I measured attenuation of almost 20 dB at 3500 kHz, though the loss dropped to about 6 dB at 4000 khz.
Running without the trap is impossible, as the atmospheric noise at the IF frequency dominated the 80 meter signals. But giving up 20 dB of signal to attenuate perhaps 60 dB of IF blowthrough isn’t a great deal either.
As 80 meters is a primary band of operation, and perhaps a tunable IF if I use converters for the other bands, I may need to change the IF of the receiver. The junk box has a few flea market filters at around 5 MHz from junked Swan and Drake rigs, and I doubt this receiver is going to get used on 60 meters, so that may be a much better option.
A review of the IF amp situation
So I’m questioning the choice of my IF frequency, and still wondering if there is enough IF amp gain. Adding a second IF amp stage could help, but maybe a check of gain through the filter is warranted as well.
Near as I can tell, I have a filter matching issue dogging me. I had tested the old Heathkit filter on a spectrum analyzer, and saw a 30 dB loss before properly matching the 2000 ohm impedance of the filter to the 50 ohm impedance of the analyzer. A couple of L networks confirmed that properly matched, the filter loss was nominally 10 dB, about right for a filter of that genre. But installed in the radio, I may be seeing as much as 25 dB loss though the filter – it is tough to get an accurate number doing a quick voltage check with the scope because of matching and in-circuit measurement issues using 50 ohm test gear in vacuum tube rigs – something to learn more about for sure.
Now what? I could either try the original half lattice filter design, as I know that is properly matched, and I’ve found a couple 3395 crystals that will work. Or I could do some guessing as to the impedance of the tube circuits, and try to use a couple of L networks to better match the Heathkit filter.
I did both. And interestingly, in both cases, with a better filter match, my IF amplifier turned into one heck of an oscillator. Signal levels were pretty good when the IF transformers were tuned slightly off resonance, but the scope showed a big whopping 3395 khz oscillation as either transformer approached resonance.
I’m not certain what this is telling me, but my suspicion is that in using exisiting chassis mounting holes and cramming the filter and IF circuits pretty close together, I made a collosal blunder.
A way out of the rabbit hole?
Now what? You can see what’s happening next below. That black box is a new IF channel sub-chassis built in an old MFJ antenna analyzer box (remember this is a junkbox project – I’m not proud of the way it looks but my old dead MFJ antenna analyzer is now serving a useful purpose). The chassis has a second junkbox Heathkit filter at 3395 khz, and 1, but optionally 2 6AU6 if amps (more gain, and better AGC range perhaps).
This chassis is arranged linearly, with the input far from the output, with internal shields in what I hope are critical locations, and can be located as sitting in the picture at the top of this post, or, if it works well, can be dropped into the current IF amp/filter space if I move the oscillator tube.
Having a sub-chassis like this will let me experiment with the 1 vs. 2 tube IF strip, and IF filter matching, as well as easily change to a different IF filter if I change the IF frequency. So that is this week’s project. Hope something good happens……..