Wadley Loop RX

Small and slow progress lately, having completed the 37.5 Mc Wadley loop section. The first iteration gave only a few millivolts of RF at the 6BE6 second mixer, nowhere near enough on grid 3 to get any decent mixing action. Frustratingly, the grid input coils, and anode loads were VERY broad in tuning for a peak, heavily loaded by the valve, a good sign that they were having trouble with transition time issues. Turns out, changing the two IF amplifiers from 6BA6 to EF80 frame grid valves, the gain increased dramatically , I now have 2v p/p , this over a bandwidth of around 400kc, all coils now with sharp tuning peaks.. The 7 pin pentodes appear to be very limited at higher frequencies, possibly loose construction tolerances? I may also need to replace the 6BE6 heptode mixers s with something better also. ... perhaps ECC81s?

Any drift in the first VFO will be cancelled over this bandwidth, ie, +/- 200kc or slightly less. The coils required were wound on ex TV formers, two 10 turn coils on each former separated by 1.5 inches to get the right coupling. By injecting a 37.5 meg signal into the first mixer, mixing products were heard at the expected plus and minus via the tuneable IF, so it is working so far. Now its time to build the harmonic generator, and the VFO. We are coming into summer, so progress will probably slow.

Can you use a series combination of a high pass filter cutoff at 39 MHz and a low pass filter with a cutoff at 40 MHz to get the desired 1 MHz band pass?

It might eliminate a lot of the tedious fiddling with the coupling and make steeper skirts on the filter.

73,

Win W5JAG

L/C Filter

Thanks Jag, all sorted.

The filter block experimental prototype took HOURS, there is a lot happening with few parts. Obviously, the coils need to resonate at the correct frequency. Then, because each individual tuned circuit resonates at just one frequency, we need several to widen the passband. If each tuned circuit is too higher "Q" then there will be a gap between each peak, this is called ripple, so we need more TCs to "fill in the gaps". Then they need to all be coupled together somehow. Proximity will couple these inductively, but too close will cause "double humping", too far apart will increase "Q" of each and cause ripple. Capacitive coupling is tricky, each coil at resonance is very high impedance, and even vey small stray capacitance has a very large effect. But, we need some capacitance to swamp out the strays, to make the design more robust and more repeatable.. The individual "Q" of each TC is largely controlled by the L/C ratio. More coil, less capacitance makes for a higher "Q", less coil, more capacitance will have the opposite effect. Phase changes too between inductive and capacitive coupling can cause dips and peaks in the passband.

I see no way to model these things, I chose the "suck it and see" method.

Using a spectrum analyser with a noise source seemed the only way to actually see what is going on.

First problem, the SA has 50 ohm input/output impedances, so direct coupling will short out the filter, A temporary two turn coupling coill was wound on the two outer filter coils as an artificial tap. With each TC having 13 turns, that gives an impedance ratio of around 50 to 1, or 2500 ohms , probably around the same loading as a pentode control grid at 40 megs or so.

The coils were wound on the formers and resonated with a 15pf cap , the slugs allowing fine tuning. The spacing between each TC was changed until the individual peaks began merging into one. The coupling capacitances were then varied to massage the response curve into what is required.

Thats why it took 5 hours to get it right, and keeping it as simple as possible.

At this stage we have 5 tuned circuits at 13 turns on a 1/.4 inch former, spaced at 18mm between centers. The resonating caps are 15pF, as are the interstage coupling caps. The input TC is inductively coupled only to the second TC, removing a pronounced dip in the passband center point. Once I add a screening box around this, its all likely to change.....

Perhaps QRP could try modelling this on his computer simulator??? Is it even possible???

Those last two photos are crap, I have no idea how to remove them????

5 LONG hours and

a swear word or three got me this.

Finally have designed a reasonably simple bandpass filter, 1 meg wide, down 20dB at 500kC either end, no passband ripple, just the ticket for both the receiver first IF and the Wadley loop amp, the same design will do both, one being tuned to 40 meg Cf, the other at 37.5 megs. Two cascaded should net another 10/15 dB attenuation on the skirts.

There are two gain stages, one between this filter and the next, and the second to match into the second IF mixer. The signal here must be over 5 v p/p to give good high level mixing. Looking at the response curve, just one may be enough for the receive IFstrip, Each harmonic of the VFO is at 1 meg intervals, so the loop should not have any response to the ones each side, so at Fc of 40 megs, we want it deaf to anything on 39megs and at 41 megs, otherwise these signals will cause bleed through of any station on adjacent bands above and below causing birdies. Two of these filters cascaded should be enough. Here is the prototype experimental filter......

I added a temporary coupling coil front and back to allow the 50 ohm spectrum analyzer inputs/outputs not to load the filter, being valve stages these are high impedance, so a quick and dirty impedance match was needed. At 40 megs, valve signal grid impedances are beginning to drop, I may need to tap these down on the input coil..

And, just when you think we have seen it all, I came across this......

https://www.youtube.com/watch?v=Mfly1pGgpqY

A little progress today, but ran out of solder......