I have just bought a NATO surplus CW filter.
Seller's description (edited down) is: "CW ... quartz filters of the English company Cathodeon.... The filters are intended for use in professional radiocommunication equipment.... They are made in the McCoy arrangement.... The center frequency of the filters is 1.75MHz.... The frequency response of the CW filter is 400Hz."
Compared to a lot of other offers on the watery auction site, this was reasonably priced and perhaps promising. It should enable me to experiment with the various ARRL designs that are based on a 1700KHz IF, as well as the one mentioned earlier by VK6HP. It only requires a modest shift in IF, and small changes in the winding of the inductors or their associated capacitors.
Trouble is though, 1.75MHz is a perfect divisor of the bottom of each of the original ham HF bands. Is that going to cause a problem by definition? Figured that out a couple of minutes after clicking "buy"....
Martin,
I don't think you will have a problem. In fact I want one of those filters!
That said, here are some things to consider:
1) Having the filter at 1750 kHz means the BFO will be at about that frequency, and you WILL hear the harmonics. This isn't necessarily a problem because you probably won't be pushing the lower band edge that closely.
The trick is to set the BFO slightly BELOW the IF - say, 1749.250 kHz. That puts the second harmonic at 3498.5 kHz (1500 Hz below the low end of 80) and the fourth harmonic at 6997.0 kHz (3000 Hz below the low end of 40).
I know this from experience. The Southgate Type 7 uses an IF of 1400 kHz, and the BFO is at 1397.250 kHz. The fifth harmonic of the BFO is at 6996.250, and causes no problem - but it IS there on 40 meters. The tenth harmonic can be heard very faintly on 20 meters but is so far away that it's not an issue.
2) To use the Cathodeon filter (or any such) to best advantage, you need to know its characteristics - input/output impedance and loss. Small errors in impedance matching don't matter much, but big errors result in odd filter responses and excessive loss.
In my homebrew rigs using packaged filters, I've been lucky to have the filter characteristics, so I could design the circuit to match what the filter wanted. The matching circuit I've had great success with is a simple parallel-tuned circuit using a powdered-iron toroid, with a link winding to transform the high impedance of the tuned winding to the low impedance that the filter wants. A trimmer capacitor permits tuning for a match.
3) The 1750 kHz IF is perfect for a band-imaging 80/40 receiver - you've probably seen several designs for such receivers. With a 1750 kHz IF and a local oscillator tuning 5250 to 5400 kHz, the receiver will tune 3500 to 3650 kHz and 7000 to 7150 kHz, which should be more than adequate for CW operation.
However, there's a catch.....
The weakness of band-imaging receivers is that the only thing that prevents you from hearing signals from the "wrong" band is the front-end selectivity. The tuned circuits before the mixer must pass the band you want and reject the band you don't, otherwise you'll hear strong 80 meter signals when tuning 40 and strong 40 meter signals when tuning 80. This is why many of the band-imaging designs have a double-tuned input circuit.
4) There were several ARRL receiver designs of the 1960s that used the 7360. They work well, but the 7360 has become rather scarce and high priced. A suitable replacement is the 6JH8, which shares none of the problems of the 6AR8. There's also the 6ME8, which is somewhat less common. Note that these tubes have different pinouts and are not directly interchangeable.
IMHO, for a first superhet, a better choice is a Pullen mixer. It is simpler to implement and can use common dual triodes. For some reason the ARRL never gave it any ink, but the RSGB Handbooks of the 1960s did. You can always start with the Pullen mixer and then try a beam-deflection mixer if you so desire.
I can send you several articles on the Pullen mixer if you are interested.
More to follow....
73 de Jim, N2EY