The W7ZOI active low-pass filter for CW.
January 7th 2016. The perfunctory lash-up you see above, is of a circuit designed way back by the renowned Wes Hayward, W7ZOI. It was published in the U.S. magazine ‘Ham Radio’ in April 1974. We currently have a project to build a receiver to go with a couple of QRP (low power) CW (= Morse) transmitters we knocked up a few months ago. Our attention span is usually very brief, but surprisingly, having started at Christmas, here it is the 7th January and we’re still going at it! The receiver will be simple – otherwise we’d never be able to make it – but attention is being paid to details, in order to get over the deficiencies simple receivers tend to have. Which are usually numerous. 8^) And, as we all know, it is much more difficult to home-brew receivers than transmitters.
One of the things we’ll certainly need is an audio frequency filter for listening to Morse code. In fact, an audio peak filter to emphasise the desired signal is an indispensable feature of any receiver, especially for CW. The main station here is an expensive Kenwood TS-590, and this is indeed equipped with means of selecting and also varying the pitch of the incoming signal, and amplifying the result, not to mention noise reduction, noise blanking &c., &c. The operation of these controls soon becomes more or less second nature, but I have found there is a tendency to keep fiddling with the settings to further optimise reception. However, this is all too often counter-productive. Because we usually operate low power (5 Watts or less), and work other low-power stations, fading (QSB) is a major problem. If one has become too selective when the incoming signal is loud, then when it fades down, it usually disappears altogether. Naturally, this always happens just before the other station is about to end their transmission and hand over to you. So you can’t be sure that they have actually done so. What usually happens, is that you wait until they must certainly have handed it back, and start sending to them. You send something like ‘TNX INFO DR OM’ – which of course really means “I have not been listening to what you just sent 8^) – but by then they have started calling you to see if you are still there. So you’re both transmitting at the same time. Result? Failure of the QSO, leading to:
Well, we don’t really want any of that, do we?
No. What we – I – need, is a simpler way of isolating the desired signal. Especially for my receiver project, because it will be of the Direct Conversion type. (More on this on a page to be written later.) Hence my search on-line for peak filters of that kind.
Wes Hayward, W7ZOI, has for many years been pre-eminent in designing highly-reasoned, efficient and often commendably simple amateur radio circuits for the home brewer. So when we found the above, it was a Must Try. Of course, it’s a retro circuit, being over 40 years old; but it’s simple, and uses still-available and extremely inexpensive parts, so we gave it a whirl.
Diagram, with acknowledgements, from website http://kambing.ui.ac.id/
Actually, W7ZOI’s circuit must have originally been from somewhere else, but hope it’s OK to copy the above here. The red box emphasises a dotted line, in which further identical stages may be added.
The red box contains the input circuit, while the following blue boxes are each stages which will put in a peak, and also amplify it. It is possible to have several in cascade, the final signal being taken via the 10µF electrolytic at the far right.
The one I made – in the photo. above – had three peak stages. Centred on 600 Hz, its performance was excellent – indeed, almost over-kill! Here’s a sample of it, on 7 MHz in this January afternoon, with the TS590 on full CW bandwidth, & RF gain about halfway:
Right click & open in new tab: 600 Hz test.
An audio oscillator was taken, with its output at a lowish lever, and fed it into the filter in 13 steps. The output voltage was recorded on an ordinary multimeter. Of course, the x axis should be logarithmic, but we haven’t cracked Excel to that stage yet. 8^) Still, it’s clearly a superbly effective circuit.
There was only one slight drawback: the centre frequency of ~575 Hz was a little higher than I prefer. Back in the 1980s, one of my main receivers was the venerable (and gorgeous) HRO 5-T, and you could adjust the crystal filter to give you whatever pitch you liked for Morse. We soon decided on about 400 – 450 Hz as our preference. Well: it should be simple enough to lower this filter a little. In spite of all the graphs &c., we aren’t really all that technical, and it was not obvious to us which capacitor needed changing to lower the centre frequency. But after much googling, there was a distinct impression that it was the 6.8 nF from the base of the 2N3094s to deck.
To prove this, we built another, single stage circuit on a re-cycled bit of board…
… and plotted it with its original value of 6.8 nF:
Then, the 6.8 nF was changed, on spec., to a 10 nF:
Lo: Sod’s Law must have been temporarily in abeyance, as we had hit 450 Hz pretty well on the nose! It didn’t take long to swap the three 6.8 nFs to 10 nFs. Another test showed that the 3-section filter was still on 450 as near as dammit. We won’t bore you with the chart, even though I’ve discovered they’re quite fun to make - it reminds me of being back at Technical College, incalculable ćons ago… Instead, here is another sample off 7 Mhz, this time 1000z, 8th January. (It’s true that in morning conditions, the TS-590s inbuilt filter is excellent with loud signals; but it’s set here wide open & with full RF gain, so only the W7ZOI filter is operating.)
Right click & open in new tab: 450 Hz test.
It might cound a bit sepulchral to you, but it’s just the job for me. (See Appendix 1.) So there we are. Of course, there are times when signals come sailing in and one does not need a filter at all. So we can tidy this board up a bit, & put it in a nice box, with a switch & it will become a useful module for our receiver project. Thinks: I wonder how much current it draws, and whether it would run on 9 Volts? I don’t know; I’ll go and see… [5 minutes later]: Yes, it seems perfectly happy on 9 Volts at 10 milliamps. So, how long will a PP3 battery last? I don’t know; I’ll go and see… [5 minutes later]: Apparently a standard alkaline PP3 is rated at 550 mAh, so it wouldn’t be too bad. And it would save an extra lead sprawled over the operating position.
Very well, then. The board will go into a box with two phono sockets on it for in and out, powered by a PP3 battery. It’ll also need a pot. to control the input to the board, as it may be run from a headphone socket. Also, of course, provision must be made to switch in none, one, two or all three of the filter stages, as required. I wonder if it can be done with three miniature toggle switches rather than a rotary switch? And the first one of them can turn the power on too. Toggle switches are much easier to flip casually, rather then grab hold of a knob & turn it. Indeed, this easy flipping of toggle switches has been known in the U.S.A. for a very long time: their toggle switches are usually UP for on, and DOWN for off. That means, if you fidget about a lot and accidentally press down on a toggle switch, you will turn something OFF, rather than ON. Whereas in the U.K., they are the other way round. Look:
This is an escutcheon from on of my stock of new toggle switches. It is probably Japanese, so naturally observes the U.S. tradition as described above. In general, I think I have to agree with the U.S. approach. Though it all depends on statistics, doesn’t it? How many accidents have occurred when something was accidentally switched OFF, as opposed to how many have happened when something was accidentally switched ON. In theory, we would expect a deleterious event to be more likely when something was switched ON. Like a friendly neighbourhood mobile but psychotic chain-saw. OTOH, if you accidentally switched OFF somebody’s life support system, that might be difficult to live with. But I digress…
Appendix 1. Here write about the two different freqencies at wch the human ear seems to work best.
Page started 7th January 2016.