Shortwave Noise Filter (literally).

January 28, 2006 7:31 PM

ShortWaveFilter (literally): My outboard noise canceller is actually increasing the noise...

I live about half a block away from a busy thoroughfare in Cambridge, MA, in a small subdivision of interconnected condos. Because of the close proximity of other homes to my radio/antenna, the whole electromagnetic mess of the neighborhood comes out on my shortwave loud and clear, as you might expect.

To work around this, my family bought me a Timewave ANC-4 noise cancelling unit to interface with my Kenwood R-5000 radio and Wellbrook ALA-1530 outdoor loop antenna. (The daisy chain goes ANTENNA > ANC-4 > SHORTWAVE.)

In principle, the ANC-4 works by "tuning in" the surrounding interference (controlled by a Noise Gain pot), at which point you use a phase-nulling pot to pinpoint and create an identical counter-signal to all the noise, thus cancelling it out.

The setup of the ANC-4 is a little wacky, insofar as it comes with a built-in whip antenna to pick up the immediate ambient noise, and also has a plug for connecting a longwire antenna that would attract outdoor and remote neighborhood noise. It comes with both, although the latter is just a cheap piece of speaker wire clearly thrown in as a token gesture to get the unit up and running.

The ANC-4 is doing an excellent job at attracting noise -- more noise, in fact, than the Kenwood even picks up on its own -- but it isn't cancelling it out at all. I've tried every subtle variation between the "Noise Gain" pot and the "Noise Phase" pot under the sun, and it only creates more and more noise. When I turn the ANC-4 off altogether, the noise disappears, and the reception is clearer than ever. There is a LOT of noise on the bands without any sort of filter, but the ANC-4 sure isn't "hearing" it.

It occurs to me that the ANC-4 may need a better antenna to be able to detect the noise, but because of my space situation, I can't really provide an elaborate longwire for the noise canceller any more than I could put up a longwire for the radio itself.

Any thoughts on why this thing isn't working? Feel free to ask more questions. Or, of course, you could come to Cambridge and I'll pay you to test the cotton-pickin' thing yourself.

posted by mykescipark to technology (5 comments total)

A broad band analog "noise canceler" such as you describe flies in the face of the theory of operation of superheterodyne receivers, such as your Kenwood. The basic idea of a superhet is to tune (and possibly amplify) RF signals broadly with a relatively low "Q" front end circuit, convert that output to an intermediate frequency which is always the same for any tuned RF frequency, and using a circuit highly optimized for the known IF frequency (high "Q" or tight bandpass characteristic), do the demodulation on the IF stage, thus extracting the maximum useful signal present in the original RF signal. In a modern superhet set such as your Kenwood, enough feedback between the IF stages is employed to make it nearly a perfect detector, and it will be capable of rejecting all but the tuned signal while maintaining a signal to noise ratio in excess of 50 db. That's at least an order of magnitude better selectivity than anything in your ANC-4 is providing (based on 40 db cancellation spec of the ANC-4). And of course, the ANC-4 is only primarily aimed at providing a deep null filter for relatively low frequency local noise sources like power lines. It won't really help much with RF spectrum pollution.

As you are learning, it is hard to beat a really good superhet...:-) Unless you are willing to go to extremes.

Your problem is analogous to those of radio astronomers who are faced with pulling extremely weak signals of interest out of the soup of competing, and often far more powerful signals in the electromagnetic soup surrounding us all. They go your ANC-4's tricks many times better in a number of ways. For starters, they supercool their front ends to get rid of most thermal noise in their antenna/first stage detector chains, particularly when looking in the far-infrared and upper microwave regions, where infrared star emissions are the swamping phenomenon. They also tune very narrowly early on, and use the maximum practical antenna sizes and configurations. And then, they do a lot of post processing on the signal. Unfortunately, the only one of these tricks you can readily implement as a short wave listener is to tune more narrowly early on.

Instead of taking the broad swath approach your ANC-4 attempts, try to characterize the local noise sources that are interfering with your shortwave enjoyment, and reject only the worst of those with some optional filters that can be added to your Kenwood. The high selectivity of your triple conversion superhet circuit in the Kenwood will pleasantly surprise you, once you keep it from being swamped. But in addition to this, if you are being swamped with low frequency noise, its worth taking a walk around your neighborhood with a small sledge hammer and an AM radio. No, really. It is.

Most powerline noise sources are due to cracked insulators on distribution poles or transformers. All power companies are obliged to respond to reports of noise interference, and make reasonable efforts to mitigate it, and they want to do so, because finding and fixing these problems also makes for lower leakage losses from their lines, and forestalls accidental electrocutions. The time honored way of finding these things is to wait for a few days of very dry, sunny, cool weather. Taking an AM pocket radio, walk around until you find a point of particularly bad reception. Locate the nearest power pole, walk over and give it a good whack with your hammer, enough to shake the pole slightly. Bad insulators will sing out, some times audibly, but always with a satisfying burst of static in the AM radio. Call the power company, tell them you are a HAM operator, and want to report an interference source coming from their lines. They'll send out a crew. More than once, I've met crews at the offending pole, and demonstrated the problem, which they were happy to fix.
posted by paulsc at 9:14 PM on January 28, 2006

What he said.
posted by neuron at 10:15 PM on January 28, 2006

Paul - Comprehensive and excellent response. I thank you. However, I regularly conduct at least one-half of the test you describe (minus the hammer):

Prior to rigging up the Kenwood, I like to take my Sony ICF-2010 portable out to the highest point in the city (Brookline) and had a listen to the bands in a more "clean" environment than I get in Cambridge. Occasionally I do this because (a) I know I'll actually get some serious listening done, and (b) I like to have an accurate benchmark for local reception conditions from my home shack.

To complement this activity, I have certainly done the walk around my own neighborhood, and I find that I have to walk all the way out to the middle of a huge baseball field, as far away from possible as the streetlights and power grid overhead, in order to approximate the conditions from my hilltop spot in Brookline. Even then, the neighborhood noise leaks through - noise which is absent from the radio only a few miles further South in Brookline (in an equally population-dense neighborhood, I might add).

Although I am no great radio scientist, I think this verifies that I am dealing with highly localized noise of the sort you describe (power lines, dimmer switches, electric blankets, etc) - and not just ionospheric messiness of the sort I presume you are intending by pure "RF spectrum pollution," which I characterize as sunspots, geomagnetic activity, ionospheric disturbances, and other non-man-made things that wouldn't be affected by a simple vertical move of (basically) peanuts from trough to hill. Anything of that magnitude of ionospheric noise wouldn't care if I was in Brookline, Cambridge, or Cape Cod, would it?

(Am I making any sense at all?)

Oh, and one more thing: Your suggestion is that the ANC-4 is only for use with analog tuning circuitry, am I scanning that correctly? Who on earth would make such a thing in this day and age? I've read about it being tested with other Kenwood products even more recent than mine, with allegedly great results. Perhaps here, again, I am confused - but I'd rather waste your column inches addressing the above part first... ;-)
posted by mykescipark at 12:41 AM on January 29, 2006

I took a look at your ANC-4 data sheets and schematic, and I thought about your response post a bit. I used to live in Waltham and Lowell, and so I have some familiarity with your geographical situation. Totally as an aside, but as a suggestion for a little day trip you might be interested in sometime, I'd suggest a visit to Hammond Castle up in Gloucester. Not so much for the castle itself, but because its builder, John Hammond Jr. was a radio pioneer whose contributions were of such value to the U.S. military, that they obliged him with the creation of a low interference radio zone around the castle, which persists, to a lesser extent (thanks to cell towers, etc.), to this day. If you have some portable receiving equipment, I think you'd be surprised what you can hear around the castle, and it's kind of the ultimate solution for the problems you're having, but unfortunately, not something you can easily replicate in your location. Still, it can give you a whole new look at what the natural spectrum noise floor really is.

And while we're taking care of extraneous issues, let's deal with your last paragraph. Most "modern" radio receivers, iincluding your Kenwood, are, in their detectors, still basically analog devices, because the superhet circuit is the best way of receiving real time tunable radio transmissions. Oh, most modern receivers do use a lot of digital circuitry, to develop precisely controlled digital clocks sources, in place of the old analog oscillators, and to provide control and display interfaces, as your Kenwood does. And the Phased Locked Loop (PLL) has been a tremendous boon to receiver stability, and relies on such precise oscillators. Even some detector designs and certain filter networks in the RF and IF stages have been improved by digital advances, and in the most modern designs, digitally analyzed and controlled gain circuitry. But there is a limit to what you can do with a signal source such as an ordinary Single Side Band (SSB) shortwave transmission, as our friend Shannon has long noted.

It's a matter of some debate, especially among communications theorists and quantum physicists, whether Shannon's law is the inviolate limit for channel capacity it has long appeared to be. Designers of end to end communications systems can seemingly circumvent his theoretical fence by using knowledge of the signal outside the actual channel to provide reconstruction of a signal beyond what Shannon described, and this is what is currently going on at the leading edge with military communications and deep space probes, both of which encode error correction schemes in their signals, as keys for reception processing. But that is not the way shortwave radio works, nor is it something your Kenwood will likely ever be equipped to do. So, for the purposes of this discussion, we are children of the lesser god Shannon, and need not worry ourselves with the interesting but arcane magic considered vital down at the Pentagon these days.

But to return to your issues.

The ANC-4 has its place, and might be useful for solving specific problems, which I'll get to in a bit. But I fear that it is never going to be a "fire and forget" solution, which you set up carefully, and can then expect to work reliably to make your abode the radio equivalent of Hammond's castle. In the best case, it can help greatly to null strong, well characterized nearby noise sources, preventing the Automatic Gain Control (AGC) circuitry of your Kenwood from being overdriven, thereby returning your Kenwood's sensitivity to something like it would have been outside the presence of that interference, and this is the anti-swamping role for which it seems designed. But as a general purpose device, it does so with the equally likely probability of making many situations far worse (as you are learning), and at its maximum effective filter settings, it will always have the potential to "break over" into this bad behavior, making your overall problems worse with it, than without.

Potentially, in the best case, the ANC-4 can do something your Kenwood will find difficult, in that the ANC-4 is designed entirely to circumvent Shannon limits by comparing a noise source to a signal source, and summing out only the noise, with minimal impact on closely adjacent frequency signals. Your problem is delivering a noise signal of such stunning purity (only noise> and exact phase replication to the mixed signal source, that it can do this infallibly and instantly. Unfortunately, physics are not in your favor, unless your noise case is very simple.

For one thing, unless the noise source of interest is both pretty regular and pretty far removed in frequency from your signal, it is going to be well nigh impossible to construct a noise gathering antenna that gathers only noise, while maintaining an acceptable phase relationship, upon which simple summing type cancellation depends. The reason for that is that the noise antenna will need to be within something less than a wavelength of the signal antenna to be symmetrically affected by noise in the same way as the signal antenna, and to the same degree. For long wave sources of high intensity, you may be able to get an acceptable physical separation of the noise and signal antennas, and construct a noise antenna which, by physical design, is much more sensitive to the strong noise source than the signal antenna is, which would be the best possible case for the ANC-4 to provide some benefit. This would be the case if you lived next door to a 50,000 watt AM clear channel broadcast station like WSM in Nashville at 650 KHz, and wanted to listen to BBCeu on 12095. You could set up a simple AM long wire antenna as your noise source several yards away from your shortwave antenna, set your ANC-4's controls for maximum nulling of the strong WSM signal, and you might get the 40db of null of the WSM signal that the ANC-4 claims, if your transmission lines to the ANC-4 were phase matched as well. Assuming your Kenwood was otherwise well shielded from the WSM air signal, its RF stage would then see 40 db less WSM, and its AGC circuit would hopefully not be swamped. You might even reliably get the BBC, assuming nothing else was a reception problem and you were delivering an acceptable level of the 12095 BBCeu signal to your Kenwood for it to tune.

But let's now say that an obnoxious new neighbor moves in next door, who is an avid CB'er, who operates a badly tuned but powerful linear amplifier on his CB rig. It's totally illegal, and he doesn't know or care enough about linear amps to fix the emissions he's splattering all over the neighborhood, and for some reason, you don't want to take up the matter with the FCC, and get no joy by trying to make common cause with broadcast engineers over at WSM against this new cretin. You find that your AM noise antenna does pickup this new source of interference, but not as well as your shortwave antenna, and you determine to add another specialized antenna to your anti-noise system to provide the best noise source possible for your ANC-4 from this new source. So, you do a bit of experimenting, and find that you can put up something that gets a whopping good sample of his illicit 49 Mhz primary, as well as clipping splatter sub-harmonics of his signal at something like 25 and 12 Mhz, and you make a little combiner that also pads the relative signal level differences you get from the strong WSM signal and the weaker CB'er, to approximately the same absolute values, and feed this new noise sample to your ANC-4, along with the feed from your AM long wire. What is the likely result?

That's actually kind of hard to say. You'll probably find that you do get some cancellation of both the new neighbor and the WSM primary, but not nearly as much of either as you hope. You'll be riding your ANC-4's gain controls a lot, and your Kenwood will be a lot less likely to be pulling in the BBC noise free. The reason is that, by putting up another noise antenna in your radio field, even ones that are physically as frequency selective for the identified noise sources as I've postulated, you're presenting a combined noise signal that is now phase smeared in comparison to your signal antenna, for both of the noise sources you are trying to cancel. You might find an acceptable physical location for both noise antennas where everything will work out, but it is going to be tricky, and I doubt you'd ever get the ANC-4 to get much more than 10 or 15 db of cancellation of the CB'er, while still rejecting the very clean but very powerful WSM signal to an acceptable degree. The ANC-4 is just not that sophisticated a device. So, since the WSM signal is so strong, you might decide to drop the long wire AM antenna altogether, and just use the combined signal from your new anti-CB'er antenna, since WSM is bound to ride in on it too, although not nearly with the amplitude it had from your AM long wire. This at least eliminates the phase smear at 12, 25, and 50 Mhz your multi-point antenna experiment may have introduced, but you've probably dropped your WSM noise signal level 30 to 40 db by doing that. Still good if WSM was riding in at 80 db, and it might be a good solution, but that's not the ANC-4's only problem with this setup. Even at full illegal output power, and considering the differences in relative sensitivity to the signals of your new fairly short 50/25/12 probe antenna, the CB signal level is more than likely still dwarfed by the very strong WSM signal, in absolute terms, and you've no way to tone down the WSM signal appropriately so that you can squelch the combined interference signals from both sources with the single gain control on the ANC-4, unless you also add a high pass filter with attenuator that discriminates against and level matches WSM appropriately in your noise sample input line, ahead of the ANC-4. So you do that, and maybe all is well, until such time as your other neighbor's leaky but serviceable microwave oven lights up, at which point you're going to need to add another notch filter serially for that... I hope you see the problem here, which is that, as the signals you are trying to reject vary in number, power, orientation, and frequency, the problem of providing an appropriate noise "sample" that the ANC-4 can null with your signal source quickly becomes much more than just non-trivial. And I'm leaving out entirely any discussion of all the possible mis-adjustments you can make on the ANC-4 which will get it into a "break over" state where it is actually sending out noise of its own generation on your noise antennas, but there are warnings in its manual about this. I am also ignoring the fact that the ANC-4 will always introduce a minimum insertion loss of at least 6 db (according to its manual) in your signal line, just by being in your setup, thus cutting out entirely very weak stations that might otherwise be listenable on your Kenwood. You're getting away from what the ANC-4 was really designed to do, when you ask it to do very many things all at once. You need a better general solution.

That solution is actually your Kenwood's superhet circuit, along with appropriate in line filters in its RF and IF stages. The reason I say this is simple. What your Kenwood has going for it, that the ANC-4 doesn't, is the amazing power of negative feedback in the superhet circuit combined with an ability to incorporate multiple sharp filters and pads, in an ordered sequence, in that superhet. In essence, it is already the outboard filter network you would have to add to your ANC-4's noise antenna in the case I've just describe, but much more selective and steep, because of the superhet. In the circumstance I've described, I think you'd find your best solution would be to forget about getting the ANC-4 to take care of the CB'er, and the microwave oven, and just let it kill WSM. WSM is well characterized, clean, ever present, and a whopping big problem on its own in that locale, but the ANC-4 can probably deal with it effectively, with very little adjustment, and if you get rid of WSM externally, your Kenwood will probably be able to reject the other stuff acceptably, if you sharpen up its filters appropriately, and tune carefully. Even better, with the right filter setup, you might find that the Kenwood is fine on its own, giving you back the 6 db insertion loss the ANC-4 would otherwise claim.

But of course, to do this, your Kenwood has to be appropriately equipped, and operating properly. I mention this because Kenwood apparently stopped making this set in 1996, and if yours is of that vintage or older, and has a few thousand hours of operation on it, it may well be time for a "tune up" of your tuner, as some posters in this review thread mention. IF filters age with time and heat. Optional RF in line filters may never have been installed, or were installed in an order not optimal for your current problems. Oxidation of chassis joints happens over time. All in all, spending some money on getting a knowledgeable tech with the right equipment to bench your tuner may be the best thing you could do, as a first step. It's a good set, and probably worth spending some time and money to maintain. As to sources of help in doing that, you might be in touch with some HAMs in your area through the ARRL or the Boston Amateur Radio Club. HAMs in your area are also an invaluable source of advice about noise sources in your area, and techniques for dealing with them. As a group, they are also wonderfully helpful, often even going so far as lending test equipment, or making on-site visits to understand and help correct reception and interference problems. And/or you could try some R-5000 Internet resources such as these.

Once you are sure that your Kenwood is working up to factory specs, and maybe know for sure what optional filters it may already have, you could actually use it to examine and characterize the noise that is making you most unhappy, assuming you don't have a lot of filters in it already, or switch them out, as I think your Kenwood allows you to do from the front panel. Basically, you just tune slowly through the bands of interest over several sessions, and make note of any whopping big noise spikes you get. You learn about your radio neighborhood, and you try to see if there are broadly discernable patterns in the big noise spikes, that correlate with time of day, operation of equipment or activities you can see in your neighborhood, etc. It helps if you can get a reliable measure of noise strength and even direction, but this may call for setting up some additional gear, such as a directional antenna array, or outboard attenuators and S meters. Local HAMs may be of help, if you can get them interested. At any rate, sooner or later, you figure out the one or two worst problems you have consistently, and rig your ANC-4 for maximum cancellation of those sources, assuming, of course, you can't get them to cease and desist (see my previous comment regarding power line noise abatement, and learn about contacting the FCC for interference resolution).

I apologize for the length of this, and I wish I could be more supportive of your faith in the ANC-4 as a general solution. I hope at least I given you some things to consider that will prove ultimately fruitful.
posted by paulsc at 1:03 PM on January 29, 2006

I think this wins the Best Answer In The History Of Metafilter award.

That's a lot to chew on (although entirely welcome), so for right now, I'll just say THANKS!
posted by mykescipark at 6:38 PM on January 29, 2006