For a few years now I've had high-speed cable modem service from the local cable franchisee. More recently I've decided to become more active with amateur radio. In setting up a very modest station I noticed varying high noise levels on several of the amateur bands, most noticeably on 160 and 10 meters.

At first I thought it might have been just normal residential type random noise events manifesting itself as ignition noise or what might be perceived as fluorescent light noise. I am lucky enough to have a very good spectrum analyzer at home and started looking around my home with some near field probes. I also used my ham antennas as inputs to the spectrum analyzer.

What I saw astounded me.

Over the frequency span from 1 MHz to 30 MHz, I noticed what might best be briefly described as a radar signature spectrum located at several frequency bands. It was a typical sin(x)/x response, indicating a pulse source of some magnitude generating RFI, which was physically fairly close by.

Well, my first step in many of these investigations is to turn off my local power at the main circuit breaker box. As soon as this was done all vestiges of the noise disappeared, both on the radio and the spectrum analyzer. I started turning things back on to the AC power and finally located two really bad offenders.

I have a Sony LCD monitor for my computer, which uses a 16-volt switching power supply to power the monitor. I also have a cable modem made by Ericsson (which is now supported by Aastra), which also has a small switching power supply. Both of these power supplies are in the form of wall-warts - the Sony supply being a bit larger.

I decided to see what could be done to improve the performance of both of these switching supplies in the area of RFI. I might also mention that I am a degreed electrical engineer, having a number of years experience in RFI related issues, as well as other areas of design.

Soooooo, off to the races...

Luckily the Sony supply had a couple of screws holding the plastic cover on the unit. I removed the cover to find a nicely constructed copper enclosure. Roughly rectangular in shape with the ends open on each end where the AC power cord and DC cord exited the enclosure. As soon as I put the near field probe anywhere near the supply I saw the radiated noise spectrum.

I also have access to absorbing clamps used to measure cable radiation. These clamps are surprisingly and accurately correlated to a near field FCC site we use at work for qualifying our own designs. The absorbing clamps also yielded the same noise spectrum.

At this point, I decided to try some simple filtering with discrete components on the inputs and outputs. I tried some ferrite cores on the power leads but nothing seemed to work. The Sony supply was relegated to the junk box and replaced with a small linear supply. Problem number one solved.

The cable modem switcher was next. This proved to be a bit more of a challenge. This supply provides about 7-8 watts of power to the cable modem. It also has two output voltages, 5 volts and 7.5 volts. Replacing these with a single linear supply is not so easy.

I first contemplated designing my own linear supply with two monolithic regulators for each of the DC voltages. I quickly realized however that I was more interested in taming this switcher. Perhaps I regarded it as a bit of a challenge!

Neither switcher had any kind of filter network on the input AC power, other than a common mode choke that is. (Okay, so a common mode choke is a filter, but in this case they didn't seem to be doing much in terms of conducted RFI.) The clamp-on ferrites I added didn't seem to work either. I installed them on the output DC cable wiring. (Ed. Single clamp-on ferrites and beads generally not effective at HF. Toroid cores are generally recommended for these lower frequencies.)

One interesting observation with both supplies was the frequency range of 1 to 30 MHz yielded much higher radiated interference compared to what I saw above 30 MHz. The relative difference was sometimes as much as 40 dB. It was almost like these supplies were designed to favor the radiated emissions requirements since the lower frequencies need only be tested for conducted emissions.

I can't provide absolute field strength intensity in dB/m as my calibration data for the absorbing clamp and near field probes is not directly correlated to a near field range. (What I'm calling a near field range is the 3 to 10 meter setup used for qualifying class A and class B electronics to the FCC specs.) What is important, I believe at least in these two examples is that the radiated energy was quite severe in the 1 MHz to 30 MHz range, an area the FCC does not require radiated emissions compliance. Only conducted emissions are tested below 30 MHz, and I have sneaky suspicion that both of these devices would fail if retested in this area.

Any way, what I did do with the cable switcher was to open the plastic box and add a power line filter module to the input AC line, then connected the common (ground) lead from the DC output to the safety ground terminal on the filter module. The AC and DC sides of the switcher were still completely isolated from each other. This was the fix that worked well in this particular installation. The broadband spectrum of noise was reduced well below what was initially observed, and as far as I can tell, CANNOT be heard on any of the amateur bands in the radio.

Ideally, the cable modem should probably have a linear supply, and I will probably get around to designing one sooner or later. My points for writing this is to pass along some information as to potential sources of noise that other folks in the amateur radio community might be seeing. Granted they may not be as easily solvable as my case. And certainly the manufacturers of these types of equipment are not going change their power supply designs. But there may be some supplies that are more noise friendly than others. How to find out about them is more than likely a gargantuan effort. Maybe some type of database could be generated from interested parties????