Improving WSPR Station Noise Performance, Part 2

In a previous posting I wrote that I thought a majority of 20m WSPR stations were having their receive performance substantially reduced due to high noise levels. I suggested that even without a calibrated S-meter or signal source the balloon WSPR stations currently floating around the earth could sometimes be used to assess station performance. I tried to demonstrate some evidence that most 20m WSPR stations do not function as well as they might.

I went on to mention the WSPR map Hearing/Heard-By ratio as a possible metric, a finger-in-the-wind, that corroborates these balloon measurements. In that posting I also indicated that although I thought the typical station performance was being severely compromised that there was good news – there is very likely something that can be done to improve it.

So, at the outset, I’m suggesting that there may be some simple ways to identify how severe one’s problem is, using either or both the WSPR balloons when they are within LOS of a spotter or by watching one’s Hearing/Heard ratio. A nearby LOS balloon should produce an S9 signal on 20m into a vertical dipole. A signal this large with even an urban noise level, one that the ITU considers “high” should produce very large positive S/N spots. Similarly a station even with typical ITU suburban noise levels should produce a Hearing/Heard ratio significantly greater than 1:1. There are a lot of factors that go into the ratio such that stating a single value as an all-time litmus test is probably a bad idea. But it generally seems that a 'good' ratio is almost always higher than 1:1. I have some ideas for scouring the database to give better clarity on some of this.

If your station fails the above tests, as I believe a majority will, I think it’s worth understanding how unwanted noise might reach the receiver. This can help eliminate it. Rather than throwing up one’s hands and saying “It’s all that digital noise now-a-days, there’s nothing to be done.” as seems to be a common response, I think it’s worth taking action and understanding what’s going on and seeing what can be done.

My own situation on 20m has looked pretty bleak for the last few years. When I first moved to the present QTH, a suburban environment with residences, no large industry and with rural agricultural area nearby, I had what I thought an acceptable location. I had antenna restrictions due to the HOA but with the antennas I could manage on MF through UHF, operating the ham bands was still interesting. But a few years ago things began to get really bad. My noise level on 20m began to rise so much that I could often hear few or no signals on an open 20m band. It wasn’t worth even trying to QSO under those conditions and even WSPR yielded only a few spots, even though N6GN often was heard fairly well. My noise floor was terribly high.

I could identify some of the sources causing this noise. I went through my own housekeeping and cleaned up devices over which I had control but even after that, I often had S9 noise, -73 dBm in an SSB bandwidth. I moved most of my operation up to VHF and UHF. That helped some but even at 2m my noise was normally 15-20 dB above the thermal noise from a room-temperature termination or -174 dBm/Hz. By 70cm, my noise floor was “only” elevated by 5 dB so I’ve operated a lot of 70cm WSPR. This has been interesting but recently I wanted to experiment with some new antenna designs and I wanted to do some of this testing at HF or lower.

As I previously described, one of the antenna designs I’ve been using on 20m is quite small, only 75cm (30”) tip-tip. Furthermore it isn’t nearly so efficient as a full-size 20m dipole. But it does have an advantage; it is small enough to be easily rotated in both azimuth and polarization, It’s a great tool for investigating local noise! On WSPR, this antenna didn’t initially perform too well. I was spotted only slightly less than a full-sized antenna but didn’t spot many other stations at all – pretty much the same experience I’d been having with conventional antennas.

It was a this point that I decided to quantify signal and noise levels coming into the receiver. Not only did I want to measure the efficiency of the antenna-under-test, I wanted to know the absolute noise levels and to be able to account for how and why they arrived so large at my receiver.

My noise was vastly larger than what reference literature said 20m band noise should be. Clearly the signals I wanted to hear were being buried. I then began to try to separate radiated noise, noise that was actually coming into the antenna I was testing from noise that was arriving in a different way, noise that was coming in through an “antenna” I hadn’t known or thought was even present. I found one of these paths in a very big way in the form of common mode noise. In the rest of this posting, I want to describe common mode noise. This has been described by others before now but since I believe the problem persists for 20m WSPRers I’m trying to restate it in the hope of helping others. If things go well I may describe the particular antennas and hardware that I’ve found so effective on WSPR and other modes here at N6GN.

This blog format doesn’t allow me to embed drawings, photos and graphics in-line, so as I discuss common mode noise, I’d ask you to click at the bottom of this page on the named file indicated so that you can look at it while I refer to it in this text.

The first figure is CM_noise1.pdf. On the upper left, I’m trying to show a simplified antenna, support and receiver. The antenna might be a dipole or a Yagi. It’s supported above the receiver and has a feed line coming down where it is terminated at the receiver. This antenna might be a dipole or Yagi fed with coaxial cable.

Looking to the right I’ve drawn in some more detail. There are a number of components here, some of them are desirable and some not
1. Desired Signal
2. Desired “Band” Noise
3. Undesired local radiated noise
4. Undesired local ground noise currents, includes rcvr connections:mains,LAN,other grounds...
5. Undesired Common Mode Displacement Current

(1) should be obvious. It is the DX signal we want to hear. (2) might be a bit surprising but I also class it as “Desired” since it is actually a DX signal. We want our antenna to be sensitive to DX signals of all types. If we are to selectively reduce DX noise sources, it will require directionality or polarization to reduce them. (3) is noise that is like (2) except that it is very local. This is noise that is produced by something within ten’s or maybe hundred’s of meters of our QTH. (4) are noise currents in the earth and ground systems below us. These act like noise in an “image antenna”, an antenna that is a reflection of the real structure e.g. antenna, feedline, and support, we use. Notice that there is a corresponding noise current in the real structure due to these ground currents. (5) is noise current that comes from very nearby high voltage sources. This is a high impedance path in contrast to (4) which is generally a low impedance one.

I’m presenting here a view that essentially says “Your antenna isn’t what you think it is.” While I’m doing this I should also point out that I’m actually also saying the same thing about the ground. It’s not nearly such a simple thing as many of us would have it be. Although there are lots of ground models and lots of variation among them, though ground is sometimes viewed as a lossy dielectric, in actuality it is non-uniform in time, place, depth and its effects vary with frequency. Because virtually all our antennas are influenced by ground in some manner(perhaps highly directive EME antennas excepted) this whole business is not precise and I’m not trying to present a precise description.

So (1) and (2) are from currents that we really do want to detect. We want those currents to flow through the load that our receiver presents. But (3)-(5) are all interferers. They are all due to something nearby and they are not part of the desired, normal part of our antenna system. These last three all act to produce noise current in our total structure. It is this result that the drawing at the lower left is meant to convey. Our actual antenna, the totality of “stuff” that can produce current in our receiver is more complex than the simple view many people may hold.

Another key distinction between the wanted and most of the unwanted noise sources is their mode. In the case of wanted signals and band noise, current into the two sides of our two-conductor feedline (shown as coaxial cable here but it could be balanced line just as well) is differential. Exactly the same current that goes into one conductor of the feedline comes out of the other. These currents are equal in magnitude but opposite in direction as they come from the symmetric dipole antenna. They are exactly 180 degrees out of phase.

The unwanted signals, except for radiated local noise that might actually be coming in the dipole antenna and perhaps noise due to some of the current in the soil under our antennas, are all common mode. Common mode signals are all of the same phase, not opposite like differential ones.

At the lower left I’ve simplified the whole mess by trying to show that the entire structure, what we previously called the antenna plus the support, feedline and ground are actually all parts of one rather complex antenna. This actual antenna, all that which collects and delivers signal to our receiver, is probably more complex than we generally think.

In the next figure, Coax Closeup.pdf, I’m trying to begin to show what we might do reduce the noise from the undesirable sources mentioned above. What we want to do is to deliver differential antenna currents to our receiver but minimize the amount of common mode current. Applying this to 20m WSPR, we want the total unwanted noise current at the receiver to be small compared to signal and band noise currents. We want to hear signals and noise that are propagated from great distance rather than anything that is produced locally. This is how we will “work the weak ones”. Generally we want the total of all these currents to be no more than about one tenth of that coming from the balanced antenna on top. Its power should be 10 dB down from the ‘band noise’ that is at the level of our noise floor goal. If and when we get our noise floor down to this level, our antennas will be working as well as they can be, considering their gain.

The device we use to accomplish this isolation of currents is a passive “balun” - a balanced to unbalanced converter. There are different designs and different forms but they serve the same purpose. The balun’s function is to isolate the antenna above itself from the common mode currents that may be below. These currents may have different source causes but we don’t want any of them to reach the receiver. To do this, the balun creates a high impedance path for those common mode currents.

This is equivalent to saying that only differential (balanced) current due to desired signal&band noise sources exist, goes into, and comes out of the transmission line going to the receiver below.

The entire antenna structure may operate with high noise voltage present but this common mode excitation will not produce significant current in the receiver below if common mode current is sufficiently suppressed by the balun.

Once the common mode noise current is reduced to on the order of 10 dB below desired band noise current, local QRN sources except those actually radiated into the dipole itself do not significantly impact S/N ratio at the receiver. We will then have a receive antenna, the one we thought we were using, operating as well as it can. In my experience, once this is done, most excess noise is eliminated at a typical QTH. In my case, eliminating these current sources improved my ability to hear weak signals by more than a thousand times – more than 30 dB or 5 true S units!

Wait a Minute, My Coax has a Shield. How can this other noise get in?

Perhaps I’ve lost you in the above detail and you are still not convinced that you could have a common mode noise current problem. After all, isn’t the shield one of the reasons for using coaxial cable? To respond to this question, I’ve detailed one more figure Balun Currents.pdf.

With this drawing I’m trying to show how even perfect coax can deliver noise to a receiver if there is not good balance. The signals we want are shown as Ia (blue) and Ib (green). They are equal in magnitude with one going into the coax line and the other coming out. Also shown is the unwanted common mode noise current. It is due to the noise currents on the support or on the outside of the coax as already described. Unless the balun can provide a high enough impedance to prevent it, which is another way of saying “make Ia and Ib magnitudes exactly equal so that there is no Icm” then unwanted noise will proceed down the coax to the receiver where it may raise the noise floor above the level of band noise associated with Ia and Ib. The shield on the coax does not provide isolation on the inside from current on its outside.

For these reasons, simply installing an antenna on a support and connecting to it with high quality coax is no guarantee that its receive performance will not be terrible!

Perhaps in another post I’ll describe some approaches that have worked for me to produce excellent baluns. These include flux-coupled and transmission line transformers, both untuned and tuned that have been used with a very small, relatively high efficiency antenna providing good transmit performance at the same time it dramatically reduces 20m noise floor.

Glenn Elmore n6gn

Contact information is at the bottom of the N6GN pages
July 2017

PDF icon CM_noise1.pdf39.21 KB
PDF icon Coax Closeup.pdf22.02 KB
PDF icon Balun Currents.pdf22.55 KB