Checking Signal-to-Noise Ratio The Correct Way

Many of you may be familiar with the HF+ Discovery, Airspy's excellent, highly sensitive, high dynamic range, software-defined radio. I acquired one about three years ago to compare with my SDRPlay RSP1a. There is no competition there, the Discovery wins hands down in virtually all categories.

SDR Sharp, known as SDR#, is Airspy's SDR software made for their product line of receivers, including the HF+ Discovery. It is an outstanding piece of tech and includes many frills like the innovative Co-Channel Canceller which can effectively cancel or null one signal amongst several others on the same channel. I use SDR# exclusively here for reception of mediumwave and shortwave.

Herein we will describe a tip on how to accurately check signal-to-noise ratio using Airspy's SDR# receiving program. More Airspy SDR# receiving tips may appear in this blog, so check back often. I have several of them in mind.

In all the talk about signal strength and signal-to-noise ratio (SNR) lately, the community is finally coming around to realizing that SNR is what makes the difference between copying a station or not. Surprisingly signal strength, usually measured as the strength of the received carrier, may have little to do with it.

Noise floors on today's receivers are high, though there are ways to mitigate this using low-noise antennas and other sundry items in our toolkit. The goal is to set up an antenna and feed system environment with a low noise floor and capture enough signal to rise above that. Don't focus on S-meter readings and brute signal strength. The oft-held directive usually heard is, "Allow the antenna to set the noise floor". Translation: We want enough antenna to capture the signal, but not too much antenna to raise the background noise floor to unacceptable levels.

Signal-to-Noise Measurements

So first, how do we measure SNR? And second, how much signal-to-noise difference do we need? We are not talking about tedious lab measurements here, but simply, "How far above the noise is this signal I'm receiving?". Casual measuring of signal-to-noise ratio while receiving on an SDR is often not done correctly, and sometimes SDR software itself does not report it correctly. It's a value that's interesting to know, and tells us a lot about the signal, our background noise level, and the recoverability of the audio we might be hearing.

To get a sensible reading we must set our bandwidth properly and position it correctly in relation to the signal's carrier. The SNR reported will be the signal power contained within the currently set bandpass compared to the average noise floor also within that bandpass. The measurement problem arises when the station's carrier is contained within the filter bandpass. As we will soon see, the carrier does not convey intelligible information, but its presence may add as much as 20 dB or more to our SNR reading!

So therein lies the mistake: Don't center the bandpass on the carrier! In fact, keep the carrier outside of the bandpass altogether.

dBm Talk, And How We Measure Signal Strength

dBm is the common scale used to quantify received signal power. The dB part, or the decibel, is a logarithmic ratio of one value to another. The 'm' part means relative to a milliwatt, or decibels below (can also be above) one milliwatt expended power into 50 ohms. Let's be clear here about dBm and signal measuring. Most SDR receiving software, and SDR# is one of them, does not have calibrated spectrum scales in dBm. The scale shown is in dBFS, or dB relative to Full Scale. Full Scale is where the SDR's analog-to-digital converter overloads. The meat of our testing here will refer to relative values in dB when comparing loops to each other. That is the important thing to note in the discussion which follows.

Let's talk about the makeup of an AM modulated signal, that which we usually receive on our SDR.

The AM Modulated Carrier

The total AM modulated signal consists of the lower sideband (LSB), the carrier, and the upper sideband (USB). They occupy bandwidth on the RF spectrum. Any time information is impressed upon a carrier it occupies bandwidth. The carrier itself is very narrow, and may only occupy 100 cycles (Hz), maybe even less, depending on how pure it is. All three are easily seen on an SDR spectrum display if we zoom in on the signal a little. Find a signal on your SDR and expand your bandwidth out to about +/-5 KHz either side of the carrier peak and you will see the sidebands fluttering up and down. The audio intelligibility is contained in the sidebands, not the carrier, and can be extracted in total from either sideband. Also note the lower and upper sidebands are mirror images of each other.

AM modulation showing sidebands

If the carrier is fully modulated, called "100% modulation", the carrier (the spike in the middle) will contain 50% of the station's power, and each sideband will contain 25% of the power. If there is no audio signal modulating the carrier, then there will be no power in the sidebands.

To properly measure the signal-to-noise ratio of the signal we're trying to hear, we need to measure how far above the noise the entirety of one of the sidebands is. Thus, we must isolate one of the sidebands in the bandpass.

Here's how, using SDR#:

1. Be sure to use "snap to center" tuning, that is, when you click on the spectrum, the tuning point is always centered in the spectrum window.

2. It's always good to be in DSB (double sideband) tuning mode.

3. Take a look at the zoomed in signal and note how wide the modulation is from the carrier peak to the general edge of the audio variations on either side.

4. With your mouse, grab one of the edges of the bandpass and drag it to reduce the total bandwidth to this edge. It may approach 10 KHz, or +/-5 KHz.

5. Note the bandpass width and roughly divide it in half. Example: If 10 KHz, use the mouse to narrow it further to 5 KHz. We want the width of one of the sidebands. Important! Stay away from encroaching splatter from adjacent channels.

6. Retune the receiver, offsetting the main tuning either positively (for an upper sideband check), or negatively (for a lower sideband check), so that the bandpass contains only the sideband. Important! Make sure the carrier peak is just outside the bandwidth stripe. Try to keep a margin of at least 100 cycles (Hz). It's imperative to not include the carrier in the SNR test. It conveys no intelligible information and greatly inflates the SNR reading to a meaningless value for our purposes.

7. Last, hover your mouse within the bandwidth stripe on the spectrum display. Let the SNR reading settle in for a few seconds. You will see several values. The SNR value will be your signal-to-noise value for the set bandpass width. The Floor value will be the noise floor.

Let's try some examples.

In the example, we see WPIE-1160 in Trumansburg, NY, a 5 KW station at 94 km distance. Our real signal-to-noise ratio is 17.1 dB with a noise floor of -116.6 dBFS. We have 17.1 dB of signal poking its head above the noise. That's plenty, but not overly strong. Our original screenshot of us hovering over WPIE's carrier with the bandpass centered would have us believing our SNR was 44.1 dB. That would be considered an exceptionally strong SNR. But it includes the power in the carrier. Remember, the carrier itself carries no audio information.

Click image for larger version.

WPIE-1160

One more example, WSKO-1260 in Syracuse, NY a 5 KW station at 121 km distance. WSKO is extremely weak, and marginally copyable just out of the noise.  Our real signal-to-noise ratio is only 8.2 dB with an extremely low noise floor of -126.4 dBFS. That is minimal and about the weakest signal which we are able to extract copyable audio. Our original screenshot of us hovering over WSKO's carrier with the bandpass centered would have us believing our SNR was 29.6 dB, under normal circumstances a very comfortable listening level.

Click image for larger version.

WSKO-1260

You will find that signal-to-noise ratios of 9-10 dB to be about minimum for extracting audio intelligibility from a signal. 15 dB becomes comfortable, 25 dB armchair copy.

Remember, centering our carrier in the bandpass is our mistake in measuring SNR. A carrier with no audio modulation or one which is weakly modulated may show an SNR of 40 dB. It's a meaningless value unless you wish to know the SNR of the carrier itself to the noise.

Try this method of checking signal-to-noise ratio. It will give a much more accurate representation of the receiving condition.

Updated Portable Greyline Map v1.9.1 2024

The Portable Greyline Map has been updated to version 1.9.1 2024.

Look for the download at the upper right of this page.

Knowing where there is daylight and darkness over the earth at any given time is somewhat of a necessity when DXing shortwave or mediumwave. There are online maps, sure. But what if you had your own personal daylight/darkness map you could save to your computer and click on any time you wish?

The Daylight/Darkness Greyline Reference Map is produced by my Radio Data MW program.

Included is a GoogleMap-based, HTML-driven map which shows the current daylight/darkness state over the earth. Being a Google-based map, it is zoomable and scrollable. You may choose map view or satellite view. If desired, a home location may set by scrolling the map and clicking "<<Set Home". This will allow you to return to the same reference point as long as the current map page remains open. Depending on your browser, latitude and longitude may be saved across map restarts.

Draw tracks, check distances and bearings to points, all displayed in the marker tooltip

You must have an internet connection to view the map.

The file download is at the upper right of this page. It's a small 19 KB. For more details on usage, be sure to read the greyline_readme.txt file contained in the download.

Portable Greyline Map v1.9.1 2024

Eliminating Noise With The 1:1 Voltage Balun

This little gadget might save your DXing career, the 1:1 isolation voltage balun. Follow along as we talk about bothersome electrical noise, weak signals, and how we might use this balun to dramatically lessen your noise problem.

The 1:1 Isolation Voltage Balun

I've been fighting electrical and digitally generated noise here for a few years now. It's been a tough go. New battles rise up often. Most recently, the next door neighbor bought a new appliance which is throwing even more hash across the AM broadcast and shortwave broadcast bands. What to do? Give up radio? I entertained that idea for awhile.

Over these years, I've tried it all - impedance matching, ferrite sleeves, choke baluns, taping up coils of coax, strategic-positioning of antennas, probably more. Moderate success in noise elimination was sometimes gained but total success has eluded me. 

Noise levels here, even with on-ground antennas, can run as high as S9 or more on a traditional signal strength meter. Welcome to the modern era, dear radio DXer. S9 is your new "atmospheric" noise floor in a manner of speaking. If your desired signal isn't a little or a lot above that level (we'll call that the signal-to-noise ratio, SNR), you're not going to copy anything intelligible from it. If below S9, forget it. You will hear nothing because the signal is buried in the noise.

Most modern receivers, and that includes SDRs, will produce meaningful, copyable intelligence from a 1 microvolt signal if the noise floor is low enough. A 1 microvolt signal will show S3 on our strength meter. That's pretty low, but common when dialing across the bands. In order to hear that signal our noise floor must be lower than that.

Let's calculate this out. The 1 microvolt signal coming into our receiver is -107 dBm on the power scale (-107 dB less than 1 milliwatt into 50 ohms). Let me illustrate how infinitesimal a signal that is. The decibel or dB scale is a logarithmic scale. Every increase (or decrease) of 10 dB multiplies (or divides) our signal's power by a factor of x10. In ratio form, our 1 microvolt signal at S3 is only 0.000398 the voltage strength of our S9 signal we talked about earlier.

Depending on the signal and the receiver, many receivers can produce intelligence from a signal if it's at least 10 dB stronger than the background noise. Thusly, to receive our 1 microvolt signal our noise floor must be at least -10 dB below, or not more than -117 dBm. With the proper receiver, hardware, antenna, and some luck, we might be able to accomplish this.

One of the tricks of lessening noise pickup is to get your antenna as far as you can from any noise source. Whether on-ground or in the air, put it out in the far corner of the yard as far away as possible. I've found results usually start to improve at about the 100 ft. distance. In the extreme out on the farm, I've run the coax feeder out to on-ground wire antennas which were 200 ft. away. Great results in noise-abatement can be had at that distance. Don't worry about long runs of coax. In the bands we are concerned with, attenuation is still borderline microscopic with RG-6 common cable coax. RG-6 is available and cheap, at 75 ohms, which makes nary a difference when receiving at MW and SW frequencies.

Seriously consider the on-ground antenna, that is, an antenna which lies on the ground. Yes, it goes against common sense, doesn't it. Old hams and SWLers pre-1980s are of a mindset to hoist as much wire into the air, as high as possible. Today, often what you get is S9++ noise on your receiver. I've experimented with many on-ground antenna types, among them the LoG (loop-on-ground), the DoG (dipole-on-ground), and what I'll call the VDoG (vee-shaped dipole-on-ground). After a lot of experimentation, I became a convert to the on-ground antenna school of thought. You must start thinking in SNR (signal-to-noise ratio), not brute S-meter readings.

My current antenna is what I call a VDoG, the vee-shaped dipole-on-ground. It's positioned as far out in the yard as I'm able to get it. It's fed with 80 ft. of RG-6 coax up to a second-floor window. The antenna itself is two sections of 22 ft. #18 stranded insulated wire fed at the middle, arranged to form an angle of about 80 degrees with the open end facing south. It shows a little gain towards the open end, about unity gain on the sides, and a bit of a null at the back, or north. From here in western New York, it's good for mediumwave DXing, southerly along the eastern seaboard all the way to Cuba.

Even with this low-noise antenna, in my eco-system this antenna hears a lot of noise due to all the electrical buzz flying around, amounting to about a steady S9+ across the AM broadcast band. The noise is essentially electrical hash coming from mine and the two adjacent houses, the power lines which transit across the back edge of the yard, and the ones feeding the properties.

The VDoG, the LoG, and the DoG are balanced antennas. If you want to lower your noise problem, stick with a balanced antenna. Avoid end-fed wires, even if layed on the ground.

Common noise-abatement thinking would be to place a choke balun at the antenna feed point and then perhaps clip a few ferrite snap cores to the coax end coming through the window, just before it connects to the receiver. I installed an MFJ-911H 4:1 current balun I had at the feed point and snapped some cores on at a strategic point. It produced little improvement, helping a little more at the higher end of the MW band than the lower.

The 1:1 Voltage Balun

A single item I accidentally stumbled upon about a year ago was what killed the noise dead in its tracks. That was the 1:1 voltage balun, which will isolate the entire feed system, including the coax shield, from the receiver. These are sometimes called galvanic baluns.

Of course, this type of balun can be home-constructed, and in the past, I've wound these myself on ferrite cores and used them with varying success. On larger cores they become an unwieldy mess, and the connecting wires are a noise magnet. Size matters. The extra exposure of a larger object to RF hash matters. Searching eBay one day I found compact, miniature versions of these from China already constructed on a very small circuit board, complete with SMA connector on each end. The entire board is only about an inch long. They are 1:1 ratio, 50 ohm impedance in/out. More importantly, they have no physical connection between input and output, and no grounding. The board uses a Chinese clone of the Mini Circuits 1:1 balun and costs about $8 apiece with a little shipping added to that. I ordered one. It worked so well I ordered two more.

The 1:1 Balun in use with the HF+ Discovery

Positioned correctly in the coax feed to the receiver, the resulting reduction in noise was miraculous. Positioning is important. Inserted right at the receiver input (SDR here) using a double-ended male union produced the best results. Looking at the schematic diagram this little balun should be bi-directional, meaning either connector should be able to be used for input or output. It didn't seem the case. Connecting it one way resulted in a little better noise reduction than when reversed.

Here are graphics of the results, the before balun and after. Graphic shots were taken with an Airspy HF+ Discovery tuned to 860 KHz at 4 AM local time, lots of skywave arriving, showing the MW band span from 500 KHz to 1000 KHz. The noise floor goes from an unmanageable average -70 to -80 dBm (~S9) to an astoundingly low -110 dBm (S2) across the band and holds close to that figure across the shortwave bands to 30 MHz too! Look particularly at the frequency range 500 - 620 KHz. The crazy electrical hash bubble there has even been completely removed.

Before application of 1:1 isolation balun

After application of 1:1 isolation balun

Most signals run 25-35 dB above the -110 dBm noise floor, peaking in the -70 to -80 dBm range, equivalent to S8-S9 strength. But S8-S9 means something here, as our noise floor is at S2! Signals are strong and well out of the noise. DXing is possible again.

I won't suggest that the 1:1 voltage balun will work for total noise abatement in all cases. It works for me in my environment, as you can see. If you have severe noise problems, consider trying a low-noise antenna and this low-cost, miniature isolation balun.

Here is the eBay link to the 1:1 isolation balun:

https://www.ebay.com/itm/166955230066

If the eBay link is dead, try searching eBay for "0.1M-550M 1:1 Isolation Transformer High Frequency Transformer SMA Connector".

I hope this article helps with your noise abatement.