snr-and-noise-eng

IV3PRK Pierluigi “Luis” Mansutti

160 Meters: DXing on the Edge

S.N.R. - Signal to Noise Ratio… and more on noise.

When making any sort of comparison, it should be noted that for DX contacts on 160 meters, two factors are involved: - signal strength and noise - or, in combination, - a S/N ratio. Thus, there is a minimum S/N ratio for a successful contact: if contacts are not made, then either the noise was too great for the signal strength, or the signal strength became too weak for the prevailing level of noise. The signal strength part of S/N ratio has been fully threated on this page of my site:

You find there all the loss factors playing a role in a multi-hop propagation path: free-space distance loss, ionospheric absorption losses and ground reflection losses. With the formulas provided by Carl, K9LA, and K. Davies in his basic book “Ionospheric Radio Propagation”- National Bureau of Standards 1965, I compiled a simple excel spreadsheet to perform all these calculations. Unfortunately I can't upload an Excel file on this website, so this is a screenshot example:

Honestly, this loss calculator must be intended mainly as a teaching tool to explain the K. Davies propagation principles, which are still valid but date back sixty years. These principles describe the classic propagation mode through ionospheric and ground reflections, common to all HF bands. However, in recent years, it’s been found that on 160 meters, the best long-distance contacts often happen through a special ducting mode that sometimes forms between the E and F layers.

Nowadays, advanced programs on powerful computers handle these calculations. The 3D ray-tracing engine in Proplab-Pro version 3 calculates ionospheric electron densities and three-dimensional layer gradients directly, instead of relying on pre-built profiles. With its high-resolution global topographical database (with a resolution of less than one square km.), it can also precisely compute ground reflections by determining ground tilts, like reflections off mountain ranges. A three-dimensional ray tracing is very complex and time consuming. With the ray advancement, a lot of data are updated: distance in km., height and angle of the ray, ionospheric absorption in dB and signal strength at every ground reflection point.

This is the most important data for path analysis and is given in dBµV per meter – a field strength voltage, which is convenient to convert in power dBm (below 1 milliwatt). The conversion formula is given by Carl, K9LA: dBm = dBµV/m - 77.21 - 20*LOG(MHz) + Rcv. Ant. Gain (for simplicity, we keep the receiver ant. gain to zero, simulating an isotropic antenna also at each reflection point), and I added these relations - calculated for 1.8 MHz – to the following table:

The table shows the voltage and power relationships of the RF signals found at the receiver input and can be used to convert between the many different units used to express signal strength. You can check them also with the spreadsheet Receiver_Levels.xls that comes on the CD with ON4UN’s “Low-Band DXing” fifth edition.

The table can be useful to follow the path analysis with 3D ray tracing fully exploring ducting and spotlight propagation conditions occured during the 1997 S21XX and P29VXX DXpeditions (Part 1, 2 and 3).

Part 4 analyses the 160 m. FT8 operations by S21DX in their 2024 DXpedition and, thanks to the printed reports, it's very useful to understand spotlight propagation and S/N ratios given in that mode. Click on these PDF files:

NOISE

Now that we’ve covered all the aspects of signal strength entering our 50-ohm receiver system—the first part of the S/N ratio—let’s take a closer look at the second element: noise. Under the broad category of “noise” fall two key issues: receiver sensitivity (internal noise) and external noise.

Much of what follows comes from an article in WorldRadio, August-September 2005, by Carl Luetzelschwab, K9LA, available on his website:

as well as from Chapter 3 of ON4UN’s “Low-Band DXing,” 5th edition, 2010, a must for every dedicated Topbander.

Receiver sensitivity and noise floor.

Sensitivity is the ability of a receiver to detect weak signals. The most important concept related to sensitivity is the concept of signal-to-noise ratio (SNR or S/N). Reception will be good or bad depending on the strength of the signal with relation to “the noise.”

It is generally accepted that comfortable SSB reception requires a 10 dB signal-to-noise ratio. CW reception requires a lower S/N, and any moderately experienced CW operator can rather easily deal with a 0 dB S/N.

A really good operator can dig CW signals out of the noise at –10 dB S/N in a 500-Hz bandwidth, mainly because his built-in “brain filter” narrows the noise bandwidth much further. This shows the inherent advantage of CW over SSB for weak-signal communications. Now, with FT8 and new technologies, the computer-operator can deal with -24 dB S/N, and everything is easier, but that’s another story…

One measure of the sensitivity of a receiver is called its minimum discernible signal (MDS), or noise floor. It’s the level of a RF signal that increases the no-signal audio output by 3dB. In other words, the RF signal level generates the same audio output power as the internally generated receiver noise.

Our receivers are not perfect - they have the thermal noise - internally generated in different ways. It’s the noise we hear when there is no antenna, but a dummy load connected to the receiver. The internal receiver noise is produced by the movement of electrons in any substance (such as resistors, transistors and FETs that are part of the receiver circuit) that has a temperature above absolute 0 Kelvin (0 K or –273° Celsius). Absolute zero is where all electrons have stopped moving. Above 0 K, electrons move in a random fashion, colliding with relatively immobile ions that make up the bulk of the material. Their random pulses produce what is called thermal-agitation noise, or simply thermal noise.

Let’s ignore the Boltzmann equation and calculations, which express the noise power, and go directly to an example: at an ambient temperature of 27°C (~300 K), in a 50-ohm system with a receiver bandwidth of 3 kHz, the thermal noise power is: p = 1.38 × 10^–23 × 300 × 3000 = 1.24 × 10^–17 W.

This is equivalent to -10 LOG (1.24 × 10^–17) = –169 dBW or –139 dBm (139 dB below 1 milliwatt) and is equivalent to 32 dB below 1 µV or –32 dB µV. This is the theoretical maximum sensitivity of the receiver under given bandwidth (3.000 Hz) and temperature as highlighted in the previous table.

More sensitivity can be achieved by reducing the bandwidth or by cooling your equipment. If in this example the bandwidth is reduced to 300 Hz, the noise floor level becomes –149 dBm, simply because there is 10 times less noise power in a window that is 10 times narrower. We have gained 10 dB by reducing the noise by 10 dB! This explains the big advantage of CW over SSB when signals are weak.

Anyway, on 160 m. CW DXing I always use 150 Hz filter, thus gaining further 3 dB, so calculated:

-149 dBm -10 LOG (300/150) = -149 -10 LOG (2) = -149 - 3dB = -152 dBm.

…and what's about FT8?

Nowadays a lot of activity is carried on digital modes. FT8 is made up of 8 tones, only one of which is on at any time and is separated from the next by 6.25 Hz. Together they take up 50 Hz, but the equivalent noise bandwidth of an FT8 detected tone is only 6.25 Hz and this is to use in the calculations.

Thus, moving from my CW 150 Hz filter to FT8 mode, I could gain 14 dB so calculated:

-152 dBm -10 LOG (150/6.25) = -152 -10 LOG (24) = -152 -14 dB = -166 dBm.

Be aware that SNR for FT8 in WSJT-X is computed based on 2.5 kHz passband, as the transceiver works in USB mode. That is why you see numbers like -20 dB. Factually, the SNR of the 6.25 Hz wide detector is a factor of 2500/6.25 higher, i.e., 26 dB higher. (Tnx info from Jim Wolf, KR9U, on Topband Reflector, August 2019).

Now, a question arises: which is the MDS of my receiver?

The ARRL measures MDS in their product reviews. For example, in the product review of my Icom IC-7610, in October 2018 QST, the noise floor (MDS) on 1.9 MHz in a 500Hz bandwidth is -132 dBm.

(Just fo comparizon, these were the similar MDS tests of my other previous rigs: Yaesu FT-1000MP -123.2 dBm; Ten-Tec Orion 565 -122.8 dBm; Icom IC-756 Pro II -131.1 dBm; Kenwood TS-870 -124 dBm).

What does this mean? It says a signal level of -132 dBm increases the no-signal audio output by 3dB when using the 500Hz filter. Does the MDS change with different filter bandwidths? Yes, it does, by 10 times the log of the ratio of the bandwidths (we use 10 times the log since we’re dealing with noise powers).

A 3KHz filter for SSB lets in more noise, so one would expect the MDS in SSB to be worse by 10 times the log of 3KHz/500Hz = 7.8 dB. Thus, the MDS would be -124.2 dBm with a 3KHz IF filter. In other words, with the wider SSB filter, a signal must be 8 dB stronger in order to be copied. On the other side, if you switch to CW using a 150Hz filter, you will limit the noise by 10 times the log of 500Hz/150Hz = -5.2 dB, lowering the MDS to -137 dBm. Thus, you will be able to pick-up weaker signals, up to 5 dB, that means 13 dB better than on SSB.

An interesting question to ask is “How does my Icom IC-7610’s MDS compare to the lowest theoretical noise power?”

The lowest theoretical noise power comes out to be, at room temperature (25°C = 298°K) in a 1Hz bandwidth, -174 dBm.

If an Icom IC-7610 had a 1Hz filter, the MDS would be better than the -132 dBm value by 10 times the log of 500Hz/1Hz = 27 dB (it’s better because the 1Hz filter lets in less noise). This works out to an MDS of -159 dBm in a 1Hz bandwidth. Note that it’s 15 dB away from being “perfect”, which is the -174dBm theoretical limit and this 15 dB difference is the Noise Figure of Icom IC-7610.

You can check all the process with a spreadsheet calculator (RX_Noise_Factor_and MDS_calculator.xls) available on the CD that comes with ON4UN’s “Low-Band DXing” fifth edition. Here is a screenshot showing the results of the example in the text. ===>>

But now, after having learnt everything about internal noise, noise figure and noise floor (or minimum discernible signal), another question arises: Is the MDS a very important receiver parameter on the low bands? No, it’s not! because usually the level of the input noise to the receiver - atmospheric and manmade noise - is relatively high, and generally much higher than the internal noise level of the receiver. The real acid test, as far as receiver sensitivity is concerned, is as follows: If you hear a very noticeable noise increase when you connect an antenna to the receiver, your receiver is sensitive enough! You should check sensitivity at the quietest time with the narrowest selectivity you use on every antenna you use. Any more sensitivity will just make it more difficult for the receiver to handle very strong signals causing intermodulation problems.

External noise.

So far, we have discussed a receiver and its internal noise. Now let’s hook it up to an antenna to see what external noise does to our ability to hear. Since external noise has a great impact on propagation, it has been studied extensively. One excellent reference on noise is Recommendation ITU-R P.372-7 (the old CCIR Report 322), which is appropriately titled Radio Noise. There are three sources of external noise that can impact our HF operations: man-made noise, galactic noise, and atmospheric noise due to lightning discharges. Let’s look at man-made and galactic noise first. The ITU document includes a plot of man-made noise and galactic noise versus frequency, and this is reproduced in Figure 1. All noise powers are monthly median values and were measured with short vertical monopole antennas. And as indicated in the vertical axis legend, they are in a 1Hz bandwidth.

There are several important pieces of information with respect to HF operation in Figure 1.

- First, the environment you live in determines how man-made noise will impact your QTH. Ideally, you would like to be in the quiet rural (D) environment.

- Second, as you go lower in frequency, the noise increases. So, if you are a low band aficionado, noise is critical.

- Third, unless you live in a quiet rural environment and prefer the higher HF bands, galactic noise (the E curve) is not an issue (galactic noise generally doesn’t go below 10 MHz as it doesn’t get through the ionosphere).

- And fourth, if you have a Ten-Tec OMNI VI (the rig of K9LA) with its extrapolated MDS of -162 dBm in a 1 Hz bandwidth (the dashed F curve, which assumes the MDS is constant for all the ham bands - a pretty fair assumption), man-made noise limits your ability to hear.

Translating to my IC-7610 transceiver, the MDS would be -159 dBm (red dashed line), just 3 dB high, not a great difference. Here we understand why also a better MDS (i.e. lowering the dashed line) won’t help, as on 1.8 MHz all kind of external noise is always much stronger than the internal one - better known as the noise floor.

Now let’s address atmospheric noise due to lightning discharges - better known as QRN.

What this represents is the constant drizzle of noise propagating into your QTH from lightning discharges worldwide (it is estimated that there are two thousand thunderstorms occurring worldwide at any given moment). The ITU document has worldwide maps of monthly median atmospheric noise in 4-hour time periods for the four seasons - for a total of 24 maps. Each map gives the noise in dB above -174 dBm (the lowest theoretical noise power) at 1MHz, along with two other plots that allow you to calculate the noise at other frequencies and to show how the noise varies statistically. As it’s going to be too complicated, let’s just summarize K9LA’s comments:

- In the evening hours of winter months, the noise versus frequency is roughly halfway between the rural C curve and the quiet rural D curve in Figure 1 up to about 10MHz.

- As would be expected, the atmospheric noise propagating into my QTH is greatest during the summer months (lots of thunderstorms).

- And since it’s mostly a low frequency phenomenon, it’s greatest when my QTH is in full darkness (atmospheric noise propagates just like our signals).

Now we have information about our receiver sensitivity and estimates of the external noise at our QTH.

In my small village, unfortunately due to some overcome activities, man-made noise has grown in the last years and I can’t consider it as “quiet rural” anymore… just half a way, as indicated in Figure 2.

The noise data are the same as those in Fig. 1 but converted from 1 Hz bandwidth to 500 Hz with the usual relation (10 times the log of 1Hz/500Hz) = 27 dB of more noise, worsens to -93 dBm.

Sharpening the IF filter to 150Hz for CW DXing, we get a noise reduction of 5.2 dB, to -98 dBm, while on FT8 (6.25 Hz) we would gain 13.8 dB more, reaching -112 dBm, which is still about 20 dB above the minimum discernible signal, confirming that on 160 m. the problem is external noise and not MDS.

Summary and final comments by Tom W8JI.

Carl, K9LA, concluded his article with this statement: «there are a couple of issues that I didn’t address here, and the key could result in “your mileage may vary”. One issue is that you may have one troublesome noise source that dominates your incoming noise: a near-by power line (been there, done that), a neighbour’s electric blanket for their cat (also been there, done that), etc. Until it gets fixed, the data in Figure 1 may be useless».

Let’s wrap up this discussion on noise with the words of Tom Rauch, W8JI, from his website

—one of the best sources for low-band information—quoted exactly as they appear also in ON4UN’s “Low-Band DXing” 5th edition:

«The noise that limits our ability to hear a weak signal on the lower bands is almost always an accumulation of many signal sources. Below 18 MHz, the noise we hear on our receivers (even at the quietest sites) comes from terrestrial sources. Receiver noise is generally a mixture of local ground wave and ionosphere propagated noise sources, although some of us suffer with dominant noise sources located very close to our antennas.

«In urban locations, noise arrives from multiple random sources through direct and ground wave propagation from local sources. One or more sources can actually be the induction-field zone of our antennas (in most cases the induction field dominates at distances less than 1⁄2 λ). Urban locations are the least desirable locations because typical external noise average 16 dB higher than suburban locations. There is often no evidence of winter night noise increase on 160 meters, since ionosphere-propagated noises are swamped out by the combined noise power of multiple local noise sources. Much of the noise sources are utility distribution lines, because of the large amount of hardware required to serve multiple users. Other noise sources are switching power supplies, arcing signs, and other unintentional manmade noise transmitters.

«Suburban locations average about 16 dB quieter than urban locations and are about 20 dB noisier than rural locations. Noise generally is directional, arriving mostly from areas of densest population or the most noise-offensive power lines. Utility high-voltage transmission lines are often problematic at distances greater than a mile, and occasionally distribution lines can be problems. The recent influx of computers and switching power supplies has added a new dimension to suburban noise. There is often a small increase in nighttime winter noise at exceptionally quiet suburban locations. This increase occurs when propagated terrestrial noise equals or exceeds local noise sources.

«Rural locations, especially those miles from any population center, offer the quietest environment for low-band receiving. Daytime 160-meter noise levels are typically around 35-50 dB quieter than urban, more than 20 dB quieter than suburban locations. Nighttime brings a dramatic increase in low-band noise, as noise propagates in via the ionosphere from multiple distant sources.

«Primary local noise sources are electric fences, switching power supplies, and utility lines. Their typical daytime noise levels, measured on a 200-foot omnidirectional vertical, are around -113 dBm with a 350-Hz bandwidth (in Tom’s very quiet location).

Noise power increases about 5 to 15 dB at night, when the band “opens”. As in the case of suburban systems, directional antennas reduce noise power. Nighttime is the “equalizer” reducing the advantage of location as distant noises increase with improved propagation».

In quiet areas the total noise will be higher during the night, since additional noise arrives by atmospheric propagation, adding to the local manmade noise found during the daytime. In urban residential areas the local manmade noise is so high that you never can hear such propagated noise. This means that the receiver sensitivity is essentially a moot point. Almost any receiver is sensitive enough in such a hostile environment!

Noise levels can, and do, vary tremendously from one location to another. We have many bad 160-meter locations, and only a few very good ones. Indeed, the difference in manmade noise levels can be up to 50 dB!

Assuming that under most circumstances all noise is evenly distributed in all directions, a well-engineered special receiving antenna will receive up to 12-15 dB less noise than an omnidirectional antenna.

To benefit from this directivity advantage, the receiver will require 12-15 dB better noise figure (more sensitivity).

In addition, such directive noise-reducing antennas are usually low output antennas (such as Beverages with typically –10 dB gain) and often used with long “lossy” feed lines (say, 1 dB loss). Add all of this together, and you need 12 + 10 + 1 = 23 dB more sensitivity (lower noise figure), so a lot of your surplus sensitivity for 160 meters has gone away.

Sometimes we may need a preamp (typically with 10-15 dB gain). Much lower is the output of small loop antennas - even -55 dB with my Waller Flag - where I need a low-noise KD9SV preamplifier with 50 dB variable gain. Don’t forget that a preamplifier amplifies the atmospheric noise as well as the signal, so the ratio will remain the same. Such preamps cannot amplify only the signal (and not the noise); hence they should only be used to compensate for loss, and not to receive a stronger signal.