At first, I was just looking around on the web for some simple Rules of Thumb that compare the weak-signal performance of commonly used analog and digital modulation types. I was mostly focused on FT8 and FT4 but I also wanted to compare SSB and CW. I failed to find a simple comparison of these modes but I did find a number of good articles that compared some but not all of them. This article is my attempt to aggregate the available information into something easy to understand.
Disclaimers
I decided to leverage the work of others and to not try deriving everything from basic principles. I am telling myself that I am perfectly capable of doing the analysis but that I would never find the time to actually complete it. (Yeah, that’s my story, and I’m sticking to it.) Where the articles disagree, I tried to identify which one(s) had the most convincing analysis or rationale and used those values.
My goal is to compare common modulation types primarily in terms of weak-signal performance. This means focusing on how well a signal can be detected with low signal-to-noise ratio (SNR). I have ignored other factors, such as signal fading, frequency drift, multipath distortion, etc. Also ignored are factors such as the information rate provided by the modulation type and the required signal bandwidth. This is focused on having the ability to pass just enough info to make the contact.
Literature Survey
Searching the internet provided me with a number of good articles that have examined this topic, listed below in the References section. My approach is to compare the results of these articles and aggregate them into a concise summary. These minimum SNR values are listed in the table shown below, along with my aggregated conclusions in the righthand column.
Most of these articles presented SNR data in terms of a 2500 Hz bandwidth, with the goal of providing an easy comparison between modulation types. SSB is the widest signal discussed, and it roughly fits into a 2500 Hz bandwidth, which is often the IF bandwidth of the receiver being used. Some authors make this explicit by tagging this SNR as SNR2500. It is common practice in communications work to normalize the bandwidth to 1 Hz, which indicates the modulation’s bandwidth efficiency. However, we’ll stick with SNR2500.
I started with the article by PA3FWM [Ref 1], which provides a look at many of the modes I was interested in comparing. Unfortunately, this article does not include FT4 and FT8. N6MW [Ref 2] has a good treatment of FT4 and FT8 as well as minimum SNR values for SSB and CW. These lined up well with the PA3FWM values, so that was a good sign. N6MW referenced the foundational article about FT4 and FT8, published in QEX, written by the FT4 and FT8 developers [Ref 3]. The KB9II article [Ref 4] focuses on VHF weak-signal performance and provides minimum SNRs for SSB, RTTY, CW, and PSK31. He introduces the concepts of SNR (average) and SNR (peak). I used the SNR (average) numbers in the table. The KF6HI [Ref 5] article provided another set of SNR values that lined up pretty well. Finally, I came across a presentation by K0LB and KK4SNO [Ref 6] that includes a slide summarizing SNR performance. Because it is slideware, it does not include much about the sources of their numbers, but it seems useful to include them in the table.
The authors have somewhat different approaches to determining their SNR2500 numbers, mostly related to the assumptions used. You may want to read through these papers to gain a better understanding of the fine points. Overall, there is good alignment on results, with a few exceptions.
SSB
The single-sideband SNR2500 values are a mix of 10 dB and 6 dB. Frankly, I think 10 dB is a bit high for “minimum SNR” because I’ve spent quite a bit of time making weak-signal VHF/UHF contacts with the signal right at the noise level. I’ve squeezed out radio contacts with SNR much less than 10 dB. I looked at the rationale supplied in the articles for this value and it is mostly just assumed. So I went with my own experience and chose something smaller, 6 dB, aligning with KB9II and KF6HI. Even this number might be a bit conservative.
RTTY
I found only four values for RTTY, and they vary quite a bit. After studying the articles, I judged KB9II to have the best justification, so I went with -9 dB. I suspect that the actual decode performance may vary depending on the type and quality of the detector.
CW
The SNR2500 numbers for CW varied significantly, over a range of 10 dB. One way to estimate CW performance is to use the bandwidth of the receiver and compare it to 2500 Hz. Using a typical CW filter bandwidth of 200 Hz, SNR2500 = 10 log (200/2500) = -11.0 dB. However, it is well-known that the human ear/brain combination provides additional signal processing. The classic article by W2RS [Ref 7] covers this topic quite well. Using actual on-the-air tests, the article explains that the skill of the operator can introduce a variation of 3 to 6 dB. Another interesting note is that if the operator knows in advance the type of information they are expecting (such as the callsign of the other station), it provides a 3-dB advantage.
We can and probably will debate the SNR2500 value for CW until the cows come home, but I decided to adopt -12 dB in the right-hand column. This is probably conservative for a highly skilled operator.
FT8, FT4
For FT8 and FT4, I used the N6MW values, which come directly from the FT4 and FT8 paper [Ref 3]. I rounded off to the nearest decibel to be consistent with the rest of the column.
JT65
The JT65 values are quite consistent. An article by K1JT [Ref 8] says JT65 SNR is “roughly -28 to -24 dB in 2500 Hz,” so I put -24 dB in the righthand column.
WSPR
WSPR is a popular beacon mode and the king of weak-signal reception. Signal reports are collected worldwide and shared via WSPRnet.org. WSPR performance will vary depending on the specific settings used on the software and we have some variation in the table. The K1JT & W1BW article [Ref 9] says, “The WSPR protocol is effective at signal-to-noise ratios as low as –28 dB in a 2500 Hz bandwidth, some 10 to 15 dB below the threshold of audibility.” So I used -28 dB in the aggregated column.
Conclusions
The rightmost column in the table provides a reasonable comparison of the listed modulation types. I don’t claim that the values are perfect, but they should be helpful in understanding the performance of these modes. These data show that SSB is the least sensitive mode, followed by RTTY and PSK31. As mentioned earlier, the CW number is open to debate but it performs better than RTTY and PSK31. This brings us to FT4 and FT8, which are commonly used WSJT protocols with reasonable throughput. (FT4 and FT8 using 7.5 and 15-second transmit/receive intervals.) JT65 operates at lower SNR, but it is really in a different category, It is designed for Earth-Moon-Earth contacts, using one-minute intervals. WSPR is also unique as a beaconing system and not designed for two-way radio contacts, but it does have the best SNR performance on the list.
When using this data, keep in mind that most of these modes degrade slowly so there may not be a sharp cutoff at an exact signal level. The values are Rules of Thumb, accurate to within a few dB.
Thanks to Jim/K5ND and Bob/WØBV for reviewing this article and providing feedback.
73 Bob K0NR
References
-
Signal/noise ratio of digital amateur modes – Pieter-Tjerk de Boer, PA3FWM
-
FT8 Modulation and Decoding – A Dive into SNR interpretation N6MW
-
The FT4 and FT8 Communication Protocols – Steve Franke, K9AN- Bill Somerville, G4WJS – Joe Taylor, K1JT
-
A Comparison of Common Digital Modes for Weak Signal VHF Communications – John Matz, KB9II
-
Signal to Noise Ratio, definition and application to Radio Communications – KF6HI
-
Digital Modes in Amateur Radio – Larry, K0LB and Scott, KK4SNO
-
The Weak-Signal Capability of the Human Ear – Ray Soifer, W2RS
-
EME with JT65 – Joe Taylor, K1JT
-
WSPRing Around the World – Joe Taylor, K1JT, and Bruce Walker, W1BW
Earn the FCC Technician license or upgrade to the FCC General license.
I’ve been reading a number of reports from the areas affected by the two major hurricanes (Helene and Milton). The North Carolina experience is particularly interesting because people have experienced the loss of communication and electrical service for several weeks. I can imagine this same thing happening in other parts of the country, including my area. As one example, read the 
