Mixing Wide and Narrow Modes
Why are wide bandwidth modes like phone separated from narrow modes like CW on most bands?
Why are modes generally arranged in frequency allocation so wider modes are more restricted in available spectrum?
These are both good questions. Looking at spectrum restrictions, we find CW and other narrow band modes are generally allowed to operate any frequency within a certain license class while phone or wider emissions are more limited in available operating frequency ranges. Only recently has phone been expanded to a wider areas of bands. That's the old established regulations. This makes us wonder, why can't we just allow narrow modes and wide modes to mix?
Some wide bandwidth mode operators, or unskilled non-technical people, have deluded themselves into thinking we can have unrestricted segmentation within bands. They often cite the modern spirit of deregulation, where common sense rules. We all understand deregulation. Like the banking and mortgage deregulation, where common sense and self control prevailed to produce a strong stable banking system. Or deregulation that brought free trade and so many jobs to our country, first consumer electronics in the USA boomed, then our steel industry took off, then general, manufacturing, and finally textiles blossomed from deregulation. Deregulating Wall Street worked very well also, as did deregulating electrical power generation and distribution in California.
Citing the spirit of deregulation, some operators propose we can self-regulate Ham bands. Oddly, these operators tend to be strong signal operators who prefer the widest possible bandwidth. These operators have convinced themselves, or at least want us to believe they are convinced, they can hear and avoid weak signals.
Operators often do not consider the technical problems caused by interleaving dissimilar bandwidth and dissimilar mode signals. There are some very basic engineering reasons why modes are separated. Let's look at how the system really works.
Receiver Sensitivity Limits
Noise is a broadband signal. The amount of noise, or noise power, reaching the detector in our receivers is directly proportional to overall bandwidth. The narrower the bandwidth, the less noise power reaches the detector system. Signal power, so long as the signal is narrower than the IF bandwidth, does not change with selectivity.
Excluding narrow band interference, broad spectrum noise almost always limits the weakest possible signal that can be heard. Almost any modern receiver, even the least expensive, is sensitivity limited by external noise picked up by the antenna system.
Unless our receiver is so poor it overloads and internally generates spurious signals, noise that limits reception is not reduced by front end filters or "antenna Q". Excluding the relatively exceptional case of receiver front end overload, a given receiving antenna system and receiver can only be "quieter" and hear weaker signals better when the receiver has better overall selectivity in the narrow IF filters or DSP filters.
Here is an example. Let's assume we have a 500 kHz wide front end filter on a 2 kHz wide IF bandwidth receiver. Let's assume this receiver is not overloading from poor design or exceptionally strong off frequency signals. Weak signal performance (signal-to-noise ratio) will be exactly the same whether antenna system bandwidth is 50 MHz or 50 kHz. This is because the receiver IF filters set the ultimate selectivity, and that selectivity determines noise power by determining the ultimate noise bandwidth.
Wider receivers have less useable sensitivity than narrow bandwidth receivers. Anyone who serviced two-way radio systems back when wideband FM was used probably remembers it was impossible to have a wideband FM set equal the sensitivity of a narrowband FM set of similar design. The wideband FM receiver typically had several dB less weak signal sensitivity than a narrowband FM receiver even when both used identical front ends. The reason for this is very simple... noise is a very wide bandwidth signal while the desired signal can be made very narrow. The wider the receiver the more it fills with the broad spectrum noise that surrounds any receiving location. As a matter of fact once a receiver is wide enough to pass the spectrum necessary to convey information, any further increase in bandwidth reduces system range. This is why Collins and other companies (including telephone companies) settled on 2 to 2.5 kHz bandwidth for SSB voice transmissions. While wider bandwidths sound better; they actually decrease communications system range or efficiency.
We can observe the increase in noise power with an HF transceiver having variable selectivity with constant receiver gain. If we tune to a signal-free area of the band with a wide filter and note noise level, we will hear noise drop in proportion to any reduction made in IF filter bandwidth. This is because, as we narrow the IF bandwidths by half, noise power drops by half. Go from a 6kHz filter to a 3kHz filter with no other changes and noise power drops by 3dB.
Assume we have a signal carrying information in a 3kHz bandwidth. We listen to that signal with a 9kHz bandwidth IF filter. We have three times the noise power necessary, or 4.8dB more noise than necessary. If we reduce bandwidth to 3kHz while making no other changes, we improve S/N by almost 5dB. We can hear a signal that is almost 5dB weaker. This effect applies to any mode, and is true up to the point where the receiver's ultimate bandwidth is less than the width of the useful transmitted information.
This happens because noise is almost always of significantly wider bandwidth than the information bandwidth of the incoming signal. The chart below shows the change in noise power as receiver bandwidth is changed:
Let's assume you and I have the same receiver, we are using the same IF filter bandwidth, and we have the same noise level. Let's also assume you and I have similar antennas and propagation.
We start out both using 8kHz filters listening to a 100Hz wide CW signal from a 100 watt transmitter. If you switch to a 500 Hz filter, the S/N ratio of your receiver improves 12 dB. The bandwidth change of 500/8000 = .0625 .0625 * log10 = 12.04dB
With your reduced receiver bandwidth the station we are listening to can decrease power from 100 watts to 6.25 watts. He will have the same initial signal-to-noise ratio to you, but I will now receive with 12dB less S/N!
If you further decrease bandwidth to 100Hz, you now gain another 7dB of S/N over my ability, for a total difference of 19dB. You can detect a 1.25 watt signal almost as well as I can hear a 100 watt CW signal!
This illustrates one major problem that occurs when dissimilar bandwidth systems are mixed, even when they are the same modulation type. This is a valid technical argument in favor of rules or bandplans that sort signals by bandwidth. Sorting by bandwidth prevents a wide bandwidth "alligator" from not hearing what is going on and accidentally QRM'ing narrow band signals. It isn't a matter of how careful the operator is. The problem is rooted in the fact that wide bandwidth receivers cannot hear as well as narrow bandwidth receivers. Keep this important effect in mind when we discuss transmitter bandwidth and splatter!
By the way, a "white paper" referenced by ESSB operators claims readability improves with increased bandwidth. That's absolutely correct, except the paper assumes zero noise floor or infinite signal-to-noise ratio! Indeed if we have unlimited signal-to-noise ratio a bandwidth increase beyond normal communications channel bandwidth will often increase our ability to recognize the difference between certain sounds, but this is only true when the noise floor is far below the level of the weakest part of the speech spectrum. Under normal communications conditions any increase in bass and treble actually decreases range, and it can make copy nearly impossible when stronger noise or weaker signal levels are present.
Claiming we can work more DX or enhance communications range by using wider audio bandwidth might be good salesmanship, but it is far from true. Collins and other communications system leaders weren't headed by dummies, and a white paper written by a company that sells broad bandwidth announcement systems for offices that says wide bandwidth improves the system is not exactly and unbiased or reliable technical resource.
How wide are typical medium quality SSB transmitters? Here is the actual spectrum bandwidth of an old IC-751A on SSB at 100 watts with normal speech. The transmitter filter was measured at 2.7kHz at -6dB points.
This display is 1.2kHz per division.
99% of the transmitted power is within 2.04 kHz of bandwidth. 2.4kHz below and 1.0kHz above peak emission frequency, the signal is -43dB. -43dB bandwidth is less than 3.4kHz.
The occupied bandwidth is 2.04kHz.
A non-HiFi Viking Valiant with modifications to remove modulator and PA modulation linearity flaws with 90% modulation has an occupied BW of about 8.6kHz under the same recorded voice stimuli. The -43dB bandwidth is more than 18kHz! Many people blame the bandwidth on audio system distortion, but a major cause of needless power level at wide bandwidths is non-linearity in the modulated stage.
In order to have low distortion the power output of a plate modulated RF power amplifier has to follow the square of anode voltage change. Tetrodes don't ever do that very well because the screen voltage has such a large influence on power. The solution is to modulate one of the grids and the anode at the same time. If we carefully pick the ratio of audio voltage applied to the anode and grid (generally the screen) we can greatly improve modulation linearity. Unfortunately it is almost never close to perfect. As such, most tetrodes make very poor modulated stages so far as distortion products are concerned.
This creates a problem. We not only have the original audio bandwidth to contend with, we have distortion products extending out for a considerable distance. While amateur AM transmitters are the worse offenders, SSB transmitters are not without problems.
Mixing Wide and Narrow modes
If we move the IC-751A transmitter's signal (or any other reasonably clean SSB rig) close enough to a normal bandwidth SSB receiver's frequency without overloading anything, we will start to hear the other signal on modulation peaks. Part of the receiver passband is overlapping the extended parts of the transmitter's passband. The occasional peaks are heard as short "spits" of sharp noise.
If we have two normal SSB BW systems, each system has equal footing. The noise floor, sites being equal, is identical. As we move the operating frequencies closer and closer together, and if the systems have about the same IM performance and noise floor, each operator will hear the other at about the same time. It's very easy for either of the operators to know when they are bothering the other operator when the systems have equal radiated power, bandwidth, and local noise.
When we mix a SSB system (or any wider system) with a CW system (or any significantly narrower bandwidth) system, a much different situation develops.
Let's look at a CW signal (it could be PSK or any other narrow mode) and a normal SSB signal (it could be AM or ESSB).
The CW transmitter has the majority of its energy within a few hundred hertz, the exact bandwidth dependent on the rise and fall times and slope of the rise and fall. In a properly engineered transmitter very little energy extends outside of a few hundred Hz total bandwidth. By the way, it is important to note this bandwidth is NOT set by the speed of the CW. It is set by the shape of the rise and fall of the carrier and other transmitter characteristics.
The SSB receiver, on lower sideband, can be set so the CW signal is either just a few hertz above the dial frequency or 3kHz below the dial frequency of the SSB transmitter. In most cases very little if anything will be heard from the CW transmitter. This is not the case for the narrow mode operator.
The CW receiver, having somewhere around 10dB less noise power due to its narrower bandwidth, has 10dB more ultimate sensitivity than the SSB bandwidth receiver. This means the narrow bandwidth receiver can detect or be bothered by 10dB weaker SSB distortion than a regular SSB receiver could hear. If a good clean CW receiver is 5kHz away from the same IC751A, it will be bothered by frequent spits from the SSB transmitter. The same spits would be masked by the increased noise floor of a wider receiver.
I have a very quiet rural location. On 160 meters I sometimes have difficulty copying weak DX as far as 8kHz from strong SSB stations. Often this is not because the SSB operator is "overdriving" his transmitter, but rather because transmitters (even when not HiFi or overdriven) have fairly wide bandwidth low-level emissions. Some of the very wide spurious emissions are only reduced 50 to 60dB from peak power. Of course the SSB station generally has no idea his transmitter is bothering anyone, since the stations he bothers are well outside his receiver's passband. Also, for the same site background noise, he has a higher noise floor because of his receiver's increased bandwidth. He can hear as weak a signal as a narrower receiving system can detect.
I ran a daytime test on several S-9 75-meter AM amateur stations, and in my quiet location I could detect spits that would bother weak CW signals up to 15kHz up and down from their carriers. This problem wasn't nearly the same severity when using SSB bandwidth because the wider receiver bandwidth increased effective noise floor by 10dB. The 10dB higher noise floor of the wider receiver filter masked the splatter with broadband noise.
Transmitter bandwidth is a second compelling reason to not mix voice and narrow signals of any type. The bandwidth issue stacks on top of the noise floor disparity caused by receiver bandwidth.
Here's a good test. At a time when few people are operating, use your own receiver to learn the difference between detecting cross-modes. Find a very weak CW signal with no one else nearby, and switch to AM without changing bandwidth. With all things equal, you will find it very difficult to tell a weak but readable CW signal is present. At best you only hear a little change in background noise level as the transmitter is keyed. If you try to detect a very weak carrier, you turn on the local oscillator to get a beat note. It has always been known things work this way!
PSK and FSK are even worse, there are no lengthy breaks in carrier level.
The detection disparity adds another level of problems to mixing modes. Some modes just do not detect other modes real well.
This effect adds a third level of disparity between modes!
Figure this one out. You are on CW and a SSB operator comes on the same frequency and politely asks, "Is this frequency in use".
What does the CW operator hear? Nothing but undecipherable noise! Even if through some miracle the CW op was using a SSB filter, the SSB station would have to be nearly zero beat with the CW BFO frequency to be understood. The CW operator would have to stop his QSO, switch to SSB, and hope the SSB station could hear him say "yes, the frequency is in use".
What is the common reply when you are not Johnny-on-the-Spot with an "it's busy" reply? The other guy says "I asked first, you should have said something sooner."
What if you are on PSK, and a CW op comes on? You might not even know how to copy or send CW! How would you alert the CW station, or how would a CW station alert the PSK station? How would a SSB station talk to a digital op, or a digital op talk to a SSB station?
What about FM? Under normal conditions AM, CW, or SSB receivers cannot decode FM.
What about the operator who cannot copy CW? How does he recognize an ongoing CW contact?
We now have found a fourth problem layered on the other three. People cannot readily communicate between different modes.
The Operator Problem
The final problem is not so much the mentally deranged operator, they are few and far between, but rather the selfish unbending operator who only follows the letter of the law. Without enforceable bandwidth guidelines, a selfish operator could park a transmitter that would wipe out weak narrow signals anywhere he wanted. He could claim he didn't hear you, and you could not prove differently. 160 meter weak signal operators are all familiar with a group of W5's who parked on 1824 SSB just "because they could". The JA stations only have 1810-1825, many other DX stations could not go below 1810 or above 1825, and this was a frequent spot for DXpeditions to operate CW. They clearly and intentionally violated the bandplan...because they felt they legally could.
It took years and ultimately required FCC intervention to get them to move to another frequency in an otherwise clear band!
They aren't the only group. WA0RCR runs wide AM broadcasts on 1860 KHz, wiping out SSB up and down 5kHz. In the 1970's W8LZM, W8ETO, W8LAD, and a few others formed a "Window Shade Net" with the sole intention of QRM'ing weak DX. One of the original "Window Shade Net" members is still alive and actively QRM'ing DX today! He has a history of over 40 years of willful QRM and violating bandplans.
One person can easily wipe out the pleasure of hundreds when we depend on bandplans. They only need a desire to cause QRM, without enforceable bandplans they have the means.
A workable non-regulated band requires all operators to be willing to sacrifice and compromise. It requires everyone to respect bandplans, and to use good judgment.
We often hear "160 works without segmentation", yet when a petition was filed to segment the band several hundred people filed in favor of segmentation. Only a handful filed against forced segmentation, and those people were all wide bandwidth mode operators. "160 works" for those who work wider modes with strong signals, and it works exceptionally well for those who like to cause intentional QRM.
There is logical technical evidence to support this statement. Mixed modes and mixed bandwidths are both clearly problematic.
We need to encourage regulations that separate or segment areas of bands by signal bandwidth.