Checking bandwidth with receiver
Bandwidth rules Part 97.307
Note: Bandwidth measurement dynamic range requirements are based on typical signal-to-noise ratios I have observed over the past several years. They are not the extremes of what I have seen, but rather are typical values. Some information on my receiving system and noise floor is available in (NOISE) and (RECEIVING).
A receiver can be used to check BW if we understand what we are doing! The common mistakes are:
1.) Bandscopes and Spectrum Analyzers
Bandscopes or spectrum analyzers using wide filters are often unreliable signal bandwidth indicators. While such devices are always good for monitoring band activity, they often have too much bandwidth and/or are subject to overload. When bandwidth is filled by many strong signals, the accumulated signals demand headroom from amplifier and mixer stages. Many spectrum analyzers and band scopes (panadaptors) do not have narrow filters and low internal distortion. The old HP-141 series of analyzers, for example, require extreme care in use to prevent overload.
You can test bandwidth by viewing a pure unmodulated carrier on the display. When bandwidth of a carrier is nearly zero, the display should ideally show a perfect single spike of negligible bandwidth.
Bandscopes (and spectrum analyzers) with poor dynamic range or wide filter bandwidths are all but useless for determining bandwidth or signal defects.
2.) Noise Blankers
Noise blankers must be OFF when checking bandwidth or working close to strong signals.
In order to remove noise, noise blankers add a form of intentional distortion to signals, They do this by turning a switch or gate in the IF amplifiers off and on. An abrupt rise in peak input signal level over average signal level triggers the switch, and shuts the receiver off. The bandwidth of the noise detector is very wide, and this means a strong signal even 5-10 kHz away will activate the NB gate and distort signals.
Some receivers (like Yaesu's) do not fully remove the effects of the NB system, even when the NB is off! In some receivers you have to turn the NB off plus turn the NB gain down, the FT1000MP and FT1000MP MK V are examples of this. In others, like the FT1000D, you actually have to modify internal wiring to correct NB problems. The mechanism is explained in the links to the receiver mods.
You must select the narrowest filter possible to measure TX BW, certainly less than a 500 Hz filter with good shape factor for SSB measurements.
Receiver bandwidth and shape factor directly adds to the transmitter's bandwidth. This means a perfect brick wall 2kHz bandwidth receiver tuning across a perfect 2kHz wide transmitter makes it sound like the actual signal bandwidth is 4kHz. Theoretically it is possible to deduct the receiver bandwidth from apparent measured bandwidth to obtain real bandwidth, but this generally means you have also decreased the dynamic range of the receiver (or spectrum analyzer). In practice, deducting bandwidths often produces unreliable results.
The slope of the receiver (or analyzer) filter is also important. If the receiver response is -6dB at 4kHz and -60dB at 8kHz, you will hear stuff out 8kHz (plus transmitter bandwidth) on very strong signals if you are in a quiet location.
4.) Inferior Receivers
Some radios, in particular DSP only radios, have very poor strong signal performance. They can't be trusted to give accurate BW reports.
Look at tests here, Sherwood Engineering's tests, or ARRL tests of close spaced receiver performance. Many receiver are not all that good. Most Yaesu receivers have a built-in design problem in the noise blanker amplifier that seriously deteriorates close spaced IM performance even when the noise blanker is OFF.
Even the Sherwood engineering test is too wide for some receivers. The Sherwood test, for example, inflates performance of R4C's with the CF-600/6 filter. This happens because the 2 kHz test measures outside the filter passband (600Hz). The second mixer in the R4C is a horrible design, especially the early FET mixer. Close-spaced tests should always be done inside the bandwidth of the roofing filter, the roofing filter should be considered the narrowest reliably useable selectivity.
If the band is noisy you really can't check a signal for low-level IM, clicks, or splatter. The noise will cover up any weak signal defects. There must be at least 50dB headroom between the peak signal level being tested and your noise floor to check bandwidth on SSB, or 80dB of signal to noise headroom to check CW bandwidth if you operate near weak signal areas of the band. For general ragchewing away from weak signal areas 50-60dB headroom is generally enough.
Some bands are a special case because SSB operates near weak signal CW stations. 160 meters is one example. Bandwidth of higher power SSB transmitters operated near weak CW stations can be problematic. I often hear spits from Icom 756 and TS 2000 transmitters on SSB as far as 10kHz away on 160 meters.
Receiving noise floor is probably the single most common source of false "clean signal" reports to what actually are problem transmitters. If the noise or QRM floor is high, you won't hear spurious signals.
Wide-audio operators are particular victims to giving each other false assurances of how narrow and clean they are. They often use "opened-up" receivers that absorb more noise power from wide bandwidth (remember noise power is directly proportional to receiver bandwidth) and they often live in noisy environments. It takes a good weak signal narrow receiver in a quiet location to properly check bandwidth.
The result of the factors above is that some people will report a nasty signal "clean" when it isn't, and some will report a signal "wide" when it isn't.
Understanding how to test will correct problems, and help us use our own equipment better. Receivers make very good measurement devices if used properly.