Your browser is out-of-date!

Update your browser to view this website correctly. Update my browser now


Commentary: 5 kHz Bandwidth Restriction Suits AM

Pre-emphasis Response Distortion at Transmission End Can Compensate for Listeners' Poor Radio Bandwidth

Pre-emphasis Response Distortion at Transmission End Can Compensate for Listeners’ Poor Radio Bandwidth

I have long been an advocate of AM bandwidth restriction to approximately 5 kHz — long before the recent trend and well before digital IBOC and/or DRM came on the scene. The genesis for my advocacy for a restricted AM bandwidth was based upon the realization that the majority of AM radios, portables, tabletops, clock radios, automobile radios and even semi-professional receivers all had poor higher frequency (h-f) response bandwidth.

I first noticed this quantitatively in the late 1950s and early 1960s. The typical AM radio high-frequency response was about –12 dB, or worse, at about 4-5 kHz. Little has changed in the ensuing 45 years.

Although I can offer no proof, it has long been my belief that AM radios have limited high-end bandwidth because of cost and the fact that restricted bandwidth reduces the occurrence of nighttime skywave adjacent- and next-adjacent-channel interference.

Frequency response data

My advocacy of restricted AM transmitter high-end bandwidth enforced the concept of “less can be more.” That is to say, if the transmitter bandwidth on the AM bands had been restricted to 5 kHz many years ago it would have allowed the receiver manufacturers to increase receiver bandwidth while still maintaining acceptable adjacent-channel interference. The receiver cost factor would not have been solved by this action, and in fact would have reversed.

(click thumbnail)
To offer some support for my position, Fig. 1 shows frequency response data I recently performed on nine AM radios I own. The data are divided into four categories: A, the worst radio (in terms of h-f bandwidth); B, the best radio; C, the average of all nine radios; and D, the average of all radios minus the best and the worst.

The RF signal generator was homemade with a modulation bandwidth (–1dB) of 20 kHz; 640 Hz was arbitrarily chosen as the reference frequency. For radios with tone controls, the controls were set at mid-rotation. Radios with multiple bandwidths were set at the highest bandwidth. Averages were computed using RMS (power summation) techniques.

The data seen in Fig. 1 do not paint a pretty picture for analog AM. Even the “best” radio in the table, a tabletop model circa 1970, is still –9.1 dB at 5 kHz. While I did not make measurements on the two automobile radios I own, my ears tell me the AM radio response is very poor — typical for car radios for as long as I can remember — compared to the quite acceptable cassette, CD and FM modes.

The human ear is remarkable, having approximately 100 dB dynamic range from the threshold of hearing to the threshold of pain. Hence, the ear could easily “respond” to an h-f signal that is –20 dB, –30 dB or less. But that would not correspond to the typical engineering definition of “response,” which is typically –1, –3 or, in some cases, –6 dB. Nor it would it fit the standard definition of “high fidelity.”

Pre-emphasis distortion

I have no argument for those who claim there are radios out there with excellent h-f audio response. They may indeed exist. However I think I can say they are in the extreme minority, a minority so small it would not be prudent to base a transmission standard upon it. It is my belief that the data presented in Fig. 1 represent the majority of radios available to the AM listening public.

I also have little argument with those who propose a dual bandwidth system; wideband (10 kHz or higher) for daytime ground-wave propagation, and narrow-band (~5 kHz) for nighttime sky-wave propagation — except the wideband daytime bandwidth still will not gain the listener much because of the arguments and data presented in Fig. 1 — i.e., the overall system limitation is caused by the typical listener’s receiver.

There is, however, something that can be done at the transmission end to compensate for the listener’s poor radio bandwidth, and that is pre-emphasis response distortion.

Pre-emphasis has been used on the AM band for years, ever since the NRSC recommended a modified 75 uS standard in 1985-86. This pre-emphasis recommendation yields approximately +10 dB boost at 10 kHz and a +5 dB boost at 4 kHz. A different and more pronounced pre-emphasis could be used with a restricted 5 kHz bandwidth standard to achieve dramatic improvement in the AM system bandwidth — “system” being defined here as including the listening receiver.

A pre-emphasis curve peaking at approximately +9 dB at 4 kHz, followed by a “brick wall” filter with a 5 kHz cutoff could dramatically improve the overall AM “system” response while still preventing adjacent-channel nighttime skywave interference.

The Ts and Ss and other consonant sounds would return to the AM band spoken word. Music, while not high fidelity, would at least be better — noticeably better than it is now. Nighttime sky-wave interference would be made acceptable.

And, for those few high-quality radios out there, the “treble” might have to be cut a bit. They do this successfully in regions of the world where the AM band channel spacing is 9 kHz as opposed to our 10 kHz spacing and where the European Broadcasting Union and Asian Broadcasting Union establish the transmission standards.

Of course digital, whether IBOC or DRM, is now around the corner and requiring its own and new transmission standards. But as long as analog AM exists, I think the discussion above is valid.

RW welcomes other points of view