Monthly Archives: January 2012

Jeff Keith

When does ‘Cranked Up!’ become ‘Over the Top’?

In part four, I touched on the importance of using good quality source material and how a little extra attention in that area can turn so-so sound into awesome sound. A few wrote and questioned my inference that codec-compressed audio was bad, but if your station has the opportunity to use uncompressed material, then why not? It’s another step towards your station sounding better than your competition.

This issue is about FM loudness wars. To get the ball rolling I’d like to refer to some research I did a few years ago on “loudness” from the point of view of listeners radios and how the average radio behaves when tuned to a station that’s modulating above 100%.

Rewind to 1994. Our $$$$ modulation monitor sounded good if we overmodulated, but I was wondering about the radios in the hands of listeners … what happened inside their radios?  I felt, and with a high degree of confidence, that consumer radios had nowhere near the demodulator performance of our station’s modulation monitor. But how bad were they? I wanted to find out.

I’m fortunate to have a nicely equipped lab at home so making some measurements on a collection of different consumer FM radios gave me a great excuse to stay inside on a cold and snowy weekend. The staff of the station graciously loaned me around two dozen radios for my project, and in return I promised that I would try not to break them.

My research had one goal; discover how run-of-the-mill radios behaved as FM modulation levels increased above 100%. I wasn’t interested in measuring the radios’ baseline distortion performance – I intuitively knew those numbers would be all over the map anyway and therefore meaningless, at least for this project.

What line of thinking led to this project? Tradeoffs were probably made in the design of consumer radios to minimize their cost and maximize the signal to noise performance with the components used. Using the regulations on FM modulation as a guide, designers would naturally believe that if the radio’s design had sufficient IF bandwidth and audio headroom to accommodate FM signals modulated at the legal limit, they would be home free.

The upside of that assumption was that radios could deliver decent audio performance at legal modulation levels, and in fact most do.

The downside (and this is where I was going with my thinking): designers would likely leave little if any headroom for accommodating FM modulation above 100% simply because they didn’t need to. To me, this hinted that higher modulation levels had the potential to generate significant and audible distortion in a listener’s receiver.

The data I collected during my experiment represents the comparison of two signals; the modulation percentage for the very linear FM signal generator I used, and the resulting audio level out of each receiver’s demodulator. I was not surprised that most radios had significant nonlinearity above 100% modulation.

In all, 27 radios were measured. To keep this article at a reasonable length I’ve selected five of the most revealing plots to share. The first one, at left, is for the receiver that best tolerated modulation above 100%; it’s my 1960’s vintage, tube type McIntosh MX-110 which just started to become nonlinear at 150% modulation!

The remaining four plots show what each radio’s audio output level did as the modulation levels went up. The associated graphs are easy to interpret; the straightness of the diagonal line going from lower left to upper right shows how well the radio’s audio output follows increasing modulation levels. A perfect receiver would plot as a perfectly straight line. Unfortunately such a receiver is quite expensive to build and could not be sold at ‘consumer’ prices. It’s also unfortunate that that’s what we’d like to believe our station’s listeners have.

The vertical line in the center of each plot represents 100% modulation. The left-hand scale is ‘relative’ because each radio had a different audio output level and since standardizing the radio’s actual output levels wasn’t relevant to my measurements, I simply chose to not calibrate that axis.

At left are four plots, beginning with Radio #1. Radio #1 was actually pretty good and its detector and audio sections didn’t become nonlinear until 135% FM modulation. I suppose this might be expected in a ‘higher-end’ consumer model that has sold for around $400 since the 1990’s.

Next is the plot for Radio #2. This radio was a ‘better’ model that formerly sold for around $75 at Radio Shack and it’s admired by broadcast professionals for its ‘nice’ AM sound.  Note however that on FM it becomes very nonlinear at around 118% modulation.

Next is the plot for Radio #3, a ‘mid-grade’ and popular table model. This radio was fairly linear up to 110% modulation but abruptly became very nonlinear above 115%.

The last plot is for Radio #4, a common consumer grade portable radio. Its output level exhibited a very strange behavior at 110% modulation where it suddenly stopped increasing, and once above 110% the output actually started decreasing as the modulation increased further.

The important detail in each plot is at what modulation level the plot diverges from being a straight line because when it bends, it shows that the radio is no longer following the station’s modulation and is then generating its own distortion. The sharper the bend is at a particular modulation level, the more distorted the radio will be at that modulation level and above.

These plots ignore any influence from the station’s audio processing because it’s not relevant except for one thing; when a station is overmodulating, the denser the audio processing is, the more distorted the audio will be for listeners with those imperfect radios because the audio spends more time up in the nonlinear part of the radio’s level vs. distortion curve.

Do the plots provide other useful clues? Yes, and there’s one that I’m loath to admit; our country’s FCC somehow managed to get it right in 1984 when they decided to raise the FM modulation limit, but only to 110% (see FCC BC Docket #82-536 to amend parts 2 and 73 of the FCC rules…). As well as the interference concerns, the new ruling took into consideration how the radios of the day, and the anticipated radios of the future, would behave at 110% modulation.

I haven’t measured a group of newer FM radios yet. But the trend towards lower quality consumer AM radios suggests that the quality of today’s FM radios is probably inferior to those in 1984 when the 110% modulation limit went into effect. It also infers that the trend should be for the radios of today to be inferior to those I measured in 1994. Now that we know that our 130%-plus FM modulation probably isn’t doing many consumer radios any favors, what can we do?

As a sanity check, I plotted the ability of each of the 27 receivers to accept higher modulation percentages and put this data into the pie chart at left. What the chart shows is the percentage of receivers tested that are able to accept a certain level of modulation without becoming distorted. There are two numbers associated with each pie chart segment; the number inside the parenthesis is the percentage of radios tested that were able to accept a certain modulation level, and that modulation level is shown outside the parenthesis. For example, 100% of the receivers tested could demodulate 105% modulation without distortion, while only 44.4% could do so at 130% modulation.

Based on the data it’s tempting to suggest that modulating above 115% should be approached with caution because we can’t control what radios listeners are using to hear our station.  With the radio’s own ‘self-distortion’ exposed it’s also tempting to suggest that the distortion heard by listeners on their radios is probably being blamed on the station, and not on the radio (with everyone in a market modulating well over 110% listener’s have no point of reference, yes?). Could this mean that some of the erosion of time spent listening has much less to do with the stations programming, and more to do with the inability of consumer radios to handle the extremely competitive modulation levels we’re seeing today? In other words, is high modulation annoying our listener’s radios which then annoy the listener to the point that they tune out? It’s food for thought…

I decided to write an article on this particular subject in order to share the observations I’d made about how some not-too-carefully-selected consumer FM radios behaved at elevated modulation levels.  The fact that we as broadcasters have no control over what radio a listener might use perhaps leads to a very pertinent question:

What benefit, other than embarrassing the competition (which is just another form of inside joke that listeners will never get) are we getting from pushing FM modulation levels through the roof? 

I love reader feedback and welcome all comments. Please shoot me an email at jakeith@wheatstone.com and tell me what your observations have been in your market!