HD Radio stations are supposed to time-align their analog and digital signals. If they don't, when analog blends to digital a listener will hear anything from an echo to a strange phasey sound to a momentary drop-out. Misaligned levels will cause a sudden change in loudness. Correct time and level alignment can make a signal much more listenable near the limits of digital coverage where a receiver may switch back and forth between modes.
A time-alignment error of just 1 millisecond causes the spectral notches shown above to appear at the midpoint of the analog/digital blend. This artifact is distinctly audible on wideband program material.
A 0.1-ms error places the delay-induced spectral notches well above the midrange. However, response is down 3 dB at 2.5 kHz and the blend still may not be entirely transparent. The iBiquity delay error spec is 0.068 ms maximum. You can experiment with a blend simulator to hear what various delay errors sound like.
Mike Bergman and John Kean discuss time-alignment issues in an NAB paper.
I decided to measure the time and level alignment of every FM HD Radio station I could receive at home in San Marcos, northern San Diego County. On my computer I recorded sound samples from a Sangean HDT-1X tuner in split-audio mode, with the analog signal in one channel and the digital in the other. Then I used a cross-correlation program to measure delay error, level error, peak unlimiting, and correlation. Delay error is positive when the digital signal lags the analog. Level error is the ratio of digital to analog RMS signal levels. Peak unlimiting is the ratio of digital to analog peak/RMS values, a measure of the relaxation of analog peak limiting. Both level error and peak unlimiting may vary with program content so the values are more suggestive than definitive. Correlation ranges from 1 for identical waveforms to 0 for entirely dissimilar signals. The smaller the magnitude, the more the transmit processing differs for analog and digital. A negative value means that the waveforms match best when one is inverted. During blend such signals will suffer a level drop when time-aligned and a low-frequency loss when not.
This is the frequency response at mid-blend for a delay error of 0.1 ms and correlation = -1.
Peak unlimiting: the signals have the same RMS level, but the peak level of the digital signal on top is 5.7 dB greater than the peak level of the analog signal on the bottom. Peak unlimiting is not a measure of blend quality, but higher values identify digital signals whose processors take advantage of the lack of preemphasis and avoid high-frequency limiting.
I used a ten-second sound sample for most of the measurements listed below. Many of the delay errors are stable, but some vary markedly from day to day. In all cases I list the latest measurement.
Delay Level Peak Corre-
Freq Call Error Error Unl lation Date Organization Market
MHz ms dB dB
88.1 KKJZ 38 0.9 1.5 0.81 04-07-09 Cal State Long Beach Los Angeles
88.7 KQSC -203 -1.6 0.5 0.93 04-07-09 University of Southern California Santa Barbara
89.3 KPCC 0.82 -0.4 2.9 0.95 03-20-09 Pasadena City College Los Angeles
89.5 KPBS 3 0.5 0.2 0.99 03-18-09 San Diego State University San Diego
89.9 KCRW 0.05 -0.8 0.2 0.71 04-07-09 Santa Monica College Los Angeles
91.1 KDSC 28 -4.2 -0.4 0.98 03-21-09 University of Southern California Ventura
91.5 KUSC 0.52 2.2 0.7 0.45 04-07-09 University of Southern California Los Angeles
92.1 KSOQ -73 0.5 0.5 0.99 03-16-09 Lincoln Financial Media San Diego County
92.3 KHHT 0.02 1.7 1.6 0.90 01-02-09 Clear Channel Los Angeles
93.1 KCBS 0.41 -0.5 2.7 -0.42 03-18-09 CBS Los Angeles
93.3 KHTS 0.25 -1.3 4.9 0.91 03-16-09 Clear Channel San Diego
93.9 KMVN 0.18 -0.5 1.9 0.97 02-19-09 Emmis Los Angeles
94.1 KMYI 0.09 0.0 1.6 0.56 03-16-09 Clear Channel San Diego
94.7 KTWV 32 -0.3 1.7 0.57 03-20-09 CBS Los Angeles
94.9 KBZT 0.07 0.6 2.4 0.49 03-16-09 Lincoln Financial Media San Diego
95.1 KFRG 0.39 0.3 0.0 -0.98 04-07-09 CBS San Bernardino
95.5 KLOS -0.05 -2.6 2.9 0.95 03-20-09 Citadel Los Angeles
95.7 KUSS 4 -2.0 2.4 0.62 03-09-09 Clear Channel San Diego
96.3 KXOL 62 -0.7 1.5 0.91 04-07-09 Spanish Broadcasting System Los Angeles
96.5 KYXY 775 1.9 3.5 0.91 04-08-09 CBS San Diego
96.7 KWIZ 30 -1.7 4.0 0.98 10-19-08 Liberman Orange County
97.1 KLSX 0.05 -0.2 2.4 0.93 03-09-09 CBS Los Angeles
97.3 KSON -35 -3.6 3.6 0.97 03-16-09 Lincoln Financial Media San Diego
97.5 KLYY -62 -1.4 0.2 -0.65 04-07-09 Entravision San Bernardino
98.1 KIFM 0.07 -0.1 0.4 0.99 03-18-09 Lincoln Financial Media San Diego
98.7 KYSR 0.00 2.0 1.8 0.93 03-18-09 Clear Channel Los Angeles
99.1 KGGI 3 -1.1 2.0 0.92 04-07-09 Clear Channel San Bernardino
99.7 XHTY 9975 0.3 2.0 -0.54 02-02-09 Uniradio Tijuana/San Diego
100.3 KSWD -50 2.1 1.7 0.95 03-18-09 Bonneville Los Angeles
101.1 KRTH 0.86 -0.5 2.0 0.94 03-18-09 CBS Los Angeles
101.5 KGB 2 -0.8 1.8 0.45 03-16-09 Clear Channel San Diego
101.9 KSCA 0.07 0.3 -0.2 0.98 03-09-09 Univision Los Angeles
102.7 KIIS -0.20 1.5 0.7 0.95 03-20-09 Clear Channel Los Angeles
102.9 KLQV 9983 -4.2 3.4 0.98 02-16-09 Univision San Diego
103.5 KOST 0.02 -0.6 3.1 0.84 10-04-08 Clear Channel Los Angeles
103.7 KSCF -227 0.1 4.4 0.92 04-08-09 CBS San Diego
103.9 KRCD 2 1.3 1.6 0.95 01-02-09 Univision Los Angeles
104.3 KBIG 0.00 1.7 1.7 0.96 03-18-09 Clear Channel Los Angeles
105.1 KKGO 5 3.8 -0.4 0.53 03-20-09 Mt. Wilson FM Broadcasters Los Angeles
105.3 KIOZ -2 0.1 4.6 0.51 03-16-09 Clear Channel San Diego
105.9 KPWR -0.39 -0.4 1.3 0.95 03-09-09 Emmis Los Angeles
106.7 KROQ 0.02 -0.9 3.9 0.95 04-07-09 CBS Los Angeles
107.1 KSSD -5 -1.0 3.7 0.98 04-07-09 Entravision San Diego County
107.1 KSSE 15 -0.5 3.1 0.92 12-05-08 Entravision Los Angeles
107.3 XHFG 9971 -6.4 5.3 -0.39 04-14-09 Uniradio Tijuana/San Diego
107.5 KLVE 0.02 0.2 2.7 0.98 03-18-09 Univision Los Angeles
107.9 KWVE 17 0.3 2.4 0.97 03-20-09 Calvary Chapel of Costa Mesa Orange County
Organization Number # Stns # Stns Median Median Mean Mean Mean Mean
of within with Delay Level Delay Level Peak Corre-
Stations iBiquity Inverted Error Error Error Error Unlim lation
Spec Waveform Dev ms Dev dB ms dB dB Magnitude
All 47 12 5 2.00 0.8 644.50 -0.3 2.1 0.83
Clear Channel 11 4 0 0.20 1.3 0.65 0.1 2.4 0.78
CBS 8 2 2 0.63 0.4 72.72 0.0 2.6 0.83
Univision 4 2 0 1.03 0.8 2496.27 -0.6 1.9 0.97
Lincoln 4 2 0 17.53 0.6 -26.97 -0.6 1.7 0.86
USC 3 0 0 28.00 2.2 -58.16 -1.2 0.3 0.79
Entravision 3 0 1 15.00 1.0 -17.33 -1.0 2.3 0.85
Emmis 2 0 0 0.28 0.4 -0.10 -0.4 1.6 0.96
Uniradio 2 0 2 9973.00 3.4 9973.00 -3.0 3.7 0.47
Median delay error deviation is the most important statistic. It is the median delay error regardless of sign. Half of the signals have a delay error magnitude smaller than the value, and half larger. Median level error deviation similarly ignores sign. Mean delay error and mean level error include the sign and may reveal systematic bias.
April 17, 2009
88–108 MHz