Postdetection Filters for HD Radio Signals

Nearly all stereo decoders use a 38 kHz square wave to demodulate the L−R signal, which occupies 23 to 53 kHz in the stereo composite. An unintended consequence is that the waveform's fifth harmonic demodulates power near 190 kHz. HD Radio digital sidebands, which appear at 129 to 198 kHz for hybrid service mode MP1, can cause an annoying background noise when the fifth harmonic demodulates them. Extended hybrid signals, whose spectrum begins near 116 kHz for service mode MP3, cause additional noise due to the third harmonic at 114 kHz. A lowpass filter between the FM detector and stereo decoder can eliminate this HD Radio self-noise.

To maintain audio frequency response and stereo channel separation, conventional postdetection filters have flat amplitude and group delay over the 53 kHz stereo composite passband. But high stopband attenuation requires a complex filter. This Windows program takes a different approach. It models a simple all-pole filter along with a stereo decoder. You specify minimum stereo separation and maximum frequency response error. An optimizer then seeks the filter with maximum stopband attenuation. Instead of using a standard filter response, it directly optimizes the pole locations. It constrains the spectral flatness of the audio output, but not that of the filter itself.

The program uses a Differential Evolution optimizer. The resulting optimal filter is down 3 dB somewhere between 35 and 45 kHz. Roll-off this low attenuates the L−R signal, particularly the upper sideband. But like the vestigial sideband system of NTSC television, the sum of the demodulated sidebands is nearly constant. This approach requires four poles for contemporary digital sideband levels.

Here's a filter example. Sep is stereo separation in dB. L|R, L+R, and L−R are the demodulated levels in dB for single-channel, in-phase, and antiphase signals. Fc is the −3 dB corner frequency. Sen is stereo separation sensitivity, the largest degradation of worst separation for ±1% changes in C1–C4, the active filter capacitor values. Lev is the decoded audio level change with no compensation. MP1 is the RMS filter attenuation over 175 to 198.4 kHz where MP1 digital sidebands decode to 15 kHz or less. MP3 adds 115.55 to 129 kHz for MP3 mode. MP11 adds sidebands beginning at 101.74 kHz for MP11 mode, which very few stations currently use.

To calculate the stereo figures, the example cascades the filter with the composite response (amplitude and phase) of a Sony ST-S555ES tuner measured before its own postdetection filter. The IF uses two SFE10.7MP3-A 250 kHz Murata filters. The Sony response, down 1.5 dB at 53 kHz, should be typical of tuners that use similar filters. To improve accuracy, measure the composite response of your own tuner. Phase data is optional.

The optional buffer provides high input impedance and restores the original decoded audio level. Add a 0.1 µF ceramic across the IC power pins. Connect the FM detector output with DC offset to the filter input. Connect the filter output to the stereo decoder input coupling capacitor. An NE5532 op-amp will work for input above 2 V (typical) or 3 V (worst case). To handle detector output closer to ground, use an op-amp with wider input voltage range and gain-bandwidth product of at least 10 MHz, such as a quad LM6134 (DIP) or OPA1679 (SOIC). In dual-supply systems, connect the op-amp negative power pin to the negative supply.

This is the filter response without the buffer gain.

Select and parallel parts to come close to the target values. Then enter measured values in Check.exe to check the filter response as implemented. Adjust the component values until you're satisfied with the response.

The stereo decoder pilot lag capacitor compensates for an internal phase lead. Increase its value to compensate for group delay in the postdetection filter. The capacitor is C1 in the circuit on the left and C2 on the right. Sometimes it is on the other side of the series capacitor (almost always 0.047 µF), and some tuners omit the pilot lag capacitor altogether. You may find it on these pins:

 Pins  Stereo Decoder
 1  5  CXA1064 LA3450
 2 12  LM1800 LM4500A TCA4500A µA758
 3 11  HA1156W MC1310 SN76115
 3 12  AN7470 HA11223W KB4437 PA1001A µPC1161C µPC1173C µPC1235C
 3 13  AN363 BA1332 HA1196 HA12016 KA2261 KA2265 LA3350 LA3410
 3 18  LA3400 LA3401
 4     AN7472S
18     µPC1223C
19 20  LA3390
20     LA3380 LA3381

Adjust the stereo separation trimpot and pilot lag capacitance for maximum separation at 1 kHz. You may need to parallel a resistor with one that limits the separation trimpot range.

Three-Pole Filter

Eight years after initial authorization at −20 dBc, the FCC permitted an increase in digital sideband level to −14 dBc for most stations and −10 dBc with special permission. The four-pole filter should fully quiet any HD Radio signal, with the possible exception of those at −10 dBc using MP3 extended hybrid mode. At −20 dBc a simpler three-pole filter renders HD Radio self-noise inaudible for signals using MP1 hybrid mode. A trace of noise may remain for MP3 mode. I installed three-pole filters in my tuners before the digital power increase.

Stereo Sensitivity Enhancement

A postdetection filter can reduce noise on any stereo signal, not just one with HD Radio sidebands. Detected FM noise increases 6 dB per octave, the same rate that squarewave harmonic amplitudes decrease. Thus each 38 kHz harmonic can potentially demodulate as much noise as that in the L−R region. The IF filter will attenuate this harmonic noise. A postdetection filter can eliminate it.

For the wide IF filter (two 250 kHz Murata MXs), 50 dB stereo quieting sensitivity for a Yamaha T-1020 was 42.4 dBf. Adding a three-pole postdetection filter increased sensitivity 2.4 dB to 40.0 dBf. For the narrow IF filter (two 110 kHz Murata MHYs cascaded with the 250s), sensitivity increased 1.5 dB from 40.6 dBf to 39.1 dBf.

Three-Pole Filter Examples

This is the detected spectrum to 200 kHz for an HD Radio signal in a Yamaha T-1020 (wide IF filter).

This is the spectrum after installing the postdetection filter.

This shows the filter installed on a perfboard in the tuner. The T-1020 uses a noise-detection bandpass filter near 125 kHz to automatically select IF bandwidth and stereo/mono mode. After installing the postdetection filter, the tuner still thought most clean signals were noisy. I had to route the postdetection filter output to the noise filter and boost its gain somewhat to restore normal operation. This is typical of the complications you may encounter when adding a postdetection filter.

Here a postdetection filter is installed in a Sony ST-S444ESX. This filter uses 2.7kΩ resistors. I selected three that measured within a few ohms of 2811Ω and then used this value in the filter optimizer to determine the capacitor values. Adding a 750 pF pilot lag capacitor to the CXA1064 stereo decoder increased 1 kHz stereo separation from the high-40s to the mid-60s in dB.

This postdetection filter is installed in a Carver TX-11b tuner. I added a 1300 pF pilot lag capacitor.

This shows a postdetection filter built directly on a Technics ST-9030 PCB.

Passive Filter

The circuit uses a buffer, passive lowpass filter, and active delay equalizer. Use 1% resistors and capacitors (22 µF 20%). Separate the 5% inductors and orient them at right angles to minimize coupling. Place a 0.1 µF ceramic between the op-amp power pins. Connect the FM detector output with DC offset to the filter input. Connect the filter output to the stereo decoder input coupling capacitor.

The op-amp should have a gain-bandwidth product of at least 10 MHz. With rail-to-rail input/output it will handle any detector output. Suitable dual op-amps are the LM6132 (DIP) and OPA1688 (SOIC). In dual-supply systems, connect the op-amp negative power pin to the negative supply. Use the specified inductors. The circuit model accounts for their Q and self-resonant frequency.

2.2 mH  Fastron 07P-222J-50
2.7 mH  Fastron 09P-272J-50
Op-amp  LM6132 OPA1688

RMS filter attenuation is 63.9 dB over 175 to 198.4 kHz where MP1 digital sidebands decode to 15 kHz or less. It is 55.1 dB for MP3 mode, which includes sidebands from 115.55 to 129 kHz. MP11 mode, currently rare, has additional sidebands starting at 101.74 kHz. MP11 attenuation is 40.3 dB. The filter is not designed for this mode.

kHz  .02 .1  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15   SFE10.7MP3-A (250 kHz)
dB    55 56 86 55 50 48 46 45 45 45 45 46 47 49 51 54 55   composite alone
dB    45 51 54 52 49 47 46 46 46 46 47 48 49 51 52 51 48   with filter

These calculated stereo separation figures are for two Murata SFE10.7MP3-A 250 kHz ceramic filters in a Sony ST-S555ES tuner. I measured composite amplitude and phase prior to the tuner's own postdetection filter. The first line is for the composite alone. The model set the L+R/L−R weighting coefficient for maximum separation at 1 kHz. The next line cascades the composite and calculated filter responses with no separation readjustment.

kHz  .02 .1  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15   SFE10.7MM-A (230 kHz)
dB    67 66 51 45 42 40 39 37 36 35 34 33 32 31 30 29 28   composite alone
dB    48 55 49 45 42 40 38 37 36 35 34 33 32 31 30 29 28   with filter

These figures are for two Murata SFE10.7MM-A 230 kHz ceramics in an Onkyo T-4150, also measured before the tuner's own postdetection filter.

Changing inductance ±5% decreases MP3 attenuation as much as 2.9 dB or increases it as much as 1.1 dB. Worst Sony stereo separation degrades at most 3 dB. Halving Q for both inductors decreases attenuation 0.1 dB. Halving self-resonant frequency decreases it 1.4 dB. Neither change affects worst Sony separation. Worst Onkyo separation degrades at most 1 dB for all tests.

Replacing the 820 pF with 680 pF in parallel with 110 pF increases MP3 attenuation 1.6 dB. Changing inductance ±5% decreases MP3 attenuation as much as 2.1 dB or increases it as much as 0.4 dB.

Filter amplitude and group delay over the stereo composite passband.

Filter Comparison

The active filter requires no special components. You may be able to build it with parts you have on hand. It does require component selection and realignment of stereo separation. Best separation requires adjustment of pilot lag capacitance. Highest accuracy may require measurement of tuner composite response.

The passive filter requires specific inductors and close-tolerance parts, but no component selection, tuner adjustment, or measurement. Compared with the active filter, attenuation is 4.0 dB greater for MP1 mode and 6.1 dB greater for MP3. It is 4.4 dB less for MP11, but hardly any of these signals exist.

Service Mode Survey

MP1  65%
MP3  35%

49 Southern California HD Radio stations, February 2024.


February 28, 202488–108 MHz