L14 Lab — HD Radio#

ECE 448 · Spring 2027

HD Radio

Build a sideband power meter, identify a station’s configuration from a capture, and measure a live station on the PlutoSDR.

L14 Lab · Module 6: Waveform Applications & RF Reverse Engineering · Dr. Neil Rogers

Learning Objectives

  1. I can describe how HD Radio works and demodulate an HD Radio signal.

From Lesson to Bench

  • Pre-Lab: change one test-file setting and read the finder.

  • Bench: build a sideband meter, identify a capture’s configuration, and measure a live station.

  • Writeup: explain what you saw and what it tells you about each station.

Lesson 14 found the HD Radio sidebands in a simulated test file and measured them with L14-hd-find.grc. Today you build the sideband meter yourself from an empty canvas, use it on a second capture whose configuration you do not know, and then point it at a live station with the PlutoSDR. Last, if you have nrsc5, you can decode a recording of the station; that step is optional. The lab packet has three parts, and you turn it in once, complete, before Lesson 15. You do Part 1, the Pre-Lab, on your own before the period, record Part 2’s measurements in class, and explain them in Part 3, the Writeup.

Nothing in this lab transmits. The PlutoSDR only receives, because your meter has no transmit block, and both captures came from GNU Radio flowgraphs, not from an antenna.

Part 1: Before the Period

  • Change one setting in L14-hd-test-tx.grc, run it, and read the finder.

  • Slide the finder’s Inner subcarrier and watch the readings.

Do the Pre-Lab before you arrive. Put both flowgraphs from the Lesson 14 page in one folder. Question 1 has you open L14-hd-test-tx.grc, change the Variable upper_dbc from −23 to −18, run it once to write L14-hd-test.cu8, and then run L14-hd-find.grc and record what its three meters read; if GRC reports an error, the L5 lesson page lists the install checks.

Question 2 uses the same run: slide the finder’s Inner subcarrier from 356 down to 280, one step at a time, and record what the meters do at each step. Part 2 uses the same slider on a capture whose configuration you do not know.

Expected Numbers

Check

Expected

Your meter on the Question 1 file

matches the lesson’s finder

Hybrid sideband edges

about 129.4 and 198.4 kHz from the center

A sideband on the display

at least 10 dB above the floor

Your meter should match the finder’s readings on the same file within 0.1 dB; a larger difference means a filter setting or a center frequency is off. The lab capture comes from a different station, and you identify its configuration from its spectrum and its meter readings.

Part 2: Build the Meter

  1. Start GRC and choose File, New, QT GUI, and save the new flowgraph in the folder that holds L14-hd-test.cu8. Do not open L14-hd-find.grc; build the meter yourself.

  2. Set samp_rate to 1488375. Add Variables df = 1488375 / 4096, outer = 546, sb_center = (inner + outer) / 2 * df, sb_half = ((outer - inner) / 2 + 0.5) * df, decim = 6, and avg_len = int(0.2 * samp_rate / decim). Add a QT GUI Entry with ID fc, label Center (Hz), type real, and default value 100.3e6, a QT GUI Range with ID rx_gain, label Pluto RX gain (dB), from 0 to 70 in steps of 1, default 30, and a QT GUI Range with ID inner, label Inner subcarrier, type int, from 280 to 356 in steps of 19, default 356.

  3. Read the file: a File Source (type byte, file 'L14-hd-test.cu8' with the quotes, repeat Yes), a Throttle (byte, 2 * samp_rate), UChar To Float, Add Const (float, −127.5), Multiply Const (float, 1 / 128), Deinterleave (float, 2 streams), and Float To Complex.

  4. On the Float To Complex output, add a QT GUI Frequency Sink: complex, FFT size 4096, center frequency fc, bandwidth samp_rate, average Medium, Y axis from −110 to −10.

  5. Add two Low-pass Filter Taps variables at samp_rate: host_taps with cutoff 100e3 and transition 10e3, and sb_taps with cutoff sb_half and transition 2e3.

  6. Add three Frequency Xlating FFT Filters, each with decimation decim and sample rate samp_rate: the host with taps host_taps and center 0, the upper sideband with sb_taps and center sb_center, and the lower sideband with sb_taps and center -sb_center. Connect Float To Complex to all three.

  7. After each filter add Complex to Mag^2, a Moving Average (float, length avg_len, scale 1 / avg_len), Keep 1 in N (float, N avg_len // 10), and Log10 (n = 10).

  8. Add three Subtract blocks (float): upper minus host, lower minus host, and upper minus lower, each taking the Log10 outputs. Connect them to a QT GUI Number Sink with 3 inputs, minimum −40, maximum 0, and labels Upper and Lower in dBc and Upper minus lower in dB. Every arrow must be black.

  9. Run it on the file you wrote in Question 1, wait 5 s, and record the three readings and a screenshot for Question 3.

Part 2: The Lab Capture

  1. Stop your meter and change the File Source’s file to 'L14-lab-capture.cu8'. The File Source reads the file only when the flowgraph starts.

  2. Change the frequency sink’s FFT size to 8192 and run it. Drag a box around one edge of a sideband on the frequency sink to zoom in, read the edge’s frequency in MHz, and right-click to zoom back out. Do the same for the inner and outer edge of each sideband, for Question 4(a).

  3. Record the meter readings with Inner subcarrier at 356 for Question 4(b).

  4. If the inner edges sit closer to the host than 129.4 kHz, slide Inner subcarrier down one step at a time until the readings stop rising, and record the setting and the readings there.

  5. Read the Upper minus lower meter for Question 4(c).

  6. Find the other station in the capture, give its offset from the center, and say whether it carries HD Radio, for Question 4(d).

Part 2: A Live Station

The L14 sideband finder on the lesson's test file, labeled Simulated: L14 test file, for comparison: meters at minus 23.02 and minus 23.05 dBc and 0.02 dB upper minus lower, two flat shelves beside the host on the spectrum, and two straight bands on the waterfall

The figure shows the lesson’s simulated test file; a live station’s host is rougher, and its shelves may stand closer to the floor. We have compiled this step but not yet tested it on a live station, so expect to adjust the gain.

  1. Add a PlutoSDR Source to your meter: IIO context URI ip:192.168.2.1, LO frequency int(fc), sample rate samp_rate, RF bandwidth samp_rate, gain mode Manual, and manual gain rx_gain, with the rest at their defaults. Connect it to the frequency sink and the three filters. Then select the File Source and the six blocks after it, through Float To Complex, and press D to disable all seven; GRC will not run with both sources connected.

  2. Plug in the PlutoSDR with an antenna on its RX port; the course Plutos run in extended-frequency mode, so the FM band is in range. Run your meter and type each strong station’s frequency in hertz into Center in turn, using the station list at hdradio.com, and stop at one with two flat shelves beside it.

  3. Raise the gain until the shelves stand at least 10 dB above the floor, but lower it if the floor rises everywhere or copies of other stations appear.

  4. Record the station, its sideband edges, and both meter readings, with a screenshot, for Question 5. If its inner edges sit closer to the host than 129.4 kHz, slide Inner subcarrier down until the readings stop rising before you record them.

  5. Lower the Pluto RX gain slider in steps and watch the shelves and the readings as the floor rises toward them, for Question 5(c).

Part 2: Decoding with nrsc5 (Optional)

This part is optional, and Question 6 does not count toward the grade. nrsc5 is not part of radioconda, and it has no PlutoSDR driver, so it decodes a recording you make in GRC. Use a laptop that has a recent nrsc5 installed, your own or one I provide. These steps follow the nrsc5 project’s documentation; we have not yet tested them in this course.

  1. Open L14-hd-find.grc and switch it to the PlutoSDR as the lesson describes, then also enable its Rational Resampler and SigMF Sink. Type your station’s frequency in MHz into the finder’s Center (MHz) and set the gain you used before you run it, because the SigMF Sink records the center frequency when the flowgraph starts.

  2. Run it for about 10 s and stop it. It writes hd.sigmf-data, about 60 MB, at 744,188 samples per second, and hd.sigmf-meta.

  3. Run nrsc5 --iq-input-format cf32 -r hd.sigmf-data 0. It takes a few seconds to synchronize. Record the station name, the song title and artist, and any HD2 or HD3 program it reports, for Question 6(a). Try program 1 for HD2.

  4. Record and decode again at lower gains, a few dB at a time, until nrsc5 stops playing audio. Record the lowest gain at which it plays, and how high the shelves stand above the floor in your meter at that gain, for Question 6(b).

No Shelves?

If the meters read below −40 dBc or the shelves never appear

For a file, check that the file name in the File Source is in quotes, that the file sits in the same folder as the flowgraph, and that the File Source’s type is byte, not complex. If the shelves appear but a meter reads wrong, check the sign of each filter’s center frequency, that the two dBc Subtract blocks take the host on their second input, and that the third takes the lower sideband on its second input. For a live station, check that the PlutoSDR Source’s sample rate is samp_rate, its LO frequency is int(fc), and its gain mode is Manual, that you have centered on the station, and that the station carries HD Radio at all; many do not. Shelves of unequal height can mean the station runs one sideband weaker than the other; a missing shelf usually means a weak signal or a neighbor covering it.

Part 3: After the Lab

  • Compare the live station with the test file and say what each difference means.

  • Explain how the capture’s spectrum identified its configuration.

  • Explain why nrsc5 cannot decode the lesson’s test file.

The Writeup is three questions. The first compares the live station with the lesson’s test file, its edges, the height of its shelves, and whether its two sides match, and says what each difference tells you about the station. The second explains how the lab capture’s edges and meter readings identified its configuration, and why a station would choose it. The third lists what nrsc5 needs from a signal that the lesson’s test file does not provide.

Looking Ahead

  • Lesson 15 recovers the carrier phase and symbol timing of mPSK.

  • Lesson 20 returns to OFDM, with QAM on each subcarrier.

Keep your meter. Lesson 15 builds the carrier and timing recovery that nrsc5 performs on every HD Radio subcarrier, for a single mPSK carrier. Lesson 20 returns to OFDM, with QAM in place of QPSK on each subcarrier.