HD Radio Diversity Delay Field Observations: The Need For Automatic Alignment. Alan Jurison. Senior Operations Engineer iheartmedia

Size: px

Start display at page:

Download "HD Radio Diversity Delay Field Observations: The Need For Automatic Alignment. Alan Jurison. Senior Operations Engineer iheartmedia"

Eunice Anderson
6 years ago
Views:

1 HD Radio Diversity Delay Field Observations: The Need For Automatic Alignment Alan Jurison Senior Operations Engineer iheartmedia

2 Background The hybrid digital broadcasting solution deployed in the United States, FM and AM IBOC, more commonly known as HD Radio from DTS, Inc, allows simultaneous transmission of broadcast radio signals in both the analog and digital domain. A key component is that receivers can immediately acquire the analog signal and then transition (or blend ) into the digital signal after it has acquired and buffered. The transition process is called blending.

3 Background In order to have this blend be seamless and transparent to listeners, broadcasters employ delays on analog AM and FM signals to time align it to the digital. This delay is typically called diversity delay because the HD Radio transmission system uses a variety of techniques and buffering to add redundancy or diversity to the system for signal robustness.

4 Background This robustness and signal processing creates significant delays compared to the near-instantaneous analog transmissions, typically anywhere from 6 10 seconds depending on hardware, software, equipment settings, and data links in use. A properly configured AM or FM station broadcasting in HD will have a device in the analog air chain providing this delay.

5 Background In the 1st and 2nd generation of HD Radio signal generation hardware, the diversity delay in the HD Radio signal path was in the same device producing the HD Radio signal. Over time, engineers demanded flexibility in configuration, reduction of hardware costs and reducing points of failure (the first two generations of hardware weren t particularly reliable).

6 The Rise of the Exgine Platform The creators of HD Radio responded to industry input regarding transmission system architecture with a flexible transmission design called Exgine, which was implemented in 3rd generation HD Radio signal generation hardware and recently introduced 4th generation hardware. These systems rely less on computer-based designs and move most of the encoding into DSP (3rd generation) or FPGA (4th generation) designs. While modernizing the encoding hardware to meet flexibility demands, the encoding was broken up into two pieces. What was a single device became two, the Exporter and the Exciter. These components communicate over common Ethernet using a proprietary protocol called the Exgine platform. The connection between the two is considered the Exporter to Exciter (E2X) link.

7 The Rise of the Exgine Platform Exgine-based 3rd and 4th generation platforms are the most common deployment of HD Radio transmission systems deployed by radio stations. When many FM stations were implementing HD Radio in the timeframe, 3rd generation hardware was the hardware available on the market at that time. Many FM stations on the original 1st and 2nd generation platforms that deployed in have upgraded to Exgine based systems due to equipment age, obsolescence and overall improved reliability on the new hardware platform.

8 The Rise of the Exgine Platform As a station-level engineer during that time, ever since I started working with the Exgine system (now in its eleventh year in production hardware), I was challenged to keep the analog diversity delay perfectly aligned with the digital. Consulting with my colleagues, it was apparent that I was not alone. The lazy engineer

9 Measuring the Problem Manual alignment procedure was to put on a set of headphones and put a receiver in split mode with the analog on the left and the digital on the right, then get them to match just by listening. As you might imagine, it would take a golden ear to get it perfect. Now, equipped with the precision devices available today, I have determined I personally am not very good at setting this delay manually. Or, if I get close, I wasted half my day trying to get it right. Who has time for that?

Measuring the Problem Commercially-available signal monitoring products on the market during the initial rollout period (2005-2008)

10 Measuring the Problem Commercially-available signal monitoring products on the market during the initial rollout period ( ) that would measure diversity delay. But, the measurements being taken were single point in time values, just like the manual process.

11 Measuring the Problem The thinking at the time was the delay was constant, stable, and you just needed to set the compensating delay on the analog channel once and not need to make any further measurements or corrections. That process may have worked for setting up the single box 1st and 2nd generation systems, since we now know those platforms did not drift. But on the split 3 rd and 4 th generation Exgine platform, it can. On some platforms, the drift is more than you could ever imagine.

12 Time Alignment Specs NRSC-5 Document SY_SSS_1026s The absolute accuracy of the analog diversity delay as defined in the transmission signal shall be within ±68 microseconds (μs). This is equivalent to ± 3 audio samples at a sampling rate of 44.1 khz. The permitted range is 3 to +3 samples, or a range of 6 samples with the center of 0 (0 ± 3).

13 Time Alignment Specs That is an extreme amount of precision required to get within specification. One sample is 22.7 µsec. 3 samples means within 68 µsec. Many broadcast monitoring products show the measurement in seconds and samples. If you ve been measuring your diversity delay in seconds from any of these early products and been saying things like the station is off seconds close enough, you have been doing it wrong.

14 Time Alignment Specs At four digits beyond the decimal point, resolution of 100 µsec. Any reading that is not essentially is out of spec. If the station is off by 3 samples, that equates to 68 µsec, rounded to the nearest 100 µsec, you would get 100 µsec or seconds. Four samples is 91 µsec and would round to You could be in spec or out of spec at Not enough resolution. To achieve the accuracy needed, focus on samples, not time.

15 Time Alignment Specs

16 Simulation model and image courtesy DTS/HD Radio Time Alignment Specs

17 Simulation model and images courtesy DTS/HD Radio Time Alignment Specs

18 Simulation model and images courtesy DTS/HD Radio Time Alignment Specs

19 Simulation model and images courtesy DTS/HD Radio Time Alignment Specs

20 Blends Can Happen Frequently The blend doesn t just happen once or twice for some listeners. Don t think of it as driving out of the market on the highway and you lose HD lock, blend to analog. Think if you live and work and do your errands in that area, on the edge of coverage, the radio could be going in and out of HD frequently. This is where blending problems are most apparent.

21 Blends Can Happen Frequently

22 Blends Can Happen Frequently

23 Blends Can Happen Frequently

24 Automotive Companies Care Diversity delay blending issues are the number one HD Radiorelated complaint from auto manufacturers. General Motors temporarily removed HD Radio in some vehicle models to help fine tune their implementation to address consumer feedback about HD Radio blending. Their customers (and radio listeners) have been complaining to them about this for a long time. The customer takes the car back to the dealer and thinks there is a problem with the radio.

25 Precision Monitoring Exporter / Exciter co-located at TX site, simple reboot of Exciter shown. DaySequerra M4DDC Belar FMHD-1 Inovonics JUSTIN 808 ~425 sample drift when the exciter was rebooted with their graphing interfaces

26 Configuration and Delay Impacts Exporter / Exciter co-located at TX site, isolated E2X network, properly clocked, integrated FM+HD single processor as good as it gets? 24 hour view, steady at 0 samples

27 Configuration and Delay Impacts Exporter / Exciter co-located at TX site, isolated E2X network, integrated FM+HD single processor unless there is a clocking issue.

28 Configuration and Delay Impacts Exporter / Exciter co-located at TX site, isolated E2X network, properly clocked, integrated FM+HD single processor as good as it gets. 24 hour view, steady at 0 samples Omnia 6 Exi HD FM+HD Integrated Shown

Configuration and Delay Impacts Exporter / Exciter co-located at TX site, isolated E2X network, properly clocked, separate FM, HD processors, same model,

29 Configuration and Delay Impacts Exporter / Exciter co-located at TX site, isolated E2X network, properly clocked, separate FM, HD processors, same model, similar presets. 24 hour wideband view, steady at 0 samples, but notice the perceived drift to +/-1 sample per second Omnia 6 FM Omnia 6 HD Similar firmware and presets

30 Configuration and Delay Impacts Exporter / Exciter co-located at TX site, isolated E2X network, properly clocked, separate FM HD processors, same model, dissimilar presets. ~7 hour wideband view, Average steady at 0 samples, but notice the perceived drift of 150 samples per second. Omnia 6 FM Omnia 6 HD Dissimilar presets shown

Configuration and Delay Impacts Exporter / Exciter co-located at TX site, isolated E2X network, properly clocked, separate FM and HD processors, dissimilar models.

31 Configuration and Delay Impacts Exporter / Exciter co-located at TX site, isolated E2X network, properly clocked, separate FM and HD processors, dissimilar models. 10 min hour wideband view, Average steady at 0 samples, but notice the perceived drift of 300 samples per second. Omnia 11 FM Omnia 6 HD Completely different architectures

32 Configuration and Delay Impacts Exporter / Exciter co-located at TX site, isolated E2X network, properly clocked, integrated FM+HD single processor as good as it gets. 24 hour view, steady at 0 samples

33 Configuration and Delay Impacts What happens when we move the Exporter to the studio? Exporter GPS Synced. Properly isolated Intraplex DS-64 bandwidth of 512 kbit/sec. Same audio processor for FM+HD. No GPS disciplined 10MHz reference at the Exciter/transmitter site.

34 Configuration and Delay Impacts GPS Referenced Exporter at Studio, Exciter with no GPS reference. 4 hour view, significant drift

35 Configuration and Delay Impacts GPS Referenced Exporter at Studio, Exciter add GPS reference. 24 hour view, significant drift until 10:20am. Delay reentered to 0 manually 11:45 Significant improvement

36 Configuration and Delay Impacts GPS Referenced Exporter at Studio, Exciter add GPS reference. 30 min view Much more stable But out of spec

37 Configuration and Delay Impacts GPS Referenced Exporter at Studio, Exciter add GPS reference. ~10 hour view Much more stable But out of spec

38 Configuration and Delay Impacts Configuration and location of the equipment can have a major impact on diversity delay. We are in the infancy of learning about this topic. So far, co-locating the Exporter and Exciter seems to be the most stable approach. This was anecdotal before, but we now have proof of this with continuous real-time delay monitoring, logging, graphing. More research is needed.

39 Configuration and Delay Impacts Exporter at transmitter site certainly helps. Still needs to be GPS referenced for long-term stability. Exporter to Exciter (E2X) Ethernet network still needs to be isolated. Single, integrated FM+HD audio processor best practice. Dual audio processors of the same model, same firmware, similar presets can be managed. Clocking issues still matter. Check clocking configuration and, if applicable cabling. Not immune to Exporter, Exciter reboots causing delay jumps. Make sure they are on a UPS and the UPS is working correctly!

40 Configuration and Delay Impacts Exporter at the studio, is it bad news? Significant research is needed with these precision monitors. Typically means separate audio processors in competitive situations. Adding GPS appears to be a requirement, not optional. iheartmedia starting an internal deep analysis of these configurations to learn, see if we can tame them. (Intraplex, Moseley 950MHz QSLAN, etc). Early research showing mixed results. NAB Radio Technology Committee, NRSC doing research as well. Stay tuned?

41 Configuration and Delay Impacts iheartmedia Testing Matrix

42 The Need For Automatic Alignment The industry can t wait any longer. Consumers (listeners) complain about blends. HD automotive penetration is rapidly increasing. Automotive companies are not happy. Sit and wait for a year or two for more papers and presentations to tell you it s bad and you need to do something?

43 The Need For Automatic Alignment I m telling you it s bad. This problem is beyond the scope of any existing individual product, device, software build in the typical airchain today. It s not going to fix itself. It s an impossible task to do keep in top of manually. You can t even characterize the problem without an updated monitor.

44 AES Phase Reversal on Blend Simulation model and images courtesy DTS/HD Radio

45 Separate Processors Dissimilar Processing AES WordClock Ʈc Simulation model and images courtesy DTS/HD Radio

46 Separate Processors Simulation model and images courtesy DTS/HD Radio

47 The Need For Automatic Alignment Please read the white paper 2016 NAB BEC Proceedings Phase and audio level alignment is important too. Very detailed about the research thus far. What we know today is different than what we knew a year ago. What we know next year will be different than what we know today. The industry is working on a best practices document. At the end of the day, broadcasters need to take action.

48 Automatic Alignment Products Belar FM-HD1, DaySequerra M4.2 TimeLock Measures, sends corrections to a variety of Omnia, Orban, Wheatstone audio processors, GatesAir, Nautel exporters in the airchain. BE XPi 10esp Exporter DaySequerra M4DDC, Inovonics Justin 808 Single box inserted in the analog or digital airchain. 25-Seven Precision Delay Worldcast/Audemat Golden Eagle

49 Acknowledgements A special thanks to many at: DTS/HD Radio Belar DaySequerra GatesAir iheartmedia Inovonics Nautel

50 It can be bad out there. Questions?

51 Thanks Thank you for attending today s presentation.

A BETTER LISTENER EXPERIENCE: HD RADIO TIME AND LEVEL ALIGNMENT

A BETTER LISTENER EXPERIENCE: HD RADIO TIME AND LEVEL ALIGNMENT NAB Broadcast Engineering Conference - April 17, 2016 1 HD RADIO TECHNOLOGY AND YOUR AUDIENCE More of your listeners are experiencing your