What's With Crappy HD Encoders?

[RomanPort]

Hello!

I want to preface this saying that I don't work in radio and don't work for any of these stations. I'm just an avid listener with equipment for viewing the signals from these stations. I might not even know what I'm talking about.

In the last few months, I've noticed more and more stations with crappy HD Radio encoders in my area. All of them seem to leak noise both into the station's analog signal as well as in adjacent channels. Using a software defined radio, I can demonstrate the actual spectrum of the signal, but I can also notice a difference just by the sound out of my analog FM radios.

Here's a screenshot of KQRS-FM during a period of little modulation. You can see how much the HD Radio sidebands leak into the analog signal of itself as well as how much it's leaking into the adjacent channels.



For comparison, here's a different station, KXXR, that doesn't suffer from the same problem. The cutoffs on the HD sidebands are very close.



KQRS also isn't the only station suffering from this problem. I can clearly see it happening on KEEY-FM, KDWB-FM, KSJN-FM, and KTCZ-FM too.



The interference caused to adjacent channels by KEEY and KDWB is enough to obliterate the HD signal on a more distant station I enjoy, WHMH, which is unfortunately squeezed between two of the noisy stations here. The HD signal on this station should be easily decodable but the edges of it are being "blocked out" by the rolloff of the local stations. WHMH's HD signal is decodable for a few seconds at a time but is usually impossible to synchronize to because of the interference.



Three of those four stations earlier also show the analog signal leaking into adjacent channels at modulation peaks too. This picture of KQRS zoomed out a few months ago really demonstrates how bad this interference is.



KQRS has since fixed this analog interference for the most part, but KEEY, KDWB, and KTCZ definitely haven't. I'm also fairly confident this isn't a result of my SDR overloading, as it's persisted for months and I actually happened to have been recording when KQRS switched on their noisy equipment for the first time. Here's a video of KQRS switching from their clean backup transmitter to their noisy primary transmitter.

Anyone else see any of these transmitters in their area? Anyone know what causes this? It's especially frustrating when the interference is causing me to be unable to decode the HD of other stations.

 

[Anton P]

It looks like they run PAR2 or Hybrid Crest Factor Reduction of OFDM carriers. KXXR apears to be running older PAR1 crest factor reduction. PAR2 will throw more spectral regrowth towards analog carrier and you are seeing that in your scans.

 

[Kelly A]

Some cheap SDR tuner also may not be a way to accurately measure the spectrum, either.

 

[Anton P]

Yes. The decibel scale may not be accurate on SDR dongle as well as low bit resolution of ADC could produce weird artifacts. But I think the biggest issue here is that he is using FFT peak detection on the spectrum analyzer and waterfall which could make HD signals appear out of mask. NRSC-5 FM IBOC measurement specifies measuring Power Spectral Density using RMS detector averaged over 100 trace sweeps during at least 30 seconds. In addition newer NRSC-5-A mask allows for slightly wider spectral regrowth shoulders below -60 dBc level outside of +/- 200khz BW. So there is a good chance those stations are in mask and OP is using incorrect measurement method.

 

[Cowboy78]

This is what is known as "spectral regrowth". Essentially the digital sidebands (actually a bunch of tiny little OFDM carriers) are being subjected to intermodulation distortion in the transmitter's power amplifier.

These distortion products are due to nonlinearities in the amplifiers. For FM analog you can get away with amplifiers that aren't perfectly linear. For digital operation, however, that isn't the case. It has improved considerably, but even linearized amplifier designs at the time could only get so good on FM broadcast frequencies, and there were huge tradeoffs in efficiency. The solution for the nonlinearities was to precorrect for them in the digital exciter before it hit the PA.

First generation HD system designs were pretty bad about this. They used "fixed" precorrection set at the factory based on the mode, TPO, and frequency specified when the transmitter was ordered. Modern HD transmitter designs have adaptive precorrection that can dynamically "learn" these distortions based on an RF sample from the PA and adapt accordingly.

The usual fix in the case of a system with active precorrection is just to re-run the routines and let it adapt again. I usually just left the exciter in that mode full-time.

The other consideration (as another poster pointed out) is peak to average power. In a digital system, the peak to average power ratio is rather high. The algorithms to reduce this have become quite good and made for much more efficient and spectrally clean transmitter designs.

It can be difficult to quantify this without a direct RF sample from the transmitter and a calibrated analyzer, however. An SDR will still give you a fairly good "relative" picture though (assuming it isn't overloaded on the input).

--Shane