Applied Technology: Wheatstone baseband192
Nov 1, 2013 5:38 PM, By Steve Dove, Wheatstone Corp.
It seems quite extraordinary given our serious computer-in-the-pocket lifestyle that it has taken until just recently for FM to become (potentially) digital end-to-end. Everything between the microphone and speaker can be digital, at last, all while maintaining the legacy FM RF path.
The main reason for the airchain processor traditionally sitting at the transmitter site has been to minimize communication aberrations: A few feet of BNC-to-BNC was, for the most part, less impactful on the signal than any STL. Now with baseband192 -- where the entire multiplex (MPX) signal is generated digitally by the airchain processor, and captured and modulated digitally by the exciter with no intervening conversions or modifications -- the processor can happily sit at the studio complex where it belongs. It can be where monitoring is good and its blinkiness adored by all. All in the sure knowledge that the signal isn't going to get even the slightest bit bent before it hits air, assuming a solid enough and wide enough link along the way.
The last significant breakthrough has been the digital FM exciter. Although Direct Digital Synthesis (DDS) and similar techniques have been around a while, it has taken the serious efforts of the transmitter manufacturers (tips of the hat, fellas) to perfect the technique, in particular the minimization of sampling-frequency artifacts. What we now have is a method whereby one input digital word results in a particular exact transmitted frequency. And does so next Tuesday, too, free of most inherent analog "wiftiness."
You transmit dc and don't know it
A less than obvious benefit to this is that the end-to-end frequency response of the system can now extend down to dc. It never really could before, since an oscillator capable of being frequency modulated also needed to be reminded where its center frequency was supposed to be, usually by a servo mechanism operating at and near dc. This precluded audio modulation getting too low in frequency. Indeed, over-enthusiastic amounts of LF energy could catapult a modulator out of lock and the station off the air. (Ask me how I know.) The necessary high-pass filters imparted either phase-shift and group-delay, or if done digitally with phase-linear filters, additional broadband latency. None of which are palatable.
Additional latency can preclude the use of the off-air signal for announcer monitoring, or for as a free cue path for remotes and such. As for phase-shift, any -- that's any -- modification to the carefully and exquisitely constrained signal leaving the airchain processor messes it up; while in this matter attention is usually focused higher up in the spectrum, LF phase shift from such high-passing counts. The main effect is to partially undo the hard-won and expensively crafted peak-control, creating overshoots. Either these have to be accommodated by reducing the average signal power (precious loudness, poor dears), or further re-processed out. A pure digital path finally does away with this perennial bugbear.
And hip-hop stations can rejoice in unfettered thumpiness.
From the outset the Wheatstone processor hardware platform -- essentially common across all our airchain products -- has used 192kHz for internal connectivity, and our AES-format transceivers are programmable as opposed to constrained by the capabilities or otherwise of third-party chips. In short, implementing basband192 was a doddle -- everything was already in place.
Additionally, we have always treated SCA sources differently. We A/D them at source, then add them into the overall transmission MPX digitally. Why is this important? The common way of dealing with RBDS and SCA signals (narrowband modulations typically centered at 57, 67 and 92kHz) has been to sum them into the MPX output in analog after the fact, which is obviously impractical in an all-digital stream. Doing so defeats the purpose and advantages. Already having the SCAs in the digital domain as part of the digital MPX stream gave us a complete, ready, baseband192 solution today without heroic redesigns to contend with.
The baseband192 AES3 format at first seemed puzzling -- just using the left channel of a 192kHz sample-rate AES3 format. The engineer in me initially thought that ping-ponging left and right on a 96kHz AES3 would be probably both more readily generated and decoded -- common chips handle 96kHz with ease -- and was less problematic to transport. That said, running straight 192kHz does have the advantage of not requiring anything fiddly at the receive end to de-interleave the signal. However, since straight 192kHz was a given (determined by and obviously convenient to a transmitter manufacturer), unused capacity within the 192kHz stream exists, and has potential. Presently, Wheatstone processors are sending two separate transmitter feeds down the left and right channels of baseband192.
384kHz sampling of the MPX has been suggested, with the samples interleaved ping-pong fashion onto the existing baseband192 192kHz stream; this would allow bandwidth capability far exceeding that ever used in analog FM, encompassing in particular the once-common 92kHz SCA slot, rendered marginal at best by 192kHz sampling. This, however, makes the assumption of a need for the bandwidth.
Signal flow of a Wheatstone processor with baseband192.
Don't shoot the messenger!
A squint around a few markets minor and major with a baseband spectrum analyzer has shown modest SCA usage beyond 57kHz RBDS, and none at 92kHz. I'm certain there are some 92kHz SCAs out there somewhere, but -- don't yell at me -- this hardly seems a realm undergoing bursting growth. In fact, far from it. This is hardly surprising given the defection to more convenient, ready, and capable satellite or Internet distribution by those once interested in SCAs.
Although the additional capacity in baseband192 is a temptation, it should be borne in mind that whatever may be dreamt up for this fallow stream will require specific software/hardware to unfurl, something the present simplistic and AES3 compliant method avoids. Any additional effort and complexity beyond baseband192 is probably best put toward true network connectivity between processor and exciter.
That, as oft said, may well be "the rest of the story."
Dove is minister of algorithms at Wheatstone Corp., New Bern, NC.
Build an effective back-up transmitter site, step inside WTMD, the FCC updates RF exposure rules, and Wheatstone baseband192 is explained....