Robert - that's a fabulous bit of work! Well done - and I'm really glad that you like the sound of the "David".

The design began (a very long time ago) when a friend of mine got a job at a major market Californian radio station, as their RF Engineer. They had a lot of sites, and a huge amount of transmitting kit, and he was constantly travelling from site to site, keeping the hardware working properly. They used (mostly) RCA transmitters, and he discovered (to his cost) that "RCA" really stands for "Requires Constant Attention"!

They had a combination of Gentner Prism multiband limiter, Orban Optimod clipper / limiter, and Bext stereo coders. The processing gear at each site cost tens of thousands of Dollars, and really didn't sound too good, because they'd turned everything up to 11 in an effort to sound "louder" than their competitors.

We decided that a slow, shelving AGC (to minimise gross level differences) followed by a three to five band compressor / limiter, then a clipper followed by a filter to shave the overshoots, and a highly oversampled digital stereo coder would give a better sounding result than the really expensive combination they were already using, and could be made for just a couple of thousand Dollars! PWM control was a well established principle, and eliminated channel-to-channel errors and the need for accurate calibration.

Since Orban was seen as the "Goliath" in the industry, we called the one-box solution the "David" - a single 19" case, with minimal adjustments and plenty of blinking LEDs (blinking LEDs always impress the programme controller types at the stations!).

It has been a pretty successful product over the intervening years. It's also gone a long way to improving the overall "sound" of FM radio, and led to the end of the silly "Loudness Wars".

Incidentally - your PCB layout looks fabulous, and you've managed to include a LOT of groundplane!

Thank you for sharing this piece of history. I've always wondered who Goliath was.

David III is a robust, easy to align, clever design. The only modification I made was replacing the 1.216 MHz crystal (which is hard to find) with a 27 MHz TCXO/SI5351/Attiny.

I added a 19 kHz output, RDS input, MPX CLIP indication, Overshoot indication, borrowing circuits from the 708/715/716 series.

No issues regarding digital noise; the ground plane, individual decoupling, and careful routing of the PWM circuits worked well.

I found only one mistake around IC45B feedback — the schematic (rev. A, page 1) seems to be incorrect. I had to slightly modify the configuration, as I couldn't achieve the correct level for the preclipper and AGC.

Unfortunately, Jim hasn’t replied, and I was curious to see the correct schematic. I don’t own an original DAVID III to crosscheck.

Anyway, it was an interesting project with no commercial purpose, only to learn more about PWM techniques in the audio domain.

The Inovonics manuals and schematics are very interesting—it’s a pleasure to go through them.

P.S.: The LED play is really captivating, especially PIPP

The other crystal option is the (fairly) common 4864 kHz crystal, which can be divided by 4 for the 1216 kHz clock.

You're right about the IC45B feedback - as drawn in the manual. It's a transcription error (as I recall). There were also a few other errors, but most of them have been corrected over the years.

Your additions - the MPX CLIP indication, RDS port and so on are all very sensible. When the original was designed, RDBS was a long way in the future, and the FCC didn't seem to like the idea! There were a few other changes I'd make - with the benefit of hindsight - but most of them are trivial. As originally designed, it was a solid product and became very popular as a relatively cheap - but highly effective - option, and became the choice for the smaller operators.

Broadcast Warehouse (Roger Howe, et al) tried to emulate the function of the David with their DSPX and their digital / analogue stereo coder. I admired their adept use of the PIC microprocessors, but their products didn't quite make a dent in the market despite their aggressive pricing and perpetual representation at NAB events and so on.

Back to the coders: I was always concerned about the phase shifts introduced by the MPX output filters and the compromises that were introduced in order to be spectrally compliant. In subsequent designs, I've included a 114 kHz multiplexer which is used - inverted and attenuated - to null the predominent 3rd harmonic products. This means that the output filter slopes can be much more gentle (since there's virtually nothing at the 5th harmonic), and the phase coherence is significantly improved. As ever, electronic design is a set of compromises - increasing complexity for ever diminishing returns doesn't make the accountants (and CEOs) particularly happy!

The Zenith / GE stereo multiplexing system was always a set of compromises. Had I been devising it (all those years ago), I would have tried to "future-proof" it by having a significantly wider audio bandwidth - 15kHz has always seemed very low to me. I appreciate that - at the time - the limitations of the "microgroove" vinyl records and then-current tape recording equipment meant that there was little or no content by 12kHz, but transmissions of live concerts would have sounded much better with the audio bandwidth extended to (perhaps) 21kHz - giving a nice, gentle slope to the filters , with the start of the roll-off at around 19kHz.....

Don't get me started on the abysmal DAB system!

Thanks for your interest in the principles behind that coder. Your reproduction of it is exceptionally good - especially when I remember the early bench versions of the various sections - often built point-to-point over a sheet of copperclad, or thrown together on "Veroboard"!