Accession (phono preamp)

Looks interesting Graham, I look forward to finding out how this sits in the hierarchy, SQ wise. I'm presuming the Reflex to be currently at the top (or maybe the Revelation with it's additional EQ options) perhaps the clue is in the name? The King is dead, long live the King kinda thing?

I have seen use of star earths over the years and some rather weird ones that look a bit like 'spider plants' that make little sense and achieve even less.

The old proper way was to have a ground "Mecca" at the input "earth", and a good few pages were written in The Art of Electronics on the rights and wrongs - long before today's rise in RF gadgets.

It still left many unanswered questions especially in multiple input/output devices as where exactly the input earth should be placed. It was also very much trial and error in power distribution as I once found working into the night on a desk system for the "beeb": the system hummed when it shouldn't have. The team were getting rather hot under the collar, and the chief designer spotted the culprit...

I'd had the job of laying out the power supply board and the only way I could route the ground was from a rather thick track on the "input" side of a large 'smoothing cap'.

The system was pulling current against the charge current on the cap - hence the hum.

It was only about 3mm length of 3mm wide track, but sufficient to be a misplaced ground. Cutting the track and hard wiring to the cap's neg pin solved the problem and the system test got finished just after midnight - in time for the "beeb's" engineers appointment the following morning.

That illustrates just how awkward grounding is. If the board had used a ground plane the problem would not have occured. However, that would have required through-hole plating and the in-house PCB facility wasn't able to make them.

Low frequency return current flows everywhere in a ground plane, and in "star earthing" it just follows the track layout, and providing there are no daft errors like the above, all is fine.

However, at the commutating speed of RF which gets in so easily in these days of wireless prolification, the high frequency signals want to follow a return path that mirrors the way it came. A star earth more often than not disallows that mirrored path - the result is rising ground impedance, and the edges of the traces (tracks) become antennas - compounding the problems of RF interference.

The same happens when a ground plane is used but has a slot in it to clear an obstruction.

Therefore, the answer is to use a "solid" ground plane with the only clearance being the pads surrounds - you can't even have a slot to clear a number of inline pads because that results in another antenna.

Logic chips ("glue logic") having smaller dies have become extremely fast - rise and fall times better than 1nS - so the need for ground planes to allow what is in effect RF current to mirror the route it came is just as important - even if the operation cycles are long, the switching transitions are definitely in the radio frequencies.

Without a proper case - top and bottom extrusions without a front or rear panel - the Accession passed Keith Armstrong's tests, with a little help from a little copper tape from a screw port to the "shielded" phono sockets. The fully cased Gram Amp 2 Communicator he tested was borderline - the culprit? The star "earth".

Keith also argued that a properly shielded audio circuit complete with ground plane sounds better citing a studio mic preamp design he'd laid out the PCB for. In my own opinion, the ground plane has contributed to the clarity of the Accession. The Majestic DAC also uses ground planes but perhaps not to the nth degree of correctness as the Accession. I am definitely persuaded in favour of ground planes.

Great work Graham. It sounded very good at the Roadshow and I can't wait to hear one at home.

It's to eject music which sounds like a cat is being strangled...