Just found this forum by chance, fantastic summary by Keith - including many things that I had completely forgotten. Let me introduce myself and explain...
From 1985 to 1993 I was basically SSLs "caretaker" engineer on the 4000/6000. In the mid 80's Colin had decided that the 4000 was dead as the world was about to go digital. And so little happened until 1987 other than, as Keith said, the pseudo-balanced upgrade I added (following an original concept from Colin and Paul Frindle) due to problems we had with the installation of a 64 channel right-angle desk at SWF in Baden-Baden which picked up TONS of lighting dimmer noise. We ended up flying half the manufacturing team to Germany to rebuild that desk on site, something I will never forget!
By 1987 it was clear the digital desk wasn't "just around the corner", so Paul and I were given the "bothy" at the bottom of the garden and told by Colin and Chris Jenkins "just make the 4000 sound better". Best (and scariest) job challenge ever! For anyone who doesn't know him, Paul is a fantastic, and deeply intuitive, audio design engineer, with an extremely good pair of ears. I have cloth ears (I'm a bassist :) ) but by this stage knew the 4000 inside out.
The original concept was just to produce "Golden" retrofit cards (the G of Golden would of course later reappear) for the I/O, to give more dynamice range, and, as Keith says, to produce a rival to the Focusright EQ. The curves of the G series EQ are...ahem...very similar to the Focusright but the circuitry was completely different - wein bridge rather than state-variable which gives the G series better dynamic range (particularly in cut) and lower noise. Sadly what we couldn't copy was the knob size of the Focusright, hence the vicous gain control on the G series. We did what we could on the control law to give some sensitivity but there was only so much we could do.
Now, at this point marketing took over. Paul and I never intended for the G EQ to replace the 02 (Colin) / 132 (John East?) / 242 (Trevor Stride) choice - our ideal desk would have been populated with half Gs and half a more subtle EQ. But there was drive for a "new" desk so SSL went all G with the G series. Ah well, all I can say is...sorry, we tried! (I did manage to sort this out a bit when I designed the 9000 EQs, with a lot of very kludgy switching, not one of my most elgant circuits!)
Keith mentions the x3 and /3 - the eagle eared (?) or test fanatics amongst you may have noted that the /3 is actually /4; was this a result of deep psycho-acoustic testing in our research dungeons? Nope, I just calculated the capacitors wrong (doh!). Seemed to work though. Particularly for a bassist :) I'd love to know whether the software version incorporates my error!
Keith mentions Pink knobs for the G EQ - were they? I honestly can't remember, more likely they were red but tended to fade to pink. Colin hated pastel colours, I originally used yellow for the G EQ prototypes as a) it tied up with the "Golden" marketing theme and b) it was the only standard Sifam clour we hadn't yet used. Colin walked in to the demo studio, saw the row of yellow knobs, and explained to me at some length and with some vehemence how c**p it looked. And when I stood back he was quite right.
Where my memory completely differs form Keith's is that as far as I know the G EQ remained the standard at least until 1993 (when I left), but it was a long time ago now and I'm quite prepared to be proved wrong!
The I/O board changes were to increase the dymamic range on the line amp and get rid of the transformer artifacts on the mic amp. The mic amp is all Paul's work (I'm no great transistor circuit designer), the line amp took us weeks of playing around with ideas until we got there - after which we found a similar circuit in a textbook which had been on the shelf the whole time! Doh again. Nice circuit though, with ludicrous dynamic range.
As Keith suggests, the logic and channel amp card changes were more straighforward, but did perhaps have a bigger impact on the channel sound than you might expect. He's described them very well, the main aim was to make the FETs switch properly (mainly to prevent breakthrough, but partly to prevent distortion) and to buffer some of the dodgier switching on the old "11" amplifier card.
What is a shame is that we couldn't take the time to sort out the 651 centre section, but the problems there were so huge and intertwined that there was no way of starting without taking on the whole job. And by this time it was decided to focus R&D on the new analogue tracking desk (the "ARC") and digital mixing desk (the "DMC"). Neither of these ever made the light of day, however much of the ARC circuitry resurfaced when Dave Mate and I designed the 9000 (I designed the channel strip just before I left, and Dave designed the centre section).
Finally the VCA change, yes sadly DBX obsoleted the rather nice 202C "gold case" version with the 202X "bare pcb" version in about 1986. Various people, particularly I think Trevor Stride and MIke Large, tried to solve the distortion problems in the 202X, in the end Ultimation (which I wasn't involved with at all as I was working on the ARC at the time) kind of sorted it out. I never had anything to do with the old "black case" version (I joined in 1984), but I believe the gold worked better?
Good to be reminded of fun - if stressful - days.
And if anyone (in the UK) has a 9000J I would love to see one; because I left SSL during the prototyping stage I've never actually seen a complete desk. Someone kindly sent me the circuits a few years back (which sadly I then lost in a PC crash) and it was good to see that the channel strip design stayed as I left it - albeit including some bits of design that made me wonder what I was on at the time!