I've been 'dreading' answering this, because I know that it'll have to be a long answer, and will still miss a lot out.
None the less, here goes:
Boy, there were a lot of changes! There were more changes made during production of the thousands of consoles than I could possibly determine. -I can lay out a few of the significant milestones, but I caution anyone reading that they should NOT take this as a definitive work on the subject, and that most consoles out there nowadays have been augmented with modules or sub-assemblies from other consoles, so there is likely NO litmus test. -In addition, since some of the changes were signified by "knob cap color" (like the black-knob EQ for example) some owners changed knob caps to make the clients feel like their console was more up to date... Never judge a book by its cover, nor an SSL EQ by its color.
-One thing about the SL4000, 6000 and 8000 is that it was designed to be fairly easily serviceable, with a pretty much entirely passive motherboard, and all of the 'definitive' circuitry on sub assemblies. -the same was true of the master section also.
This made things wonderful for servicing (allowing whole sections to be swapped to swiftly fault-find, or get a module up and running again FAST), but also meant that when revisions were developed, they didn't have to re-lay-out the ENTIRE channel strip, for example... only the EQ board,, or the input amplifier card, or whichever part had been redesigned needed to be re-drawn; the rest of the production line (and they were SO successful at the time that there was an enormous facility cranking out consoles and STILL struggling to meet demand!) was able to keep churning out the rest of the modules without changes.
So... From memory, here's a quick summary:
The module comprises:
The motherboard with all of the pots, switches and edge connnectors on it.
There was one Big change quite early on, which was when Total recall was finally perfected. -The 4000 series went into production while Total Recall was still being developed, and so the early ones were produced and built BEFORE total recall was proven, and so didn't have the hardware (one extra pole on every switch, one extra gang on every pot, and the wiring and circuit board traces to carry the switch/pot wipers to 20-way 'scotchflex' connectors. these early motherboards also had push-on/PULL-OFF switches on the aux send controls.
The motherboard revision is etched into the copper trace, beginning "SL611..." and the last two digits represent the year of the revision. The letter preceding the year is a sequential revision letter, and the letter preceding that indicates the module SERIES... Thus SL611EK81 is an E-series module (there were B's and G's of course also!) revision 'K' the layout of which which was laid down in 1981. -the longest-standing revision i think was SL611EM83; an E-series revision 'M' which was designed in 1983. -They stayed with that for a LONG time.
The daughter cards... in no definitive order; just the order that I think of them:
Input amplifier card:
There were two E-series cards, and a G-series card. The original E series had a Jensen transformer on the mic input, later there was a transformerless mic preamp option offered There were versions with NO mic preamp, just two line amplifiers (for switching between tape machines, for example... or between line level studio outputs in a broadcast environment. -Japanese and European broadcasters were significant customers for SSL) In addition, the input amplifier card also houses minor stuff like the polarity-reverse circuitry and the routing matrix pan buffer, also the insert return... that sort of stuff. Even the basically-similar cards like the transformerless E series were revised a couple of times, as small improvements were developed and incorporated into the FET switching, for example.
Some people claim that the transformer E preamps were harder, others that they were softer. -Ignore everyone and form your own opinion. The G-series version was transformerless, and used a rotary switch for gain control. Since they could no longer use a 'pull for phantom off' on the gain knob (couldn't get a sufficiently high quality rotary with a pull switch in it) the gain range was extended DOWNWARDS so that a pad would not be needed, and the old 'pad' button was re-purposed as a phantom power switch. -This required some modification on the motherboard, and you'd BETTER change that switch cap out if you ever retro-fit a G-series preamp to an E-series module, otherwise you're going to have some VERY irritated engineers if the unthinkable 'phantom-damage' scenario ever occurs!
Out in the wide world, there are also a couple of consoles with aftermarket input cards made by Alan Smart's company.
There were several of these.
The original one (82E02) had brown LF knobs. The slope of the boost/cut sides altered with the degree of cut/boost. (I'm going to avoid "constant Q versus constant bandwidth" type terminology, because of the conflict between technical accuracy and user comprehension). It was an excellent design, and people liked it. HP and LP Filters were 12dB per octave.
Because it was plug-swappable, people suggested the possibility of making alternative versions. Since the Pultec EQ was always sought after, SSL developed an EQ which mimicked the Pultec's curves. They made it. they sold it. -It sounded better on paper. It's not universally loved, though it has its fans none the less. - I think it had Orange LF knob caps so that in a console with more than one EQ tye, people wouldn't make the mistake of splitting a sterep pair across different types of EQ, for example. Filters were the same as the 'brown-knob' original.
Then there was the Black Knob EQ. (82E32-?-) There is MUCH misunderstanding about what changed, but basically here's the story. -People pointed out that whenever you switched in the EQ, the filters were automatically inserted in circuit... whether you liked it or not. -Since the least intrusive filter settings still lopped everything at 20Hz and 20kHz, some significant phase shift was happening in the channel, and the top and bottom were being trimmed, even if your only reason for switching the EQ in was to shape the mid by 1dB... and the low-cut filter was the most egregious, in that people FELT the change between the EQ -even set completely flat- on a kick drum, turned up LOUD. -the 'push' of air felt different, and people were becoming very unhappy.
So the Black EQ (82E242) offered 'click' switches on the two filter knobs. -PULL switches were out of the question, since that would have been impossible to track using total recall (which only had a rotational position input to record and compare) so 'endstop switches' were used. these bypassed each filter. The EQ card was then modified to STEEPEN the cutoff slope from the high-pass (low-cut) filter, since now that the intrusion of the filter -even if you didn't WANT to filter anything- wasn't annoying people so much, they could use a more 'manly' slope, which they'd been unable to do before. A couple of small changes to minor stuff like the band-summing, but the actual EQ behavior really remained unchanged.
Then there was the G-series (82E292) EQ. Now things changed; both operationally and in terms of behavior. -It was launched as a response to the original Rupert-Neve produced Focusrite EQ, which was suddenly VERY popular. The Focusrite used limited-range frequency sweeps for the upper and lower midranges, together with a "multiply by 3" and "divide by 3" buttons. There was also a distinct 'dip' before a shelf boost (and mirrored 'bump' before a shelf cut). In addition, the Focusrite BEHAVED very differently as you increased the boost or cut... instead of the "corners" of the "mountain" (or valley, if you were cutting) staying in the same place, and the mountain/valley 'slopes' becoming more steep as you increased the peak deviation further from flat, the Focusrite kept the slopes THE SAME, and "widened the base of the mountain" by moving the two corner frequencies further apart. -O, and the Focusrite could cut or boost by 21dB per band, instead of a measly 18dB.
So, armed with this and other analyses of the Focusrite, and with an army of new engineers concluding that the SSL could use a new EQ design, the G-series was produced. The "multiply/divide by 3" buttons were achieved by re-purposing the old 'bell/shelf' buttons. Now the user no longer had a choice of bell or shelf at either the top or the bottom, and one was now permanently fixed as a bell and the other was fixed as a shelf. -For this alone, many people were unhappy with the G-series design. -But in addition, the 'widening mountain base' behaviour -while laudable when boosting, since the ear/brain has a tendency to zero-in on the steepness of a slope as somewhat characteristic to a particular 'sound'- it produced a rather nasty side effect if you were cutting. -The older EQ allowed you to 'notch-down' a note if say, a drum had a ring that was offending someone, and it was too late to fix it (fix-it-in-the-mix, anyone?) by keeping the bandwidth narrow, zooming in on the frequency and notching as deeply as you felt. It dd this without sucking out too many of the neighboring frequencies, or excessively altering the 'character' of the instrument. -the G series simply didn't let you do this... as soon as you started to notch -even on the narrowest bandwidth setting- it would widen the Q and start sucking all the life out of whatever you were EQ-ing. Subtractive EQ-ers generally HATED the G-series "upgrade".
As a result, early on into the G-series, it was 'dropped' as the standard EQ, and the black-knob variant was returned as the standard, with the now slightly tarnished G-series available as an option for those weren't bothered by its 'subtractive' performance, and didn't mind surrendering the ability to choose bell or shelf at the HF and LF sections.
G-series EQs were differentiated by pink LF knob caps.
Aftermarket EQ cards were made by AAD (Amazon Audio Developments) which were based on Amek/Neve circuitry, and Maselec, also based on Neve circuitry. There's not many of these out in the world... AAD Equalizers had red LF knob caps, and the HF caps changed from red to yellow.
The Dynamics card is based around a VCA. Initially it was based around the B&B 1537A VCA, which came in a 14-pin DIP package. Supply variances and some other difficulties forced the design to be ported over some time around 1982, to use the dbx 2151 type VCA, which came in an 8-pin SIL package. -While there may not have been a great change in sonics at the time, the later evolutions of the SIL design by the THAT corporation have improved the performance DRAMATICALLY, and I have no compunction whatsoever in repairing broken cards with the newer THAT designs, should an older dbx require replacement.
Other than the redesign to accommodate the VCA change, there has been little significant change to the dynamics card, with only incremental tweaks to later revisions.
As distinct from the dynamics VCA, the FADER VCA card also went through a couple of revisions. The format was NEARLY always the 'large-can' dbx style, of which the earlier ones had the gold cans. -Though the very earliest did have 1537's. Sometime around 1986, the gold cans were dropped for a replacement -also made by dbx- which comprised a daughter card containing EIGHT of the smaller SIL-style VCAs in parallel, and a low-noise op-amp to drive the control gates at a suitably low source impedance.
Also on the VCA board are a couple of other functions such as small fader buffer, and the FET switching for fader swap between VCA and small fader. There have been a good number of incremental revisions made to this board, but in terms of signal path the VCA swap has been the only one of major significance.
Channel Amplifier Card:
This does a few things. Not many of them are glamorous, but they're generally pretty important. -It's a bit of an unsung hero actually; -there's the group summing amplifier for the multitrack feed. -Okay, that's easy enough. -Then there's the group/tape monitoring FET switch assembly for the monitor fader. -Okay, nothing too tricky there... UNTIL you realise that you can also hear a SUM of the two, with a 3dB level-correction. -This is an INCREDIBLY useful function when it comes to checking to see if a singer is anywhere near the note before you commit to hitting the big red button of no return! (DAW-raised pansies and their 'undo' function will possibly not fully appreciate how this ability to check can be a MASSIVE assistance!) Then... as if THAT wasn't enough, there's a SECOND set of tape/group monitor switching circuitry, which feeds any pre fade cue sends listening to the monitor path... so that even if YOU don't want to hear him before the punch, HE can hear himself before the punch... or he can hear ONLY what he's doing now, or he can hear a (3dB-corrected) sum of both... and when the big red button of doom lights up, the channel amplifier card REMOVES the 3dB correction, since there's no longer any "old" signal off-tape, so it switches to 'new-only'. -It is careful NOT to listen through the machine, which take care of any latency issues, whether working with a pro-digi, a DASH, a RADAR, a DAW or whatever... I mean how freaking COOL is that??? -And for the vast majority of the time, the engineer never EVER gives it a thought. -It just works. -it's pretty complicated, but it just quietly gets the job done.
The BIG changes to this card were brought about when SSL was being asked to supply ever-larger consoles. -From 40channels being considered 'big' in the early days, to 64 channels being considered unremarkable... the summing buses (which extend the full length of the console) were struggling to sum the signals quietly in larger frame consoles. The original, asymmetrical (single-bus, also called 'unbalanced') summing system was really being asked to do too much, and couldn't cancel induced noise to which a longer set of buses (32 track uses, six aux/cue buses, four or six master buses), and induced noise was becoming a big worry. There wasn't the ability to go balanced, because that would mean replacing every frame mother board, every routing matrix switch, and every CHANNEL mother board... clearly out of the question.
As a result, in the later 1980's SSL switched from the 'unbalanced' design to a 'quasi-balanced' design, where a convenient set of spare pins (which had been grounded for inter-channel separation) were turned into ground-compensated audio ground returns, which would then allow any induced noise to be canceled. -a few retro-fittable changes to the module, one new mother board per bucket, and some modifications to the channel amplifier card, and the hum was reduced. In smaller frame consoles, there was little need to do this mod, but all consoles from about 1988 onwards were made to the 'quasi-balanced' spec.
Other modifications included changes to the output drive circuitry for better drive symmetry, and some stuff like that... not very interesting.
there's no audio passing through this board, it just looks at the master status and the function buttons on the module and 'directs traffic' according to commands (fader swaps. mutes, solos etc.) one of the best upgrades which came with the Gseries milestone was the improvement of the logic board to generate voltages OUTSIDE the audio power rail range... this allowed the switching FETS to NOT "pinch-on" when the signal gets close to clipping... no significant affect for people who don't "push-it-to-the-pegs", but VERY good if you don't want grumbling clients who tend to regard VU needles as things to "park" over on the right. -The other feature which clients ALWAYS appreciate was the 'solo-isolate' function, which was added to the group amp trim as a pull-switch. THAT feature is worth looking for a board with G-series logic cards, for sure. The logic board also handles things like track-arming and other non-audio functions.
TOTAL RECALL card:
You'll never guess what this board does. -It was offered as a bolt-on option, and it comes with ribbon harnesses which carry one wiper of all of the buttons, all of the pots, any rotary switches, and the small fader (and uncle Tom Cobley and aaaaaaallll....!) to a row of scotchflex sockets, where leaves connect to twigs, twigs connect to branches, and branches connect to the trunk, all through the mechanism of multiplexing. -the 'trunk' then "Hijacks" the fader VCA line to and from the computer, and lets the module go on using its fader without 'siezing' in one gain setting, by simply hooking the send back into the return... sticking the snake's tail back into its mouth, so to speak, so that the computer is bypassed while it uses the "hijacked" lines to write, read and allow the computer to compare total recall positional information.
Okay, that's the module, and all that I can think of right now... Hopefully you can see that there was a CONSTANT stream of small, incremental changes, many of which were introduced as "ECO's" (Engineering Change Orders) during production. Not all of them were associated with sonics, some were for reliability, some were for component supply differences.
Along with the power supply, the center section, there were THOUSANDS of changes... but pretty much all anyone ever cares about is the module, which might be a bit blinkered, but to write about EVERYTHING would take all week...
So -not a simple answer. There were innumerable changes. The ones which people most consider seem to be the EQ and the input card... -Maybe the logic card if they notice the lack of solo isolate on each channel... but that's about it.