Hi all, I was thinking of putting my thoughts on this site while embarking on a journey with a project was perhaps a good idea.
My background is high end developper and I lack experience on things like mixers.
Please refrain from negative comments about what I am telling is nonsense.
My intention is to give a peek into my way of thinking.
For me it is real, and I don't really care about people who think I am nuts, or a spinner.
I am not native English speaking, so forgive my faults in that area.

So: I finally found (by accident, as it should) the ideal mixer for modyfying.

What are the good things about the Soundcraft 200B 16 ch, what makes it ideal for me?
1) It's mechanical sturdy build
2) Modular, so easy to work on
3) All analoge, even the metering.
4) 16 ch is the sweet spot for me: 8 ch not usable, 24 channel not needed
5) Well designed: no obvious faults found in the design till now, and that is quite an exception!
6) Perfect faders and good pots, though some switches need to be replaced.
7) External (analog) power supply.
8) right time area: no SMD, no special parts
9) Schematics available! Saves tons of time, no need to reverse engineer.

The mixer is well worth the investment of time and money as it will be able to function many decades again after the modding.

An high-end persective: I like the PCB's floating in air, not mounted near ALU (for the most part).

Looking at the design, again it is all very functional and straight forward: very well done!!

First work was on the power supply. The +48V supply was totally ruined. I restored it.
The standard BD135 is substandard and should be replaced by a BD139. (see another post from me elsewhere)
Impressions: the case is vibrating: wrong!!
Best would be to mount the transformer in rubber grommets, but hanging on the front that would impede reliability.
Also the PCB was hard mounted on the bottom: wrong again, certainly in this case.
I choose to postpone working on the power supply. For the rest the power supply is well designed in general.

First I want to establish a base to start from, a point of reference: without giving tuning and careful selecting parts a thougt I have started to replace the parts that are most prone to aging: the elcaps.
I had Rubycon ZL (100mu/50V) and ZA (47mu/50V) at hand. I am deliberatly not changing the values, except for the 47mu powersupply caps: they will be 100mu.
The point is I want to preserve as much as possible the original gestalt of the mixer first. Using 1000mu caps for the local power supply caps is very intrusive.

I started to modify two input modules and the master section.
What is the best and most precious tool to improve a newly build device?? TIME!!!
There is "a order" in old audio equipment, a sense of unity. This can be explained, but not now.
When starting to change things we not only have new parts that needs burning in, but we loose the unity factor. We have to factor this in when listening at each step we make.
Changing the caps made (of course) a huge difference, but also put me on the track to eliminate all ceramic junk from the signal path, replacing it with same value polystyrene.
I had a short moment of reasonable unity of sound (where nothing stands out negatively, one can even enjoy the music), but today it sounds sh*t again: not changing the essential design, holding on to non intrusive replacement of parts pays here big time: we can just go on, and let burning in do the work. We can be sure that no step has been made that had severe lasting negative influence.
Reason to progress quite fast in this stage is that it saves a lot of time. Once the general work is done, we can take some time (weeks or months) to let all things settle, allowing the device to gain more unity of sound.
The point of this two step approach is also that we have to listen to too much layers of information. Doing the stage one modding also gives a lot of information, and helps to deconstruct the layers of structure in the sound.
I use a CD player for listening: not optimal, but in this stage good enough, and good stable reference: it will be left powered on for the time during the first stage of the modding. Same goes for the head amp. I use a HIFIMAN HE-6 for listening.
So: it's about cutting through the first layers. It is difficult to eplain the term LAYERS in this context. I will try perhaps another time.

The "loss of unity" is a big thing. There is no other way, but it can take easily take two years before things are really breaking in, and we have to wait for over a decade before part of the charm of an untouched (good) audio device is regained.
Using know parts is a big pre, and of course much experience listening to the influence of parts is very important.
As the IC's will be changed anyway, I choose to order the (for me unknown) LM4562 to replace all NE5532 types. The goal is to find good replacements for the TL072 and SE5532 with low Vos, so we can kill much of the coupling caps. Starting to esperiment with the different kinds of elcapos in phase two, where optimal balanced audio is the goal, we'd better reduce the amount of the elcaps drastically!
I believe in the malt versus blended approach, where it goes for same functionallity for the caps. (coupling being one, power supply another). It helps very much to gain unity of sound!
That means replacing all caps in the signalpath for best evaluation. This doesn't mean that first probing two won't be used: it certainly wil. Depending on the result there will be a go no go for recapping all caps in the signal path.
Better opamps up to today's standard with the very important low V offset, allows again to start the real work in a setting that is closer to the end result. I think it is the best way.

Looking forward: I guess that the first thing I will do after stage one will be desiging and building a total new power supply.
I think the result will be beyond believe, the most important upgrade after stage one.
Because of the huge impact, this should be done before the tuning with different elcaps and IC's.
Without it the tuning would be most likely to the soft side of the spectrum making up for the grain in the sound that is caused by the power supply.
One of the big problems I have with this, and almost any powersupply, is the lack of separating line and load regulation. At the end your regulator is an amplifier. How about feeding an amplifier with a sawtooth voltage??
There is another way of abstract thinking I sometimes use: WHERE IS THE MOST POWER CONVERTED?? Which parts are doing the most work? This is a kind of thinking in energies.
Using big value elcaps is kliling the air in sound. Same goes for the power supply. Those big elcaps are only good for reducing the ripple, but at the same time we reduce the loading periods, leading to very large peak currents, causing much disturbance in the elcaps.
Have a look at the power supply of the modules: a series resistor of 27R or 10R, followed by an elcaps of 47 mu. (fk is much above 20 Hz hehe!)
So do we need ultimate load regulation? It would serve to reduce crossover of the powersupply modulation to other channels, yes....
The integrated regulators are not linear, but in this case especially the lower frequencies need to be taken care of.
Would "cross modulation" be harmful?? Not so much with the current bus system: the gnd currents are anyway modulating the gnd wire.
More severe is the current flowing through the gnd wire caused by the AC component of the power supply: where the induced currents of the load is almost 100% related to the audio, the first mentioned has zero relation to the audio: this is much worse.

So: we can fret without end about PSSR of IC's, but that doesn't adress the modulation of the gnd wire. (mind you, it is not a balanced signal running throught the bus system, makes you think...)
So yes, we need an ultra clean power supply, where line regulation is the most important factor, not load regulation.

Ever thought about the fact that electrons do have a mass??
Vibration on conductive parts of an amp (and even those who are not part of the signal chain) do influence sound.
Typical sign: very compact bass, but not blossoming: also bass needs air! Further: grainy sound, lack of (micro) dynamics.

Opamps: I don't like them.
They were designed for the analog computer, where amongst othere the long tailed pair input was a miracle solution for the purpose.
They should be banned for audio...
Having said that: I have hear really very good results with my own designed power supply, beyond anything I heard before.
But besides that: they are not suitable for really high end. (the term high-end has been devaluated over the years)
On the other hand: there are no practical high end mixers for the on the road job, and we have to be practically: we have to deal with the fact that given the size, price and option set of a mixer, there is hardly another possililty. Safe for SMD perhaps, but I don't like SMD for several reasons. Never heard of an SMD high end capacitor, and we will need them in a discrete design.

By the way: I saw something about metal cans sounding better than the ceramic or the plastic housed IC's.
It's due to he mechanical stress on the chip.
There was a very know pre-pre from ELECTROCOMPANIET that only after three days sounded right. (year in, year out)
A friend of my complained that his high-end stereo wasn's sounding right lately. He later found out that the pre-pre was powerd by accident from a switched power output, that was switched each day...
The plastic has to even out it's tensions after warming up.
The culprit being the 2SB 737 if I rember it right.
I choose to buy first plastic IC's in oder to stay as close as possible to the original. experimenting with metal can's will be a mini project in itself in the second phase.
Very carefull procedures have to be followed in order to gain knowledge.
I designed over 20 years ago an MC pre-pre which was sounding surprisingly good: it used metal cans. It always has been stuck in the back of my mind that I should research the metal can thing again, so it was nice to read something about it by Jim Williams.
Unfortunately the metal can transistors were in those times (and now obsolete I guess) largely US en European, not the electrical most often superior Japanese transistors.
I bought recently a cheap second hand Yamaha power amp from the eighties. Reason: it uses Japanese transistors, and as the rest is quite well, it sounds good!

The input amplifier is very good designed I think. Once I had drawn it on a different way than the schematics it was easy to understand.
The opamp delivers diferential current feedback to the long tailed input pair.
Besides that there is a servo build in to minimize DC offset.
Offset is furthermore kept in pace through DC impedance matching of the input. The current delivered by the base of the transistors gives a substantial DC offset at each input: equal impedance means that the base DC voltage will about the same of both transitors (depending on the matching of the HFE), so that the DC balance in the preamp is not disturbed. Better would have been if the transistors would have thermic coupled: this stabilzes drift. (we are talking far below audio however)
Doing the simulation (also calculatiing yourself) shows that the DC blocking cap of 200 mu total (2 * 100 mu //) is really necessary to keep.
Sad conclusion is that each elcaps are there for a reason, even the input ones (mentioned DC balancing of the V offset at the base of the transistors), and they can't be omitted without special measures.
This is the price to pay for a very adequate solution for an amp that will perform well from very small to larger signals. (small signal you really need a lower emitter to emitter impedance, and vice versa for larger signals: this way it will sound about the same over the whole range).
Current feedback is a very direct way of feedback, and as I see it, it avoids one frequency pole.

This is the schematics of the input stage in the MicoCap9 computer simulation.
It is not meant as 1:1 schematics as it's purpose is for simulation, not for drawing the schematics in itself.

I replaced today all power supply capacitors (all remaining input channels, subgroups, aux group).
It really made a difference as expected. They are all // to the output of the powersupply.
I wanted today a big step, and as it was tactically the right time to do, I replaced the raw powersupply for my own design. This means that the transformer and the elcaps are replaced by my own configuration.
With the standard analog power supply the LM317 and LM337 see an sawtooth of 400 m.v. peak peak. (This with a 16 ch Sounraft 200B as a load, and the main elcaps were 4700 mu each)
My raw power supply deliveres the LM317 and LM337 with a sine wave 50 m.v. peak-peak. To be honest I used some black gates, but much smaller value than 4700 mu.
Listening: AWESOME!!! I already expected the big improvement, but what an effective way of improving the sound!!
Especially the high frequency spectrum is much improved. Decay of notes are much longer at the end of songs, so low volume detail is much enhanced. Much less grain, more liquid sound.
For the first time that it poped into my mind that the Soundcraft reached the HIFI state (in my subjective judgement, not the technical descirbed one).
Far, very far away though from being high-end.

This is still with the LM317 and 337 in place: of course total not suitable for the job. Technically perhaps yes, sound-wise: no way! (and it CAN be technically explained!)

At some time I will replace the LM317 and 337 for my own developped very superior discrete version. Sorry for boasting, but it took me only 20 years of very hard work to get to this design.
It is a fix: it won't be changed as it has found it's gestalt. It works miracles in discrete preamplifiers, voltage stages of power amplifiers, B+ supply for tube amplifiers, the filament supply of tubes, and... yes IC based designs.

To be honest: START WITH THE POWER SUPPLY!!! Expensive IC's and swapping and testing is pointless without FIRST building a better Power supply.
I won't give away my secrets, but I can give the information to see and hear for yourself the result of working on the power supply: try to use smaller elcaps, and use chokes. It is not per accident that I performed this step in this stage: i have learnt over the years to plan in a meticulous my progression through the process. It sometimes happen that I am stuck: I wait for the moment I am SURE about the next step, even when it takes two weeks for instance.

I know that the Soundcraft will be second to none other soundcraft once it is finsished: this due to a superior power supply.
I haven't tested yet more expensive opamps, I will soon, but I know already that if I had to choose between the standard LT072 and NE5532 and my own power supply, or the best opamps and the standard power suply I wouldn't have to think about it for a moment, I would choose hand over fist for the better power supply.

My (still to be implemented) regulated power supply is flat between 0 Hz and 100KHz and has a PSRR of -120 dB. It takes a lot of effort to erase the information from the power grid, much more than I even myself thought. The information of the power grid has no correlation to the music, so must be supressed as much as possible.

By the way: the knowledge of the PSRR of my power supply is derived form MicroCap 9 simulations: it is very dependend in my experience. Perhaps I should find a clever way to do measurment with my Audi Precision Cascade two plus audio analyzer. But to be honest: I don't care.
The point is that I believe I have reshaped the power: it has mainly "accidental noise" (don't know if this in English the right expression).
Better perhaps coincidental?
Many, no most high-end people claim that their sound is better at night. This is due to the fact that there is less disturbance on the power gird. In my amplifiers it sounds during day time as good as at night time. all due to a very (and unfortunatly rather expensive) design of the power supply.

Again: you have to look upon the power supply as an amplifier. Even TIM can occur in a power supply! (perhaps I will explain this another time)

I thought today about the next steps to be taken, and the limits of the project.
I decided that the best was not to go all the way with this project striving for high end and end in a mess, so:
I decided to build a sturdy version of my power regulator. Smaller ones, in the best case one for each module would be superior, but that is one step too far.
I want to keep it simple and to be quite easy implemented.
The PCB's are not to my standards for high-end anyway: the copper tracks don't have rounded corners: In the past I used at most 45 degree angles, better than the 90 degrees, but now I round the corners too. There is very distinctable difference in sound pre and post the rounding of the tracks.
It sounds so good already that I contemplated keeping all old opamps in place.
On the other hand we are stuck with a load of coupling elcaps. Elcaps in itself are far from ideal components, but even more so if they are not biased with a voltage.
So the planning is to get the reasonable, practical solution, easy to copy by this community, safe for the power supply. (In the future it can be ordered I guess)
After completion it will be used in the real world: excellent for a testing phase! Further development will take place in another Soundcraft so that this one remains functional, and can serve as a reference.
So, at the end the work done on this Soundcraft will be:

1) Replacing all elcaps
2) Replacing all opamps, safe for VU metering by opamps selected for low V offset so we can
3) Removing all elcaps in the signal path that can be removed.
4) Replacing many if not all of the switches for new ones
5) Building a total new power supply from scratch.

With this all the integrity of sound will be closely monitored. There is no way to predict if in the end it will take much time to get the sound in balance and you never know what you will encounter.
In that respect it is a much better idea to keep this first project very limited and postpone real deep modification for a next project which builds on this project.

Again: it is all about planning and taking carefull steps, or else find yourself in the middle of a mess. Picking an order because you start with what you fancy the most is defenitly not part of this approach! :-)

This shows the PSRR (power supply rejection rate) in a simulation in Mircro Cap 9 (MC9) of the standard power supply of the Soundcraft 200/400 in red, and of a simple version of my own design, intended to keep things simple and down to earth, in black.

Also is shown the phase shift from the output referenced to the signal at the input.

The PSSR of the SC power supply is about 75 dB, from my powersupply is about 100 dB. (that's about 20 times better)

Note the peak at 24 Khz in de SC PS. This is caused by the capacitive loading of the output. At the higher frequencies the PSSR depends mainly on the decoupling cap.

When the decoupling of the reference of the LM317 in the omitted the PSSR wil decrease about 25 dB, so de LM317 is greatly dependend on that capacitor too.

I notice that the picture is not high res enough to see everthing clear: the sweep (horizontal axis) is from 10Hz to 1 Mhz.

Here the same, but with no help of the decoupling cap at the output of the LM317 and no reference decoupling.
The difference is PSRR is now about 46 dB, so my own design being 200 times better in this regard, and much more linear in the frequency domain.

I have done some more testing, and from what I see it would be a very good thing to try to install a 1 Ohm resistor in series with the output decoupling cap of the LM317 and the LM337. (by the way I used the better spec LM117 in the simulation because there was no LM317 at hand).

This I already knew, but it was awesome to see how things improved: much better phase and amplitude behaviour. Be sure to check for sure it doesn;t oscillate, but I suspect it won't. (on the contrary: I have experienced oscillating WITHOUT the 1 Ohm in series, and no oscillation with 1 ohm in series)

I already planned to write someting about the STUPIDITY to decouple an output of a regulator hard with a capacitor. Perhaps another time I will explain it in detail, for now simulation clearly shows my point!

So are all those people who write the spec sheets of regullators wrong and I am right? YES!
Go check for yourself: It will sound better!

Oops! I made a mistake: this is the simulation of the power supply I will be using! (the previous one has a output impedance that is too high for this purpose, I use it in special circuits that don't mind)
Now the simulated situation is that my power supply is 70dB better PSRR than the standard one from Soundcraft.
I didn't have however have optimalized the power supply for the current, and when done the specs will be less spectacular, but anyway better than the standard PS.

Multiple small power supplies would extend the flat frequency range to 100 Khz, but that is one step too far as explained before.

O.K. I did the tuning roughly. The end result will be close to it.
The graphg shows a 60 to 65 dB better PSSR, about 1000 to 2000 times: not bad!
Also the frequency range is flat to about 100 KHz! (+3dB)

O.K. I tested the 1R in series with the output 22mu decpouling caps of the LM317/337.
Stable, as expected.
Sound: there is a layer of stress removed, it sounds more "earthed" and improved resolution in the high range, thougt the latter still far from being perfect.
Keep in mind this is tested on a fresh recapped SC, and much better raw power feeding the regulators of the original PS.
Listened on a very high res HIFIMAN HFE-6, a high-end headphone.
Anyway, please react here if you have tested it and what you heard (or not). In any case it is technical the better soluition.

Will explain to some extend: the output of the regulator is one end of the feedback loop. By means of a large amount of feedback, the regulator will try to see on the output it's reference derived value.
For high frequencies there is a very low impedance to gnd in the form of a 22mu cap, so the regulator produces relative big currents to mimic the noise of it's reference (and the electronics that are between the reference and input/output).
So: this could even lead to certain amplifying stages going into saturation and/or to no current moments. Anyway you don't want the regulator pumping noise into the ground, and that's what happening.

Yesterday I measured the current that the mixer is drawing: 560 m.a., this was about 200 m.a. more than I roughly calculated!
My initial goal was to use a shunt regulator, but the problem is that unloaded all power is dissipated in the shunt regulator, and also you have to choose the current the shunt can deliver, it can't deliver anything more. This would mean different dimensioning for each mier: no way that a 8 ch or an 24 ch could be dimensioned the same, and also I read here a lot about choosing different opamps which often draw more current.

So late at night in bed I was pondering the situation, and I choose to use a series regulated power supply. I even got out of bed to check a PCB from a very old project: it seemded a good base for a series regulator. Also looked in the computer to the schematics, and found out it was even from my previous PCB/SCH editor. I looked at the pile of paper schematics from that previous program and found it.

Today I spend desinging on the computer with the simulator.
I am really very happy with the result. Next thing was converting the old PCB for it's new found goal, and testing it. (amazing that I found a perfect soltion in one day, and also it worked flawlessly. I cheked for oscillation whit a capacitive load on the output, and short circuited it to test the safety circuit.

It is now working and burning in, and had a first listening.

The new series regulator is extremly simple and effective. The graph shows the PSRR. It uses no decoupling caps, no reference caps. On half (- or +) consists of only 5 transistors, including short circuit safety.
The nice thing is that two regulators can be put in series where the PSSR goes from roughly -60dB to -120 dB. Notice the very linear behaviour!
It uses no zener for reference. Zeners are very nasy buggers. If you want precision and thermal stability you have to use them, but for audio they suck.
At the end noboby is going to look on the voltmeter and look if they voltage are exactly X volts while listening to music, recording etecetra.

You have to understand that all transistors in a integrated regulator are on a small dye, and all, including the reference get very hot when the output transistor generates a lot of warmth. Also all transistors work in their non lineair area mostly, and are fed junk power.
These devices are highly non linear, and are in fact not suited for good audio.

Putting two regulators in a row will be tried perhaps much later. For the time being I am very, very happy to have found such an eloquent soluition, and it's another jewel in the portfolio.

And the sound?? We have to listen more, let it break in, but the initial listening indicated a huge improvement over the standard regulator.
I will edit this post as soon as I can draw more decisive conlclusions.

It will be VERY telling if two in a row, with a factor 1.000.0000 wideband reduction of sh*t from the power line instead of 1000 will mean a big difference or not.
It seems a quite pure test: two times the same in a row should sound about equal as far as it goes about the sound of the electronics itself.
It is to be noted that the raw power supply feeding the regulator should also be included in the PSSR, so now we have a power supply that has a better PSSR but also is very, very linear compared to the standard PS, also to DC, and also without a cap from reference to gnd. (it also has no memory in that respect)
My experiences from the past indicate that you don't want to have a PS that references from a cap.

Series regulators do tend to sound more coarse than shunt, but in this case I am not so sure anymore. I haven't been able in the past to build such a simple and linear series power supply.

Interesting! I've got a 200b, most heavily modded, and I also tried a different PSU, a fully discrete regulated DIY supply:

The σ22 Regulated Power Supply

The difference was negligable. I had stability issues with it using certain op amps, and with stable ones it sounded pretty much the same as my (upgraded) regular PSU. But my console has a seperate dirt ground and local PSU filtering caps for every single op amp.

I am a bit surprised that you didn't experience a change of sound wiht another power supply. It could well be that things are obscured by other things like common mode currents running through the gnd. Also the power supply is more complex than my new designed power supply.

The dirt gnd I assume you are feeding the gnd of the filtering caps (including the local PS ones) to a star point gnd? This would help, however the dirt is now transferred to the PS of the opamps: the purity of the PS for opamps should be seen as related to the local signal gnd. Also this can most likely provoke oscillation. Of course you can use local bypassing, but this means creating two different paths and effectively introducing a gnd loop voor higher frequencies.

Gnd currents, power lines, decoupling etcetra: it takes a lot to grasp the full picture. Also can the more subtle things one would perhaps never think of, be of more importance than at first thought.

I will write more about the problem of the power distribution at another time I guess.

Mounting a regulated power supply hard to a vibrating surface (like the housing of the standard power supply induced by the transformer) can kill sound in such way that diferences become much less audible.Mounting the PCB in rubber grommets and like-wise the transformer would be a means to reduce this influence.

Oh, and yes: looking again at the sch of the PS you used: I guess it will be slow, as the power Fets are driven by a high impedance source. Also it uses a zener, filtered however, but still...
Also I tested my regulated power supply AFTER I changed the raw power supply to a MUCH improved own version. Feeding the input of an amp (because that's what a regulated power supply is) with a sawtooth of 400mv top-top , compared to a sinusiodal 50 m.v. top-top makes a huge difference. (also on a more subtle level I won't even touch it is much more refined than a standard P.S)

I did some more tests today and the new PSU does indeed sound better. Much better actually. More clarity, punch, impact etc. Interesting.

Yes, the dirt ground goes to a star point. All op amps have local decoupling via 100nf ceramics and 10uf electrolytics on both rails going to this ground. In the original design all the dirt passes through the same ground as the signal.

There's a lot of discussion about this and other mods for the 200b at groupdiy.com

Thanks for the reply! :-)

Yet another Bulletin Board is too much for me (this is the first time ever I write about a project). I will probably go into greater detail about grounding issues, but first reaction: separating gnd from the elcaps and decoupling, and the signal is TOTALLY WRONG!! I will explain perhaps another time. (at least when not taking very special care of keeping induction to an absolute minimum)
This also will cause the easy oscillation.
I do have other ideas, better ideas, but I am not at that stage right now.
Ceramics are a total NO GO in audio. (Ouch! I will get burned for this)

The idea is to use bad (high resistance) ESR local electrolytic caps to provide isolation and stability, with the resistance of the caps countering inductive effects. Ceramics for off-band PSU filtering are common practice and recommended in most manufacturers data sheets... The advice I got in modding this console, including the dirt bus, come from people posting at groupdiy, some of them have designed some highly regarded consoles themselves.

But opinions differ, especially as far as local decoupling is concerned. For instance, Douglas Self recommends only local rail-to-rail decoupling, while others think the above mentioned approach is essential...

I'd be glad to hear more explanations from you, this is very interesting.

Hi living sounds: thanks for the reply!
As far as ceramics: I think it was about 25 years back I noticed that even a 22 pf as HF compensation in a 5534AN was sounding awful, so already then I learned that ceramic sounds awful.
Don't put a too great trust in datasheets: as said above the makers don't always know what they are doing, and with sound you can't trust them at all.
The gnd is a very interesting and very complex thing. Perhaps is an idea to use one of three (was it?) core's of the bus flat cable for return of the gnd of the local elcaps.
You should be aware however that all "internal affairs" gnd currents from the input strips have to flow through the signal gnd of the flat cable, so that's also not optimal.
O.K. dirty gnd: why is in dirty? What is the cause? You have to think about all these questions and really think in through to know what you are doing.

All R-C local power filters are forming a low impedance network from + and - to gnd.
First of all 1 m.v. of ripple at the output of the PS (power supply) causes about 1 m.a. gnd current....
So, indeed, we have a problem. So there is A LOT to say for using a separate gnd for the local elcaps.
But, I gave you the against argument also.
I understand the stance: use in audio band the mediocre elcaps, using a seperate gnd, and use local out of the audio band ceramics to the signal gnd to get HF stability...
Yes, I think this is a clever scheme, and with the standard PS it is probably the way to go. Try polysytrene instead of ceramic.
I have heard the influence of even very pure PS currents to gnd in high end, and indeed it should be avoided like the plague.
As the power supply configuration is a low Z thing, relative large current flow compared to signal flow in the gnd, and also it is 0% related to the audio.
Also problematic is that even out of audio band HF currents disturb the gnd, also affecting sound. (HF on transistors changes their curve)
So there is no easy answer. The thing that is however clear is that it is best to start with the source of the troubles, rather that fighting the symptoms: the power supply itself!
(more on that later)
At the end it is all about knowing what mechanisms are involved, thinking about the solututions and then testing and LISTENING to them, and trying to understand how what you hear relates to the physical electronic world. (well, that goes for me, other people perhaps just want the best recipe)

I am concerned of the loop between signal gnd and power gnd: it (the surface) should kept to a minimum.
Another way would be using a PCB over the total with of the mixing desk at the bottom, using connectorpins in a row (1/10 inch grid) and solder on the back wires, perhaps silver wire. Cut the ribbon cable in such a manner that each module has maximum lenght, and solder the end of the ribbon direct to the connector. (using seperate gnd for power and signal is still a choiche). The connector pins should be tactically placed, in order to get the PCB as high as possible, enhancing easy removing the modules.
I think that at the end would be my idea for improving the wiring. A lot of work, and less dependable: quite a price to pay.

Yesterday I have been working hard building a second regulated power supply to put in series with the first one, comparing two and a single back and forth several times, tuning the power supply, listen intensivle to all diffrent configurations, testing Vishay resistors in certain places, listening to the effect, comparing single verus dual PS again, etcetera.
I wasn't till late in next morning that I put myself to rest: it aks so much concentration, so much thinking is going on, that you try to get to a point where you can put it to rest.
So: it was a very interesting but very intense journey.
The dual PS configuration did sound very good, an awful lot of musical details, but it sounded like more suitable in a studio than on stage. Also I lacked enough voltage to give each regualator (that are connected in series) optimal Voltage. At the end I had to introduce Vishay, because the sound with 2 regs was a bit too thin, too grainy, and lacked dynamics.
The Vishay resistors worked their miracle, and the sound was becoming very acceptable. At a point I changed the value of a 100R degeneration R to 50R (Both Vishay) and it was clear that with 50R it sounded more normal, plain better.
At the end I listened again to the one reg solution, but now with Vishay: this is it! It sounds more simple, as in easy to understand for the ears, it has slight faults, but in general it sounds really very good.
So at the end I found myself again in 100% designer mode, including the stres, the very intensive listening, thinking, thinking and thinking. (1 CD is normally just enough or not enough to think after a serious modification) Pivotal in my thinking has been the bus system in the mixer, where the signals and the power are distributed, and the gnd currents. This might sound strange, but I approach a project as a whole, I try to have as much as possible an helicopter view during my work. it is possible that a new idea throws everything around, and designing is very much taking in account interaction of the parts and the over-all concept where everything sould work in concert. Also it helps to understand the limitations, and constantly there are decisions about the extend one wants to go in a project.
It was strange but also of course so natural to find myself in full designer mode over a power supply of a mixing desk, finding that the power stupply determined to an unbelievable extend the sound. (well, it was no suprise to me really, but each time you start you need confirmation)
Conclusion: what to do first before modding a mixing desk?? Modding the power supply!! And be sure: EVERY thing I changed could be heard.
You should treat a power supply as if it was a (high-end) power amplifier.

So at the end I am very happy with a new standard design for a series regulator. It's very linear (0 to 100KHz flat, and still good working to 1 MHz) very stable without any form of decoupling or compensation, short circuit safe, very simple, output R 0,1R low enough for many applications (Also the same linearity) and in series the PSSR goes from 60 dB (1000 times) to 120 dB (1,000,000 times) without any problem with stabilty again and same flat bandwith.

Today the master channel has been completed for the time being.
Again, this first stage we are still in is to bring the SC (SoundCraft) in a state from where we can departure for the second stage: deeper modifications.
I started with replacing the output IC's 5532AN for LM4562. Technically at least not a step back, and it has a low V offset, but also typical 10 n.a. input bias current, where the 5532 has typical 200 n.a.
I didn't remove any coupling caps yet: first want to hear the difference with a simple swap. As it is feeding direct the output, it is very easy to check for oscillations.
The surprise: I did oscillate! Back to the 5532AN: it oscillated too! It turned out that the new developped PS (Power Supply) did oscillate, not withstanding the fact that I checked many times...
Í fixed the problem after investigating a couple of ways in my simulator. This is offcourse all in the game with something new developped just one or two days ago and had not very extensive testing done.

O.K. listening to the LM4562: mwha, didn't knock me out of my shoes. Less PRAT than the 5532AN, but more refined. We have to wait and see till it breaks really in.
Second step: removing the output capacitors of said opamp, and replacing the opamp between fader and the output driver with again an LM4562 for a 5532AN. The capacitor after the fader has been removed, the much smaller I bias input of the LM4562 was really what made this possible. Also removed the output capacitor.
Listening: hmmm.... more refinement added, but a bit edgy sounding, still not as organic as with the old opamps, but it needs to break in, and... because we split the process over two stages, we don't have to delve too deep into the sound of the opamps yet, still we have to note what they did to the sound: this information can be used later.
Replacing the mixer (input) opamp 5532 for another LM4562, and removing the input cap. I don't like at all the input cap at that place. I might go for keeping the output cap from the output driver of the input strips: they isolate the strips in case of malfunction: one strip can't take down the whole mixer. Also perhaps it's not wrong to keep just a few capacitors in the signal chain at tactically chosen places. (This preserves something of the original design and character and we can try nice capacitors like ELNA SILMIC (was it? I ordered them) and others to tune the whole thing a bit.)
Remember that if only 1 strip delivers DC, the whole mixer goes out of whack. It also shortens very much faultfinding in the heath of the battle!
Listening: WOW!!! This is a REAL improvement!! Till now I was not impressed: compared to the mega change of the own designed power supply, the first steps seemed like babysteps.
I had soldered two very thin wires to gnd, and the output of the mixer, using a 10K in series to reduce the possibilty of oscillation: talking about it: at each step I control the output for oscillation.
It read a very stable 2.6 m.v. Switching channels on and off didn't make any difference.
I measured the output: differential (from pin 2 to pin 3) was measuring about 2 m.v. DC offset, good, but between pin 1 (gnd) and pin 2 or 3 about 18 m.v! NOT GOOD.
I decided that the best thing to do was to place the output caps again. So, not familiar with pro audio, I just stumbled on the ground compensated thing, and I didn't take the time to understand fully the topology of the output stage in differential form with some kind of adjustment of balance by a PCB pot. Best to leave that for the moment, and have a look at it when all parts are breaking in.
So: next step was to remove the cables of the AUX and SUB's and listen again: holy moly! Again oscillation!!
Again the power supply, so this time I tackeled it real good I think using a way that reminded me of HF compensation in Tube Poweramplifiers.
The listening was to get a new reference.
I now removed the AUX output channel, the SUB PCB's, and replaced the output elcaps in the output stage of the main channel, and the inverting opamp after the mixer: an AD823 for an TL027. I removed the output cap at the end, so it is now driving DC the fader. So: 4 elcaps out of the way per channel.
Listening: Very nice!! Where the LM started harsh, this opamp starts to break in from the soft side. It seems for now to compensate for strident sound of the LM. Of course we have to wait how it will sound after it has broken in more. Somehow the AD823 breaths quality: that is not to say that I would like it at the end: it could sound too velvet.
So we end up with all capacitors from the main channel removed, safe for 1 at the output, where it on the one hand makes the most sense: it is a very good safety build in, on the other hand the output has to work hard to drive the low impedance toplogy, that's a negative factor. (but I might chance the output stage anyway later in stage 2)
I am pondering about gnd compensation: do I want it or not? It seems not optimal to inject trash in the signalway. The nice thing is that this is stage 1, I don't have to solve this riddle now.
Over all the resolution has been much enhanced, it has certainly gained much good properties, but it is not yet to my liking, but again: that doesn't matter in this stage too much.
We want a solution with vastly reduced elcaps in the signal way. Tuning with other opamps with the same precision in V offset, and low I bias input will be done in stage 2, where also bipolar input opamps can be switched for FET opamps if necessary.
Tomorrow the input. After that I can let it break in, and in the mean time treat the other input channels and said other removed PCB's.
Monday I will receive 200 of those small buggers, ALPS switches. I plan to change them all in due time.
Still on the list: finding good cleaners for the inserts in my continent, canned pressured air for the faders, putting LED lights in the VU meters.
After all opamps have been replaced, and perhaps transistor change in the input amp, everything is working like a charm again that concludes stage 1.
It will be used immediatly for band and PA rehearsals, so it can be tested.

To start with stage 2 I will look for a 8 channel SC 200B: the 16 channel will stay at the rehearsal venue, and used for concerts I guess, and I can use the 8 channel for modifying in depth.
This will be more time ellapsed: it is about making contact with the design and looking for where it wants to go. Only if I have a really strong idea I will act on it. It could take weeks before something gets strong in my concious. Swapping opamps will not be such a deep process, but things like gnd modding, changing the output stage (or not), trying to eliminate the input capacitors of the input will only be done as I am sure that it will work, on a technical level but also on the level of sound. (thank god that I learned over the past 25 years to think first and act after it)

Again: the first stage is to get the mixer known, and to get it up to a good standard to start from: only then the funs begins!

Today I bought the magic trio of Kontakt Chemie and "Dust off" nr 67. The magic trio: WL, nr 61 and nr 60. It shoul deal with everything that has to do with cleaning the faders, switches and potmeters.
Listening today: not very impressed!
First of all be sure that comparded to improving the power supply swapping opamps is a minor event. Well, it plays out of course on total different parameters of the sound. At the end it is about all things together how someting sounds.
The sound is very organized, easy on the ears, but it does not sound open and dynamic and alive.
From what I hear there should be at the most 1 AD823 in the signal path, if at all.
The LM4563 sounds more HIFI, but at the same time more artificial, secondary. It makes much of a difference if the sounds like the music is represented, or if you really hear the energy of the music, loving it while lsitening to THE MUSIC, and not listening to a representation. I am much in favor of an energetic sound than a flawless but boring sound. At the end we want to make the mixer "a strong personality". This means for me also not complicated sounding, energetic, vibrant, powerfull or even thunderous, but at the same time with some refinement that keeps one to enjoy the mixer. It should stand out of all other mixers, having certain qualities that just can't be found in any other mixer.

About opamps: where do you think they get their better specs from? Better desing? Perhaps. More complicated circuits: likely!
So it is prefectly clear that though specs can be improved, it doesn't say it will sound better. On the contrary: I have found that there are million ways to do thing wrong, and only a few to do it right. Designing high end is for a large part avoiding a load of no-no's.

So what would be an effective way of getting the best sound at this point (of experience)? A much better discrete build power supply dedicated to the job: audio, and replacing the mixer opamp for an LM4563 and omitting the input capacitor.

In the mean time I have thought of the many ways of dealing with the main problem of this mixer, and probably all mixers: the bus system and the power distribution.
Something is getting more and more of a "good feeling" about it, and that is the plan to use a large PCB with hard wired bus system beneath the modules with short cables with connectors to the modules. Probably the best thing to do is to source ready made connectors with a bit of flatcable. This could be made almost (or entirely) like a drop in system!! This idea has to grow, but I already feel that this will be implemented in the future in some kind of way.

Another thing that popped up in my mind if there shouldn't be a byass resistor placed over the input chokes of the summing amps. I guess somwhere in the range of 1K to 10K.

By the way: For people who want to reproduce my results in the end, I will try to make this possible in the future. This means delivering the power supply or the parts, and for the rest give you the total instruction set how to modify your SoundCraft in such order that it becomes a thing of joy and value. This of course if I succeed getting to that point myself. It is now far, very far away from that point.

To be sure: I replaced said opamps without adding decoupling of the Power Supply. This is on purpose: testing with decoupling and the influence on the sound is another mini project, it is now not the right time to do so. It is all about tactics and not getting lost.

I replaced the to test input strips: I had still to replace the 4 times 100mu/10v. I replaced them by rubycon 100mu/16V.
Listening: WOW!! What a big difference!! Also the sound now seems quite well!

I cleaned all connectors at the back of the soundcraft with a three step approach and used compressed air in a can to clean all faders.
Three step approach: Kontakt 60, removes oxide, kontakt WD cleans and leaves no residue, kontakt 61: leaves a tin layer of protection and a bit of grease.
After all was done I listened: it sounded better!

Also I decided to use the group channels to judge the different opamps and modification: in this way I can compare in stereo subgroup 1 and 2 with 3 and four.
I removed all IC's in the subgroup and aux out channel and mouted IC sockets. (good quality) and put all IC's back in. Also changed all signal elcaps (Till now I only replaced the power elcaps) into (again) rubycon ZA same value, but 35V.

Don't even think about a full modification without a desoldering station!! You really need it to deliver a good job.

I got the impression that the new opamps are starting to break in, and it could well be that I change my opinion about the LM328. (and also have a more positive look of the LM4562).

This why it's important to have two stereo channels to compare.

O, and yes, I measured the length of the ribbon cable for the bus and powerdistribution: it is about 8 feet! (2.5 meters). More about that later.

Only after a big project like this is finished one knows how the optimal sequence of working would have been.
From what I know now:

FIRST bring it up to spec again:
1) clean all connectors at the back side.
2) replace ALL elcaps: use for instance the Panasonic FR series. Keep the same values of capacitance but preferably 1 range of voltage higher. And for the (local) power supply elcaps use 100 mu instead of 50 mu. Don't forget to give the power supply also a recap.
3) replace all switches that are flakey.
4) replace all ceramic capacitors for polystyrene.
5) If you are in the mood: replace the existing opamps in a socket.

Keep it powered up for two weeks, preferably use it.

So: now you have a mixer that is good working order. The transformation in sound will be huge.
This is the perfect base to start from for modifying. You need a reliable base to start with before modifying in order to judge what you are doing.

O.K. Today I removed 8 input modules, so I can work on it, and the console can be remained powered up all time while I am working on them.
It is important to for me to have as much as possible the console powered up: things need to break in.
The main section including the aux channel output and the group channels are completely recapped (that is elcaps), and have all IC sockets: I replace all original IC's one for one back in the sockets.
There are no ceramic capacitors in the signal chain. (replaced them with the rather expensive polystyrene)

Two subgroup channels I replaced the 5532AN for LM4562, and bypassed the input capcitor, for the rest the modules are the same.
This is a great way to compare the channels and compare it again after 24 hours or more breaking in, and keeping to follow the process of breaking in.
Ah! I love it!! It is always so good to have an easy and reliable setting to judge on components!

You must understand that the LM4562 is up to stiff competition! First of all the 5532AN is not a slouch, but that's not the point I want to make: the 5532AN has 3 decades of breaking in time! My estimation is that two weeks, but perhaps much earlier, will be enough to make a final judgement.

EDIT: OOPS I MADE A STUPID MISTAKE!! I wrote something about the difference between the 5532AN and the LM4562: I had the internal switches in the wrong position so that the LM sounded softer!! I totally f*cked up stating that it had nothing to do with measurable levels.
I will do a first assesment of the difference of the LM and the 5534 soon! :-)
Sorry for the people who I gave the wrong information.