DIY Vintage Neve or SSL Summing Mixer for sound console

i'm not fond of Carnhill's ( even less since they eaten up Sowters...). Anyway there is other options: check for UTM industry (cheaper than Uk products for mainlands citizen and closer to your country too) and if you want even higher quality ( but of course way pricer) Moby's products.
There is other options too... but not cheaper than Moby's offer and on par with quality level, i'll include in pm to you.
 
I agree re: Carnhill's treatment of Sowter and its employees / equipment. Is your opinion on their transformers based on objective knowledge or personal opinion? I have a small stash of original Marinair (I ought to sell them) - as well as some Carnhill replicas. I ought to run them through the AP and look at a THD sweep 20-20K with different levels and DC...

There were rumours that the quality changed when AMS-Neve stopped using them (they trademarked 'Marinair' - decades after the firm shut down...pretty cheesy, no?). I thought at one point that the distortion increased and haven't used them since. I suggested them here as I am not entirely convinced the difference between them and the OG transformers will be that noticeable, as the original Neve circuits will have huge THD whatever way you look at it. If you're looking to recreate the SE Neve circuits, it's not as if you have an allergy to distortion... There are hundreds of 500-series and rackmount units that have Carnhill and people seem to like them.
 
I hear a few citing the custom Neve consoles for Air Studio, which were highly customized. Some of the VCAs, input, output and interstate transformers were wound to specific needs and were given unpublished part numbers. These transformers are where most of the magic happens in these consoles. The EQ stacks were also very specific with non-standard parameters, mainly speced by G Martin and G Emerick. Much of the summing was done with inverting current mirrors and the bulk of 5532/34 used was run with high Vcc right on the edge of max spec (Usually +/- 22V). The outputs were loaded with current biasing to reduce 3rd order HD. There are a TON of nasty electrolytic caps in the signal path, which would put off most snobs looking at the specific locations they were used at. Coupling transformers dont like the slightest amount of.DC on their primaries (unless they're designed for single ended stages), so the only way to avoid this was with coupling caps. Real estate was tight in most of these channel strips and caps couldn't be that big. The other place where some trickery was employed is in the input stage loading after the input transformer. They almost never used resistor padding on the input side and usually

The BA182/205/211 EQ modules were an early design on the 1073 which could be configured in many ways with just about any frequencies desired. Sowter transformers sounded much warmer than the Carnhills on these channel strips. I greatly prefer the Sowter sound, especially in the lower mids.

There's nothing wrong with properly implemented and loaded NE5532/34 chips. Its when they get used incorrectly, they become an issue in the signal path. I've heard some phenomenal stuff done with these op amps. Paired up with decent iron in the signal path (I prefer Jensen), you can get some amazing results.

If you look up some of the recordings done on the Sound City Neve console, these highlight the clarity which could be achieved with this console even when the levels are pushed well beyond unity. SSL is more clinical sounding than Neve. API is a bit more veiled and laid back, but mainly when pushed harder. Trident is definitely very British sounding, especially on the top end. The Neve stuff has the best overall flavor when comparing 70s and 80s available technology. Trident is a close second. SSL is more for audiophile snobs. Soundcraft makes some nice stuff too if you're on a budget and sounds alot like Neve in some ways. I still use an old Ghost for analog transfers.
 
  • Like
Reactions: tmuikku
Well in my case it is not whose transformers are better, but where to get them cheap and available with worldwide delivery. I have already been advised several places where they can be bought and I will definitely visit them, but what I can say for sure is that I will remove most of the transformers from the schemes of the channel strip.
 
you could leave them in, but do they make so much sense that they're all over the channel. For example. There is a preamp at the input of a channel strip. It has two transformers. They are very important in this element of the channel strip for impedance matching and decoupling. But after the preamp comes the compressor, which also has two transformers. And I will try to modify this circuit to the form without transformers, because they are not really needed there and you can do with capacitor or electronic matching circuit.
I can say for sure is that I will remove most of the transformers from the schemes of the channel strip.
It will be much cheaper than ordering and waiting for LL5402 for example.
 
Forget about lundahl transformers if you are looking for flavor: they are the most transparent sounding you could buy. To the point most people using them without knowing that facts are disapointed they don't sound... and do not understand it's a 'tour de force' to be able to do transparent transformers!

It's not that easy to think you are going to have same thing soundwise replacing transformers by electronic stage, 1176 is another example. The transfo impart some character to the unit and without them the circuit does not sound the same at all: in a studio i worked there was a Groove Tube unit cloning the compression stage but with a ba283am 'clone' gain make up stage ( 1073 driver stage). The unit looked flashy ( all chrome frontplate) and sounded 'nice' in direct comparison to the original 1176 and Purple 77 above it in the rack but it didn't see much use because it sounded much more 'mellow' and this is not what to be expected when using a 1176 on voice... in fact i've seen it used only for bass. The aggressiveness of original don't come only from the gain reduction cell and time constants but from the transfo too which kind of emphasis midrange.

But don't take my words for granted, try by yourself and come to your own conclusion.

From what you describe i would not focus on cloning some reference to make a signal path, rather design your own circuit. I mean, if you offered some 1176 on each chanel I would not buy your end product.
Why? Because i don't want a 1176 on each signal! I find it crucial for voice or sometimes snare but would never use it for a kick drum or over OH on strings section. Much too colored and too fast time constant...
Something with less character would see more use on a day to day basis over more source.

This is the reason you see inserts on any serious consoles and rarely a dedicated compressor section on mid budget professional desks.

The main point you should focus on imho is the eq. This is what makes people love or hate consoles from what i've seen.

And here again it won't be a one size fit all answers: when tracking you can be in need of 'restauration' or 'surgical' treatment tools hence a parametric with multiple transparent bands can be a must. But it require knowledge and time to setup. Time is lacking in sessions. Always so you can have the best eq icircuit in the world but if it takes 15mn to set up nobody will use it.
EG: that's the reason 4000E eq are often prefered over G or Vseries: being constant gain they often sound fatter at the turn of only one knob rather than the need to setup Q on the constant Q design implemented on G and V which are more precise/accurate but require further tweaking time.

This is for recording tracking, at mixing stage it's different case: if the tracking was done correctly then there is need to add 'color' or vibe true the eq hence a parametric is not needed but a much softer treatment is required. Some eq like the ones found on older desk ( Helios, Mci, Api,..) can then be an advantage...

There is few tools that are good at everything...
 
I agree re: Carnhill's treatment of Sowter and its employees / equipment.

Such a shame all this knowledge is now lost: remind me of Roy Baty's last sentences in Blade Runner 'all this memory will be lost like tears in the rain'.

Is your opinion on their transformers based on objective knowledge or personal opinion?

Both.
Without going too much on details on the objective side of things there is differences in constructions between originals Marinair/St Yves and most 'clones' even from renowed brands.

That do not say they don't sound close enough or bad but from a subjective side of view but few sound as original to my ears in side to side comparison between them and originals.

In fact i just discovered some which comes close to the better clones ( which were unobtainable to buy by themself) i've heard to date ( not worst than differences between original for lo1166). After 20 years...
have been using Carnhill's and Sowter meanwhile ( 8751).

I have a small stash of original Marinair (I ought to sell them) - as well as some Carnhill replicas. I ought to run them through the AP and look at a THD sweep 20-20K with different levels and DC...

It might reveals things. First between originals, they were differences, then from Carnhill's you'll find others differences.

There were rumours that the quality changed when AMS-Neve stopped using them (they trademarked 'Marinair' - decades after the firm shut down...pretty cheesy, no?).

And when they used them i've heard from technicians taking care of the console that the ones in the 88r's 1073 modules and some highly priced rack units were differents than the ones you could buy as regular customer from Carnhill.
Money, money, money... Neve's name and everything surrounding him is a money maker as gear used by the beatles... makes me grind.

I thought at one point that the distortion increased and haven't used them since. I suggested them here as I am not entirely convinced the difference between them and the OG transformers will be that noticeable, as the original Neve circuits will have huge THD whatever way you look at it. If you're looking to recreate the SE Neve circuits, it's not as if you have an allergy to distortion... There are hundreds of 500-series and rackmount units that have Carnhill and people seem to like them.

Yes. But how many had real ones to compare to? I mean there wasn't so many original produced overall and people are looking for a reassuring name on gear.
And i'm not really different about this except i prefer to build them with component closer to original rather than them looking alike :Behringer clones anyone?.

Talking about Behringer and back to 1176 i would not bother to build one nowadays as for the use i have of them their clone kt-76 is more than enough to me and i could'nt ever build one for such a low price...
Same could be said about their analog synth.
 
Last edited:
The difference between the Carnhill stock TFs and the AMS-Neve ones is supposed to be that the stock Carnhill were reverse-engineered from Marinair, whereas the AMS-Neve ones were built off the back of tech data supplied to Carnhill, which they supposedly signed an NDA for. As to how relevant this is, I do not know (reverse-engineering transformers isn't rocket science and someone with the right qualifications will be able to make a replica that's 100% if they can get the right lam material).

All of ^this plays into the whole mythology that keeps AMS-Neve in business (good for them - their head designer knows what he's doing and doesn't need to copy old kit). Let's say there are subtle variations between the types. Are they actually audible or measurable?
 
  • Like
Reactions: krivium
I hear a few citing the custom Neve consoles for Air Studio, which were highly customized. Some of the VCAs, input, output and interstate transformers were wound to specific needs and were given unpublished part numbers. These transformers are where most of the magic happens in these consoles. The EQ stacks were also very specific with non-standard parameters, mainly speced by G Martin and G Emerick.

Yes. I think i've read those toroidal transformers were the last one Rupert designed for Neve products.
Much of the summing was done with inverting current mirrors and the bulk of 5532/34 used was run with high Vcc right on the edge of max spec (Usually +/- 22V). The outputs were loaded with current biasing to reduce 3rd order HD. There are a TON of nasty electrolytic caps in the signal path, which would put off most snobs looking at the specific locations they were used at. Coupling transformers dont like the slightest amount of.DC on their primaries (unless they're designed for single ended stages), so the only way to avoid this was with coupling caps. Real estate was tight in most of these channel strips and caps couldn't be that big. The other place where some trickery was employed is in the input stage loading after the input transformer.
They almost never used resistor padding on the input side and usually

Yes it's often put a smile on my face when i see audiophile chasing for electrolitycs because of their supposed nasty's on sound. If they said they tryed to increase long term reliability i wouldn't smile the same though.

Loading on input transformers is what RN was about at this step of his journey. And what he did too in his next design: the Focusrite ISA110 see a specific loading an a taylored zobel to create a wide bell resonance at 24khz on the 1538 Lundahl input transformers which if you read datasheet should not require one.

This is one of the stage trickery which gives Focusrite unit their clarity and sense of 'air' making them a favourite for close miking voices. The other one being the introduction of the tertiary feedback on the output transfo which linearize the output stage.
It's the most noticeable difference to the 'pumpkin' design from Montserrat and subsequent Neve models: i doubt he redesigned the transformers for other models, he probably used the Montserrat circuit as a 'prototype' and didn't include the input transformers 'trickery' in the 'standard' models coming next.

He resused the same topology ( and transfo) but simplified in Amek 9098 output circuitry including the transistor buffer stage after the 5534.

What makes me feels he didn't redesign for each generation is from the Amek line: the 'Medici' mastering eq is the same core circuit than the one used in the 9098 and Amek rack derivative: they all use a toroidal on output which is the same pumpkin as in the Focusrite consoles. See attached pic.

That said about non gapped transfo and dc, well he did differently in RND: the 'silk' and 'texture' knob introduce dc to the used transformer primary to gives an effect on bass signal by saturating core and there is change on the amount and frequency(?) of feedback around the transfo. I wasnt' able to confirm it's an ungapped core but i'm pretty sure it is.
On the last units there is a silk 'red' and 'blue' function and i'm pretty sure those are both trickery sperated one involving only feedback on hf, the other core saruration.


The BA182/205/211 EQ modules were an early design on the 1073 which could be configured in many ways with just about any frequencies desired. Sowter transformers sounded much warmer than the Carnhills on these channel strips. I greatly prefer the Sowter sound, especially in the lower mids.

There's nothing wrong with properly implemented and loaded NE5532/34 chips. Its when they get used incorrectly, they become an issue in the signal path. I've heard some phenomenal stuff done with these op amps. Paired up with decent iron in the signal path (I prefer Jensen), you can get some amazing results.

Totally agree.

If you look up some of the recordings done on the Sound City Neve console, these highlight the clarity which could be achieved with this console even when the levels are pushed well beyond unity. SSL is more clinical sounding than Neve. API is a bit more veiled and laid back, but mainly when pushed harder. Trident is definitely very British sounding, especially on the top end.

I don't know what British sound is. Really there is so different type of sounds from 70's/80's UK brands i don't get the point.

I don't agree either about Api sound, the one i've heard were usually way 'punchier' and aggressive than other consoles. Of course if used with all the transformers in the signal path (input, chanel output, summing stage, line driver, to reel to reel to line input then another round of at leasr 6 transfo...) then they start to mess up sound in the hard way imho.

The Neve stuff has the best overall flavor when comparing 70s and 80s available technology. Trident is a close second. SSL is more for audiophile snobs. Soundcraft makes some nice stuff too if you're on a budget and sounds alot like Neve in some ways. I still use an old Ghost for analog transfers.

I never felt SSL to be audiophile at all. The 4000 sound really metalic in the mid to me and doesn't go this high or low in freq range. Not a bad thing but not audiophile at all. The 9000 maybe. But to be blunt i often told myself better mix otb whith them. They doesn't fit my preference as the overall sound.
Not an isuue i have anyway as i doesn't have access to one anymore. 🙂
 

Attachments

  • 5.-Amek-Medici-The-Equalizer-Stereo-Mastering-EQ-Internals_1792x1792.jpg
    5.-Amek-Medici-The-Equalizer-Stereo-Mastering-EQ-Internals_1792x1792.jpg
    489.4 KB · Views: 28
Last edited:
From what you describe i would not focus on cloning some reference to make a signal path, rather design your own circuit. I mean, if you offered some 1176 on each chanel I would not buy your end product.
Why? Because i don't want a 1176 on each signal! I find it crucial for voice or sometimes snare but would never use it for a kick drum or over OH on strings section. Much too colored and too fast time constant...
Something with less character would see more use on a day to day basis over more source.

This is the reason you see inserts on any serious consoles and rarely a dedicated compressor section on mid budget professional desks.

The main point you should focus on imho is the eq. This is what makes people love or hate consoles from what i've seen.

And here again it won't be a one size fit all answers: when tracking you can be in need of 'restauration' or 'surgical' treatment tools hence a parametric with multiple transparent bands can be a must. But it require knowledge and time to setup. Time is lacking in sessions. Always so you can have the best eq icircuit in the world but if it takes 15mn to set up nobody will use it.
EG: that's the reason 4000E eq are often prefered over G or Vseries: being constant gain they often sound fatter at the turn of only one knob rather than the need to setup Q on the constant Q design implemented on G and V which are more precise/accurate but require further tweaking time.

This is for recording tracking, at mixing stage it's different case: if the tracking was done correctly then there is need to add 'color' or vibe true the eq hence a parametric is not needed but a much softer treatment is required. Some eq like the ones found on older desk ( Helios, Mci, Api,..) can then be an advantage...

EXACTLY!
It's nice to have a musical EQ when I can use it - but I NEED a precise EQ by hand when some problems need correction.
I always was happy with Allen & Heath EQs and never got warm with Soundcraft ones. These big A&H live consoles are underrated, always liked to work on these, so easy to get what I needed. You can get for sure the exact same response out of both - I just prefered how an A&H reacts to my knob turning.

I would like an analog desk with compression in every channel. We put compression in most channels in digital mixing - I need to do the same when I want to stay analog. But it needs to be flexible compression - and the one fits all part is pretty hard to get nailed.


So, I will stick with Console 1 and choose between SSL, Neve, API and a few others with a few clicks ... 🤓
 
Forget about lundahl transformers if you are looking for flavor: they are the most transparent sounding you could buy. To the point most people using them without knowing that facts are disapointed they don't sound... and do not understand it's a 'tour de force' to be able to do transparent transformers!

It's not that easy to think you are going to have same thing soundwise replacing transformers by electronic stage, 1176 is another example. The transfo impart some character to the unit and without them the circuit does not sound the same at all: in a studio i worked there was a Groove Tube unit cloning the compression stage but with a ba283am 'clone' gain make up stage ( 1073 driver stage). The unit looked flashy ( all chrome frontplate) and sounded 'nice' in direct comparison to the original 1176 and Purple 77 above it in the rack but it didn't see much use because it sounded much more 'mellow' and this is not what to be expected when using a 1176 on voice... in fact i've seen it used only for bass. The aggressiveness of original don't come only from the gain reduction cell and time constants but from the transfo too which kind of emphasis midrange.

But don't take my words for granted, try by yourself and come to your own conclusion.

From what you describe i would not focus on cloning some reference to make a signal path, rather design your own circuit. I mean, if you offered some 1176 on each chanel I would not buy your end product.
Why? Because i don't want a 1176 on each signal! I find it crucial for voice or sometimes snare but would never use it for a kick drum or over OH on strings section. Much too colored and too fast time constant...
Something with less character would see more use on a day to day basis over more source.

This is the reason you see inserts on any serious consoles and rarely a dedicated compressor section on mid budget professional desks.

The main point you should focus on imho is the eq. This is what makes people love or hate consoles from what i've seen.

And here again it won't be a one size fit all answers: when tracking you can be in need of 'restauration' or 'surgical' treatment tools hence a parametric with multiple transparent bands can be a must. But it require knowledge and time to setup. Time is lacking in sessions. Always so you can have the best eq icircuit in the world but if it takes 15mn to set up nobody will use it.
EG: that's the reason 4000E eq are often prefered over G or Vseries: being constant gain they often sound fatter at the turn of only one knob rather than the need to setup Q on the constant Q design implemented on G and V which are more precise/accurate but require further tweaking time.

This is for recording tracking, at mixing stage it's different case: if the tracking was done correctly then there is need to add 'color' or vibe true the eq hence a parametric is not needed but a much softer treatment is required. Some eq like the ones found on older desk ( Helios, Mci, Api,..) can then be an advantage...

There is few tools that are good at everything...
The 1176 in my vision is not the only compressor that will be in the channel strip. About using it is a user question. I want to do the routing so that after the pre-amp and its EQ, the person himself assembles the chain from the built-in instruments provided to him. About equalizers. I have two or three of them planned per channel. The first is Neve 1081. The second Calrec Eq. The third one will most likely be just a colorizing one like the Pultec EQ. Basically of course the console is intended as a box with all the instruments. I was also thinking that maybe we should change the scheme and not put all the devices on the channel strip, but make a more interesting routing. So that a person could choose what he wants from the processing on each channel. That would leave each channel with just the Neve 1081 and a built-in compressor that can be turned off. The bands would get an SSL compressor and a Neumann v475. It would make the design much cheaper, but I have no idea how to pull that off. Let's say I build in an 1176 or LA-3A compressor. And the user will need it on 3-4 channels. I'll have to make 56 of these compressors so they can be used anywhere. What do you think about it? Your opinion is welcome)
 
@krivium The SSLs I referenced to are the later models, being more for audiophool "nerds". It may be a visual aesthetic thing, but I never warmed up to them for the purpose of making music. When Massenburg did his EQs, I thought those were the least intrusive sounding filters I've heard, but they weren't always that musical. API sounds too bloated for my tastes. Great for some genres but not for delicate work. A+H has always had good EQs all around. Even the later DLive consoles had great sounding, effective EQ being a newer gen digital design. I prefer mixing on A+H over Digico these days. A+H sounds better to my ears, but is definitely a bit brighter sounding overall. They sound closest to a traditional analog console with typical facilities.

On a side note, I have to mention the newer Soundcraft digital rack mixers. They're very underrated and have a huge feature set. I purchased a UI24R a few years ago, which can do multitrack recording simultaneously on all channels including Aux. The dynamics can even be side chained and despite being only 48k converters, it sounds very good, considering the $1400 price tag. There's even an html based control which allows remote use with any common web browser on just about any mobile device. The lag is minimal, including the Midi control implementation.

To the OP, regarding channel strip dynamics, the compression is very specific and limiting on most older Neve. The time constants are very specific and don't really work with many sources. I'd always prefer a fet based comp in conjunction with heavy biased tape saturation for tracking drums. I've never regarded typical channel strip compression to be useful for most applications. In some cases, there is no provision for pre/post EQ to dynamics selection, so this further limits use. Outboard dynamics are always a better option IMO unless used for specific designated sources with limited time constant range.
 
  • Like
Reactions: IamJF
The 4000 uses a pair of supermatched LM394 transistors in front of an uncompensated NE5534. This is the way to go if you have to sum a lot of channels (more than 32 or so).

Th 9000 uses a balanced arrangement with high performance op amps (analog devices if I recall correctly). And DC servos instead of coupling caps everywhere.

The old discrete Neve summing amps are good for quite a few channels, but they have a sound and need +24V power.

The devil is very much in the details with a console. Getting grounding, decoupling etc. right is very much a non-trivial task. Op amps are often very demanding in terms of implementation and power supply requirements. 5532/5534s are great, but depending on the manufacturer and the production date they can behave quite differently. I prefer the older Signetics type. In many places you can get away with 4556 op amps (again, there are different ones from different times and manufacturers) and they are often easier to get to work properly.

As for the sound of mixer, I experimented a lot, and in the end I always preferred a neutral, unhyped signal path. This is not the same as "sterile", "boring" or "super-detailed", which are things that tend to happen if you do something wrong. You can get plenty of color from outboard inserted into the channels and the master bus when and if you need it.
 
  • Like
Reactions: krivium