Top end mic pre - suggest a project

Gents,

I'm updating my mic collection with some great mics like Neumanns and Schoeps. I need a really transparent preamp to match those. I was thinking about Millennia type 8-channels pre. Obviously, 4 k$ for Millennia is quite steep price.

I believe that there're not that many options available. I prefer discrete transformless design. What do we have here? SSL9k? John Hardy style pre (although it is with transformers)? C84 from 7th circle?

What would you guys recommend as a DIY mic pre project that performs as close to the "wire with gain" as possible?

Something like the THAT chip set is probably quite good.

JR

JR, thanks for chiming in.

Frankly, I don't like THATs much. They are ok for everyday job, but to my ears they can't be compared to the real high-end stuff. (Although I know they are used by Grace in their preamps. Hmmm...)

I have to recommend the 500-series format.

G-Sun said:

I have to recommend the 500-series format.

Click to expand...

500 format is cool, but it's not a preamp per se. Millennia has 500 format modules as well btw.

Besides, from the DIY point of view, it would be cheaper to make an enclosure and PSU than to buy a 500 frame ($400 and upward locally).

Check the Innersonix mic pres. THAT based, but real good. Can be midi controlled.

The Eden's have done very well in recent shootouts...

I built 8 channels of THAT 1512 and 8 more with AD797 and couldn't be happier.

Cleaner and more detailed without being harsh than RME and YAMAHA PM5D. Made blind tests with recording engineers and audiophiles and were quite surprised about results.

IMHO Neve is one of the greatest but not "wire with gain" as OP asked for.

I am working on some preamps with the INA103 from TI/burr brown. 48volt swing on this means more headroom... I'll keep you posted.  Check the data sheet will do up to a 50V swing...

I don't think there is any ready to go DIY project of a discreet transformerless design. Might be wrong, but don't remember having seen one.

Forssell used to sell a very nice one for U$250,00 for a board, but it seems it is not available anymore.

Also there was the Borbely one, but no schematics for these, unless you buy the thing. Sad thing is Borbely is out of business.

Ilya said:

I believe that there're not that many options available. I prefer discrete transformless design. What do we have here? SSL9k? John Hardy style pre (although it is with transformers)? C84 from 7th circle?

Click to expand...

If 'accurate' is what you want, the ultimate is Graeme Cohen's 1984 circuit.

Millenia Media admit to using this topology and I'm almost certain so do Earthworks.  The Earthworks is the quietest mike preamp you can buy but you will need a LOT of work to even get to HV3 standard in real life.  Layout, decoupling & earthing assumes HUGE importance if you are after SOTA.

The Seventh Circle C84 is their interpretation of Cohen.  I have some quibbles but it is a well thought out & tested product with excellent parts & performance.  For 'accurate', I'd suggest you get one of these and also their excellent THAT based T15.

You can then decide for yourself if THAT is accurate enough.  A lot depends on your mikes.  Condensors, even the best ones, are less critical than eg passive ribbons like Coles 4038.

Eden is sorta souped up THAT and probably the best of the single chip preamps.  But it doesn't have the comprehensive RFI protection in the 7th Circle stuff.  IMHO, this is one of the things that separates the toys from 'professional'.

I don't recommend any of the other common DIY Cohen copies.  None of them come even close to good THAT performance .. let alone SOTA Cohen.

If you have a 4038, I'd recommend making a dedicated ribbon special without P48V like Millenia.  I describe such a beast in my Yahoo MicBuilders Files converted from DMP3 available for less than $100 on eBay 

The Cohen circuit was made public around 1984, if memory serves me correctly. Both Paul Buffs Transamp & the Neotek Series III console preamp were out in 1974-75. The Neotek used 2 transistors on each side of the balanced input for lower noise. Cohens circuit used the same transistors but only one to a side. The Transamp used an overly complex configuration.
The Neotek pre's were noted for being the quietest & most transparent transformerless mike pre's.

I new Paul Buff in a professional relationship & later became Neoteks first dealer; when I was in business in Nashville.

The transistors used by Neotek, Buff & Cohen were Motorola MPS 4355; they have been out of production for many years. I have come up with improved substitutes for them.

Bill Wilson said:

The transistors used by Neotek, Buff & Cohen were Motorola MPS 4355; they have been out of production for many years. I have come up with improved substitutes for them.

Click to expand...

What would you use today, Bill?

ricardo said:

Ilya said:

I believe that there're not that many options available. I prefer discrete transformless design. What do we have here? SSL9k? John Hardy style pre (although it is with transformers)? C84 from 7th circle?

Click to expand...

If 'accurate' is what you want, the ultimate is Graeme Cohen's 1984 circuit.

Millenia Media admit to using this topology and I'm almost certain so do Earthworks.  The Earthworks is the quietest mike preamp you can buy but you will need a LOT of work to even get to HV3 standard in real life.  Layout, decoupling & earthing assumes HUGE importance if you are after SOTA

Click to expand...

I just want to chime in with another opinion and say that I really came to dislike the Millennia, having owned 9 channels for many years. It's definitely a taste thing but I would not describe them as neutral, to my ear and from memory there was a bright, inharmonic glare or hash across everything regardless of mic.

As soon as I heard the GML mic pre that was my benchmark for clean and I own a 4 channel now.  The Hardys are very nice too.

I'm not sure where you are based but borrow or rent some of what you are considering, it's truly the only way to know.  The Millennia is a perfect example here - Ricardo loves it, I don't.  What will you think?

Cheers,
Ruairi

you never know what will sound good with your mics til you plug them into something,

hard to say "this amp will be perfect with this mic" before you build it,

In answer to Ricardos question as to what transistor to use in place of the 4355's; 2n5087's work quite well in that circuit.
It is  interesting that neither the MPS 4355 or 2N5087's are listed as special low noise devices ie 2n4401 & 4403.

The proof is in the implementation. I made a number of mike pre's using the Neotek circuit as replacements in MCI 416 consoles.
The customers comments were; it is the quietest preamp I've ever used. I still have some of these that will be for sale shortly.

Ilya please check your private messages.

Wow, that's a lot of oppinions 

I think that I will have to make some listening test to decide what is best for me. Unfortunately, it's almost impossible to get hold of expensive pres here without buying them first (nobody stocks them). Some of my friends have Millennias, but I'm not aware of any Hardy, Great Rivers etc.

My plan is to make a sort of universal modular framework, much like the 7th Circle does it. But in the 1U enclosure with external PSU. This will allow me to build several different pres and compare them easily. A lot of work though...

Bill, speaking of 2n5087, is there any matching procedure required for those?

Bill Wilson said:

The Cohen circuit was made public around 1984, if memory serves me correctly. Both Paul Buffs Transamp & the Neotek Series III console preamp were out in 1974-75. The Neotek used 2 transistors on each side of the balanced input for lower noise. Cohens circuit used the same transistors but only one to a side. The Transamp used an overly complex configuration.
The Neotek pre's were noted for being the quietest & most transparent transformerless mike pre's.

I new Paul Buff in a professional relationship & later became Neoteks first dealer; when I was in business in Nashville.

The transistors used by Neotek, Buff & Cohen were Motorola MPS 4355; they have been out of production for many years. I have come up with improved substitutes for them.

Click to expand...

I was not aware that Cohen was credited with this topology until several years ago.  I had seen it around well before he published his AES paper. I wrote about the Transamp in my 1980 Console Performance limits article for RE/P and used that circuit topology in a phono preamp kit I published in Popular Electronics during the early '80s.

IIRC the 4355 was a medium power transistor, not really intended for low noise use, but naturally had low Rbb for the elevated current handling. I believe Paul screened them and selected out devices with higher (1/F process) noise He probably used the same transistors in his VCAs (culls?).  By then I had found some sweet very low noise transistors designed by a small Japanese company (bought by ROHM) for moving coil head amps (2sb737/2sd786 both a few dB < 1nV/rt Hz). I used the 737s in all my mic preamps back then. They went obsolete over ten years ago and there are newer parts that are OK (the moving coil parts were actually a lower Rbb than needed for 150-200 ohm mics).

To get back on topic, to make a transparent mic preamp "today" it is no longer necessary to roll your own from discrete transistors and op amps. There are multiple off the shelf canned solutions that are respectable. THAT corp, TI, etc. Of course execution matters so details like how you handle the gain leg (cap?) matters, but the modern parts have good app notes and work very well when properly used. 

It is rare that somebody asks for a clean preamp (i like that). It might be interesting to have a null shootout where we setup a good chip preamp as a reference, and subtract that from the sundry high end preamps to see how much they vary from straight wire with gain.

Sounds like fun.

JR   

Bill Wilson said:

In answer to Ricardos question as to what transistor to use in place of the 4355's; 2n5087's work quite well in that circuit.
It is  interesting that neither the MPS 4355 or 2N5087's are listed as special low noise devices ie 2n4401 & 4403.

Click to expand...

IIRC the 4355 was a general purpose medium power device. The 5087 (5089) IIRC was actually called low noise, but in my bench tests back in the day they did not deliver the goods at 150-200 ohm sources. The 4401/4403 was another general purpose transistor, but in the classic "Low Noise Electronic Design" by Motchenbacher and Fitchen c. 1973) book they thoroughly investigated using the 4403 for low noise circuits, paralleling devices for even lower noise voltage. Note PNP and NPN processes are slightly different so I suspect the npn 4401 might not be as good.

There were other transistors back then called low noise (4250) but again not very respectable for nominal mic source impedance. About the only device US makers called low noise that was low noise on my bench was the LM394 logging pair, but expensive and at 1nV rt hz not as quiet as the Japanese head amp transistors.  The modern chop sets are down in the 1nV ball park so quite good and not improved by extra effort. 

FWIW the 1973 M&F low noise design book also showed the discrete transistor in front of an op amp topology, but just showed it single legged, not with two of them balanced and feeding a differential, so arguably not a complete "Cohen" sighting, but adequately predictive IMO. 

JR 

The proof is in the implementation. I made a number of mike pre's using the Neotek circuit as replacements in MCI 416 consoles.
The customers comments were; it is the quietest preamp I've ever used. I still have some of these that will be for sale shortly.

Ilya please check your private messages.

Click to expand...

JohnRoberts said:

I was not aware that Cohen was credited with this topology until several years ago.  I had seen it around well before he published his AES paper.

Click to expand...

The difference between Cohen and other stuff using that topology is 0.7nV/rtHz.

IMHO, it is perhaps the only sensible topology for a practical mike preamp with <1nV/rtHz.  If we define 'Cohen' as something using that topology with <1nV/rtHz, that rules out practically all the DIY copies.  Once you add P48V, protection & RFI stuff, you are struggling to achieve even 1nV/rtHz .. which highlights Millenia'a achievement with this topology.

Circuit & topology are only one thing you have to get right for <1nV/rtHz.  The layout, decoupling, earthing & construction have equal, if not greater importance.  That's why my recommendation for the 7th Circle stuff.  Cohen 1984 is their C84 .. geddit?

I have a small preference for Eden over THAT in certain applications (which is reversed for other apps) but we really need to wait for Rochey's own box & PSU for its full potential.


I'm only talking about noise though I have my own strong opinions about clarity & definition bla bla for various circuits.  Noise is something anyone with a bit of nounce can reliably test for himself without an AP.  My MicBuilders stuff on modifying DMP3 has some details.[.]

To put all this stuff into perspective ... the normal 150-200R mikes can achieve their full noise potential with the 1.6nV/rtHz of a good THAT or IN163 (Eden) implementation.

Supa LN transistors like 2sb737 et al in Cohen will show little practical noise improvement over THAT for most mikes.  But if you have a low output ribbon like 4038 .. especially the 30-50R versions, you will be looking for every dB of better noise performance.  Even then, you are talking about at best, 1.5dB between good implementations of both.

Removing P48V and other stuff is likely to be more rewarding for a dedicated ribbon preamp.

This Millenium, I'm involved with experimental mike stuff that has some benefits from the 0.6nV/rtHz of the Earthworks.  I'd dearly love to know how Earthworks do it with P48, full protection & RFI stuff.  Their stuff is all encapsulated.  They've only recently started to claim their stuff is all discrete  ???

[.] If you read this carefully, you'll see some possible reasons for audible differences between preamps .. audible as in reliably detected in Double Blind Listening Tests.