Passive summing with transformer based makeup gain

Is that possible ? Using a high ratio transformer ?

well something has to drive that transformer...

Well yes and no..  transformers can step up the voltage with high turns ratio, but at the same time the impedance is converted by the voltage step up squared.  So a transformer will need a buffer, which is kind of like what a transformer input mic preamp is.

There is no free lunch... a passive sum bus requires make up gain pretty much equal to the noise gain of the active sum bus.. Six of one- half dozen another. 

Passive sum with transformer and buffer, may help develop a 'transformer heavy" sound signature, but it won't be a better summer, just another different one.

JR

One stage active summer driving a 1:1 ratio using the transformer to eliminate the second phase inverting stage usually needed. Servoed to negate the need for a big electrolytic cap between the high voltage DOA and tranny. Gain makeup amp is also transformer coupled...very nice sound placement  8). Oh...every pan pot is driven by a buffer.

Neve used passive summing into a 2:1 transformer in all their classic mixers. The trouble with a high ratio transformer to try to make up all the gain is that the output impedance will 100 times the bus impedance.

Cheers

Ian

Far be it for me to throw stones at Rupert, and I was a small competitor back in the day so perhaps biased.

I did some bench work back in the '70s looking at whether I could get some free transformer voltage gain in combination with opamps of the time, and keep it all inside the negative feedback loop... My experiment failed miserably for more than one reason (while I believe there are variants out there that work).

I can imagine how using a modest step up transformer hanging off a passive bus, back in the day could have merit over just using the opamps of the day at high noise gains alone. In the '70's we saw newer low noise discrete devices and improved general purpose opamps that changed our available tool kit, while sum bus design will always require special consideration due to high noise gain involved.  Eliminating 6dB from the make-up gain needed is not nothing but in the context of a bus with a few tens of inputs maybe not a game changer.

Note: stepping up the bus impedance was probably a major benefit of using it...  opamps of the day were a poor match for sub 1k source impedance.  Trying to get bus impedance up using large value resistors could have been problematic for stray capacitance, and crosstalk issues.

I am not a student of what is actually under Neve's hood so this is speculation on my part.

JR

JohnRoberts said:

I can imagine how using a modest step up transformer hanging off a passive bus, back in the day could have merit over just using the opamps of the day at high noise gains alone. In the '70's we saw newer low noise discrete devices and improved general purpose opamps that changed our available tool kit, while sum bus design will always require special consideration due to high noise gain involved.  Eliminating 6dB from the make-up gain needed is not nothing but in the context of a bus with a few tens of inputs maybe not a game changer.

Note: stepping up the bus impedance was probably a major benefit of using it...  opamps of the day were a poor match for sub 1k source impedance.  Trying to get bus impedance up using large value resistors could have been problematic for stray capacitance, and crosstalk issues.

I am not a student of what is actually under Neve's hood so this is speculation on my part.

JR

Click to expand...

I was at Neve in the 70s and when I left in 1976 Neve were still using entirely discrete designs - no op amps The transistors used at that time had an optimum noise figure with a source impedance of about 4K8 and the 2:1 transformer matched this to the nominal 1K2 input impedance of Neve mic inputs of the day. The same mic pre input end was used for bus gain make up except the gain was fixed. Bus resistors were typically 15K which for a 16 channel bus gave a bus impedance of less than 1K.

In the mid 70s the original NE5534 came out - it was called something different initially - TDA something from Phillips. Neve carried out extensive tests of it and some years later did start to use them. While I was there, the R&D department did a study comparing the normal Neve passive mix bus technique with virtual earth mixing using op amps and concluded that in noise terms there was little if anything to choose between them. There are several other difficulties in using op amps that Steve Dove has detailed in his excellent mixer design series in Studio Sound.

However, in a (physically) big desk, mix bus design is as much about the physical arrangement as it is about whether passive or active mixing is used. As I think I have described in another thread, the Neve mix bus was an aluminium extrusion with U shaped channels for each bus with bus resistors literally 'fed through' the bottom of the U and soldered into a single hole width of Vero board. When complete a flat bottom plate was screwed on so the whole bus was completely screened.

Op amp based virtual earth (i.e. active) mixing only works because the op amps have enormous open loop gains but that is also what gives them problems too. In tube based mixers it is not possible to achieve these huge open loop gains (at all and certainly not without stability problems) and so virtual earth mixing with tubes for anything other than a very small number of sources is not really practical.

Cheers

Ian

Thanks for the lecture on bus design, I published an article on console design back in 1980 (RE/P). IIRC I discussed a few different bus configurations and approaches, but my article was more of an overview than a comprehensive how to.
-----
I forget whether it was Steve Dove or Ted Fletcher, but i recall mailing some low noise transistors to one of them after they made some claims about the SOTA for transistor noise in studio sound, that was not even close. I found some really low Rbb,  low noise < 1nV/rt Hz bipolars from a small Japanese company developed for MC head amps. .
-------
IIRC the Phillips PN was TDA1034
------
If you read the original console series by Steve in Studio Sound in his last installment he gave me a friendly mention (for pointing out a mistake he made in an earlier installment). Steve and I later became friends, last I heard he was living in PA (US). 
-----
Indeed physical structure makes a difference, consoles can be several feet wide so real buses have plenty of capacitance and other practical concerns to deal with.

As I've said to any who will listen, noise is only one issue with sum bus performance, while perhaps the most obvious, it is academic in the context of the noise floor of typical real audio sources. What is less apparent but very real is phase shift and distortion from all the make up gain of a passive front end, or the noise gain of a virtual earth. Pretty much a six of one- half dozen other wrt to total gain.

I am not casting aspersion on Neve design, while I never was a fan of transformers for audio. I jumped on the transformer-less mic preamps for consoles when they first became practical in the '70s. In fact the lack of iron was a selling feature on my consoles. While even back then (like now) opinions varied. 

That first company I designed consoles for went belly up, while i don't blame their failure on my designs (Of course I wouldn't even if I were responsible). I can't deny they went belly up while Neve prospered for some time after that.

======

Some people are fans of the classic Neve sound so thanks for sharing... that information should appeal to some here. Sorry if it sounded like i was dumping on the design, just exploring the topology and you added information helps. If they were targeting a 5k optimal impedance for noise, that ratio makes perfect sense. 

JR

Note: the low noise bipolar devices I found back then were designed to work with tens of ohms source impedance... They would in fact be noisier with a 1:2 step up transformer...Note even they have since been obsoleted because they are actually lower Rbb than optimal for direct interface with mics, and MC phono carts are no longer mass market products.   

JohnRoberts said:

-------
IIRC the Phillips PN was TDA1034
------

Click to expand...

That's the one! For some reason I keep thinking it was the 1024 which is a completely different beast. I just looked up the data sheet which brought back another memory - that the ones used a Neve were the metal can types.

I am not casting aspersion on Neve design, while I never was a fan of transformers for audio. I jumped on the transformer-less mic preamps for consoles when they first became practical in the '70s. In fact the lack of iron was a selling feature on my consoles. While even back then (like now) opinions varied.

Click to expand...

I didn't for a minute think you were. The 70s were a pivotal time for studio electronics. Electronically balanced outs and then ins became practical which meant  a significant cost element in mixers could be eliminated. I remember having fairly heated discussion with Clive Green of Cadac about the pros and cons of transformers. At that point I think Cadac had balanced outs but still used transformer inputs.

I suspect this lead to a plethora of new mixer companies that used simplified mechanics, buses and connectors to make good spec mixers at a much lower cost than Neve. Some survived and some didn't. Rupert did go this route partially with the development of the CRC range of radio consoles (CRC stood for cheap radio console). It still had transformer inputs but cheaper transformers and much simplified mechanics.

Cheers

Ian