help me to understand: bal to unbal, 6 bd loss?

[abbey road d enfer]

i heard that going from balanced to unbalanced at unity gain produces a loss of 6db volume.
is it correct?

Back to the original question: When does the dreaded 6dB loss happens?
1) When an unbalanced source is connected to a diff input and the cold leg is left floating, leaving the NFB foot resistor in the air.
é) When the output is of the "stiff" balanced type (non floating, two legs delivering equal voltage in polarity reverse) and connected to an unbalanced input.

1) is a wiring mistake
2) is a flaw

 

[abbey road d enfer]

Analog processing for mastering is typically line level, where a noise rejection of less than 70 dB seemingly can be tolerated.

That's quite a claim! So maybe we can tell these guys at Maselec, SPL or Prism that they can relax their standards.

 

[tv]

I think they'll rather charge additionally for a "high-end mastering" conversion..

the6db loss is going to happen when one of the signal lines is shorted to ground.. or theres no signal in it (fault) when all of the signal lines are connected

if +signal input is disconnected (floated), there will be also 6dB loos
if -signal input is disconnected (floated) there will be iirc 12dB llos (with the simple receiver)


a big thransformer can't be labeled as a small analog computer imho...

 

[JohnRoberts]

I think they'll rather charge additionally for a "high-end mastering" conversion..

the6db loss is going to happen when one of the signal lines is shorted to ground.. or theres no signal in it (fault) when all of the signal lines are connected

I know I've written about this before but perhaps not in this thread.

NO... there are two different output designs that can tolerate shorting one side and not lose 6 dB.

1- the floating transformer output (note: A grounded center-tapped transformer will lose 6dB or more if one leg shorted)
2- the dual active driver with cross coupled feedback can sense the level of both outputs and adjust. You lose 6 db of max output but nominal output is correct. 

Perhaps this is why floating transformer outputs are so popular for 70-100V constant voltage audio distribution. Workers can be sloppy and short one leg (either leg) and the sound still plays properly at normal level.

if +signal input is disconnected (floated), there will be also 6dB loos
if -signal input is disconnected (floated) there will be iirc 12dB llos (with the simple receiver)

There are several different input and output circuits that can behave differently in response to different faults.  Know and understand what you have.

a big thransformer can't be labeled as a small analog computer imho...

huh... ?

I'm inclined to use 3 conductor wiring, even for 2 conductor interfaces. Putting ground in a different conductor than audio low, might reduce a problem or two.

This is pretty old news and modern canned (IC) solutions make this harder for junior engineers to screw up, as long as they follow the simple instruction. 20-30 years ago pin 1 problems were not as well understood as today. This is not or should not be a significant path limitation for most audio paths.

of course opinions vary and people will argue about anything.

JR

 

[Kingston]

No wonder there are rumours that some mastering studios convert their gear to unbalanced..

Not rumours, actual success stories. Just not related to this particular issue. The two real-world examples I know are both entirely about shortening signal path by removing everything related to balancing stages. But these people know what "ground" means and how to deal with it in relation to a signal path.

No success story can convince me of rewiring the next Bruce Springsteen tour unbalanced, although (or perhaps because) I know what "ground" means, i.e. a theoretical notion.
In order to achieve a noise rejection of 70 dB, which many diff input stages do, you need to have a "ground" connection totalizing less than 1 milliohm for a nominal 10k connection. I'm not ready to wire my studio with copper bars and power distro terminals.

The examples I know are both hard wired mastering studios. I'm not an advocate for these kinds of optimisations, just pointing them out. I mean which client is going to care for 0.01% less THD+N in a complete system. And I can certainly picture even a small live venue fully unbalanced would be a disaster.

 

[tv]

@JR those figures were actually wrt posts 18, 20
but wrt "arguing" there are lots of different usage scenarios and exchanging opinions is imo a natural byproduct of diversity

@Kingston
come on, it's not that hard to do it at home, I mean, it's not likely you will have a fully treated mains supply in the first place, and also you most likely won't have a fully treated ground connections .. so you will have to work a little harder anyway .. compared to a dedicated studio room where everything was taken care of beforehand. But it can be done. Getting the ground to be "quiet" or at least neutral enough will also benefit balanced connections.

 

[ruffrecords]

This is the real problem with modern differential outputs - because they are differential, both outputs need a reference, often called 'audio ground'. A truly balanced output does not need a reference; there is simply a voltage between two terminals. There is then absolutely no reason whatsoever for there to be an audio ground link between any two pieces of equipment. Of course you can only do this with transformers. Sometimes the old ways are the best.

Cheers

ian

 

[JohnRoberts]

The old ways are the old ways and the best they could do at the time.

JR

PS If old is good,,, I'm great.

 

[saint gillis]

That's right with equipement using input and output tranformers there are no pbms.
In our studio gear with unbalanced inputs are connected to the patchbay with negative pin tied to gnd, it is very problematic when gear with aop differential outputs are connected to them (signal loss, sometimes noise cracks etc), but the technologie JohnRoberts described :

No switch.. there are active balanced outputs with cross connected feedback that set level based on the sum of both outputs. When one leg is shorted, the level on the other leg increases to make up the difference.

is very convenient, no pbms with the RME converter analog outputs connected to unbalanced inputs gear

 

[tv]

...
differential outputs - because they are differential, both outputs need a reference, often called 'audio ground'. A truly balanced output does not need a reference; there is simply a voltage between two terminals. There is then absolutely no reason whatsoever for there to be an audio ground link between any two pieces of equipment. Of course you can only do this with transformers. Sometimes the old ways are the best.
...

+1

However, if you think in terms of Kirchoff laws, one thing becomes apparent.

Fe, like in the pic, if you have a stiff, hard differential voltage (fe. the cursed simple balanced driver) - AND - an impedance-optimized receiver (both legs impedance balanced). When you drive such a connection, - IF - you break the ground connection in point X, there should be no potential difference across that point.

(Just in theory), even with modern differential connections, you should be able to float it if above criteria is fullfilled. (But the real-life environment makes it very unlikely).

 

[ruffrecords]

The old ways are the old ways and the best they could do at the time.

JR

PS If old is good,,, I'm great.

And the new ways are the new ways and the cheapest (but not the best) they could find.

I prefer to use what's best, whether old or new.

Cheers

Ian

 

[JohnRoberts]

The old ways are the old ways and the best they could do at the time.

JR

PS If old is good,,, I'm great.

And the new ways are the new ways and the cheapest (but not the best) they could find.

That is a rather sweeping condemnation of modern design. While in fact it is also accurate. Mainstream consumers influence design engineers to be cost conscious. In my personal experience since the 1980s, transformer-less inputs and outputs were better than all but the most expensive transformer solutions, and that extreme was debatable, with good transformer design outpointing solid state in only one or two narrow aspects, but losing in most other general performance metrics. When trying to design products that consumer buy with their own money it was an easy call to deliver arguably better performance for a lot less money. I suspect product positioned for higher price points might gravitate toward transformers as partial justification for the cost.

I prefer to use what's best, whether old or new.

Cheers

Ian

My personal choice is transformer-less and the only customer segment I encountered that the majority favored (output) transformer audio interfaces was fixed install market. This is curious since they were not very critical of overall path performance metrics. They demanded output transformer isolation since it gave them more fault tolerance in wiring. Avoiding even one service call was worth the incremental cost of transformer isolation. 

JR

 

[saint gillis]

JR,

No switch.. there are active balanced outputs with cross connected feedback that set level based on the sum of both outputs. When one leg is shorted, the level on the other leg increases to make up the difference.

These outputs can give errors when both outputs are not read by following inputs.

JR

is the balancing circuit I ve tried to copy from the Studer 900 faders an exemple of what you are talking about?

  Maybe you have some other exemples?

 

[abbey road d enfer]

JR,

No switch.. there are active balanced outputs with cross connected feedback that set level based on the sum of both outputs. When one leg is shorted, the level on the other leg increases to make up the difference.

These outputs can give errors when both outputs are not read by following inputs.

JR

is the balancing circuit I ve tried to copy from the Studer 900 faders an exemple of what you are talking about?

That one is a variation on the cross-coupled principle. Because it has rather limited common-mode impedance, it sits between fully-floating and hard-balanced, meaning that for high Z loads, the voltage of each leg will be more or less stable, meaning that the 6 dB loss will be experienced when one of the legs is left unconnected. For lower load Z, it will behave like a xfmr, yielding almost no signal when one leg is not connected.
However, when properly connected, this output stage can be associated with all types of inputs, unbalanced, hard-balanced, floating, differential...).

 

[ruffrecords]

My personal choice is transformer-less and the only customer segment I encountered that the majority favored (output) transformer audio interfaces was fixed install market. This is curious since they were not very critical of overall path performance metrics. They demanded output transformer isolation since it gave them more fault tolerance in wiring. Avoiding even one service call was worth the incremental cost of transformer isolation. 

JR

Times change and with it needs. When I was at Neve back in the 70s we built consoles for many major studios and broadcasters. Most were substantial facilities with several studios, separate machine and reverb plate rooms and so on. Mains supply could be on any one of three phases, the voltage between different mains earths could be many tens of volts and in many cases there were nearby transmitters. The galvanic isolation provided by transformers was indispensable in these situations.

Today, with the huge number of project studios based in a single room operating from a single mains supply galvanic isolation is no longer essential. Despite that, there are endless posts asking how to fix hum loop problems in project studios.

I prefer transformers all round.

Cheers

Ian

 

[JohnRoberts]

JR,

No switch.. there are active balanced outputs with cross connected feedback that set level based on the sum of both outputs. When one leg is shorted, the level on the other leg increases to make up the difference.

These outputs can give errors when both outputs are not read by following inputs.

JR

is the balancing circuit I ve tried to copy from the Studer 900 faders an exemple of what you are talking about?


  Maybe you have some other exemples?

No examples handy but yes, this is one example. I used variants of this approach on many products, since it delivers 2x the signal output of SE outputs with good balance, for a fraction of the cost of a transformer that could deliver anywhere near comparable performance. The approach is also reasonably tolerant of inadvertent shorts. For example plugging a TS plug into a TRS jack would return the correct full level signal to the tip, without stressing the electronics. Note: shorting one leg limits max output to be 6dB less than both legs swinging, but gain adjusts so the output is the same until it reaches that voltage swing limit.

JR

 

[saint gillis]

This looks like a very clever improvement on many devices using balanced (or should I say "differential" right?) aop outputs, especially when the rule chosen in the studio is to connect negative wire to ground when unbalanced inputs...
  Better than this kind of design right?
http://www.douglas-self.com/ampins/balanced/balfig5b.gif

 

[JohnRoberts]

This looks like a very clever improvement on many devices using balanced (or should I say "differential" right?) aop outputs, especially when the rule chosen in the studio is to connect negative wire to ground when unbalanced inputs...
  Better than this kind of design right?
http://www.douglas-self.com/ampins/balanced/balfig5b.gif

Yes, that simple dual polarity output is OK for dedicated applications where you are the only person using the gear, but I found when designing products for the mass market you must make the designs customer-proof.  Driving one of two output opamps into ground will not make for very happy opamps, and all that signal current in ground could cause crosstalk issues. Not a good design practice.

JR

 

[[silent:arts]]

the unhappy opamps will blow fuses ...

 

[JohnRoberts]

the unhappy opamps will blow fuses ...

Generally between the opamp current limiting and build out resistance the fuses won't blow, but the opamp will get hot as a fire cracker and may fail prematurely. Perhaps a whole console full of shorted drivers could take out the fuse. Another subtle unintended consequence of shorting one output line, is that when the opamp loses linear negative feedback, this clipping/saturation can reflect back through the input resistance and introduce distortion into the good output. 

JR