Balanced output gain - confusion...

[Nora]

Ok, I don't get it.

If you send 1 volt into a 1:1 transformer primary you get 1 volt on "xlr pins 2 and 3"
If you send 1 volt into a drv134 (or equivalent) you get 2 volts measured between "xlr pins 2 and 3"

When the transformer output connects to a balanced input with unity gain, you get 1 volt on the "unbalanced side"
But a drv134 to a balanced input gives 2 volts?

Are my assumptions correct? Or am I missing something really dumb??

Is is because these opamps inside the cross-coupled output drivers have to run at least at unity gain, doubling that (+ and -) gives the 6dB gain?
So to get unity gain from in to output, you must attenuate 6dB inside the device?
Or is it because of the increased headroom?
Or am I just confused  ???

How about differential outputs. When connected to balanced loads they behave the same way as cross-coupled outputs, so 6dB gain for 'opamp unity'??

Someone enlighten me please. The intertubes have only made the confusion worse.

 

[hobiesound]

doesn't the drv134 have a gain of 6 db? so in essence it works like a 1:2 transformer.

 

[Nora]

So using a drv134, that 1646, is just like using a 1:2 output transformer?
To allow a unit to be unity gain, you'd have to use a -6dB input amp, like the ina137/that1206 instead of the ina134?
Or compensate inside the circuitry...

 

[clintrubber]

It depends on the used topology of the active circuit.

There are approaches where the outputs are 'sensed'
and what is 'found' there (in terms of signal) is taken into account
to calculate what is presented at the other output.

So if say the negative output is shorted (for instance a TS plug into a 'balanced TRS output')
the signal at the positive output is twice as large.

This kind of topology (sort of) does what a 'real TX' would do.


With the more common approach of 'negative polarity provided by an added inverting opamp'
the 'primary output' (the one from which the other polarity is derived) doesn't know what the 'secondary'
output is doing. The signals are 'thrown' at the outputs whatever what (unless the primary output is shorted and there's no output-series-resistor, in which case all goes silent).

Still with me ?    I don't recall exactly what topology those mentioned chips are using, but I thought they're of the sensing kind. And then they'd be 'TX-like', in that they keep the resulting signal-level between the two polarities the same, regardless of the termination of the outputs.


In addition to all this there might be an added amount of gain, which might be the confusing thing here.

Cheers

 

[Nora]

I'm all with you on the different behaviour of electronically balanced vs cross-coupled/compensating outups when driving an unbalanced load.

But more specifically, my question is:
A crosscoupled output into a balanced load behaves like a 1:2 transformer, ie 6dB gain (because both opposing polaritys are unity gain).
Does a simple differential output driver (two independent polaritys at unity) offer 6dB gain between hot and cold?

When fed with a 1v input, a cross coupled stage outputs 2v between hot and cold. A differential amp puts out 1v from hot to 0v and one volt from cold to 0v (well, -1v) wich, when loaded by a differential or transformer input, also appears as a 2v signal?

If so, why does a circuit, like the gssl, offer unity gain (balanced) without compensating the 6 dB? Or does it?

Am I missing something really dumb wrt the floating behaviour of crosscoupled outputs and transformers?

I've built quite some gear (mostly transformers) and wired a few studios.
I know how to wire equipment depending on the output/input arrangment.
But I've never really understood how the intern circuit gain relates to the outside world.
For example, when a unit with a drv134 output has a nominal operation level (output) of +4dBu, the internal circuit operates at -2dBu?

Ill draw a schematic if that clears things up!

Thanks!

 

[Harpo]

If so, why does a circuit, like the gssl, offer unity gain (balanced) without compensating the 6 dB? Or does it?

It does compensate by the ratio of Rin in front of the audio-VCA and Rfb in following opamp.

 

[clintrubber]

A lot of gear provides a balanced output by adding that inverting unity gain opamp (-1).
So yes, indeed, the resulting signal at the balanced output is then twice that of the internal operating level (unless compensated for).

Like the output here:  http://gyraf.dk/gy_pd/ssl/ssl_sch.gif

The G-SSL is actually not the best example here though, since there's actually only one internal unbalanced voltage signal node in the signal path, so 'the internal operating level' is actually just one node/channel.


> If so, why does a circuit, like the gssl, offer unity gain (balanced) without compensating the 6 dB? Or does it?

I figure the G-SSL (being balanced-in, balanced-out) only needs to compensate when one of the plugs is used in an unbalanced way. But when used like this, it's not compensated, see the schematic, there's no provision.

I'm afraid we're making a simple think pretty complicated here... we might bring in some Rane-pdf-note or something, they'll have spent more time to describe it in the best way  

Bye

 

[Nora]

If so, why does a circuit, like the gssl, offer unity gain (balanced) without compensating the 6 dB? Or does it?

It does compensate by the ratio of Rin in front of the audio-VCA and Rfb in following opamp.

Ok that was a tough one, but I get it ;D
Gssl is a bad example. But it clears up my mind.

I know it seems like I'm over-thinking it, but in an application with no gain control, where unity gain is required I couldn't stop wondering what happened with those 6dB.

> If so, why does a circuit, like the gssl, offer unity gain (balanced) without compensating the 6 dB? Or does it?

I figure the G-SSL (being balanced-in, balanced-out) only needs to compensate when one of the plugs is used in an unbalanced way. But when used like this, it's not compensated, see the schematic, there's no provision.

I'm afraid we're making a simple think pretty complicated here... we might bring in some Rane-pdf-note or something, they'll have spent more time to describe it in the best way  

I'm probably confusing you about my confusion by using the word "compensate" in different contexts, I just meant, like harpo pointed out, "loose 6dB because it gets added by the output"


So one could summarize: To deal with the 6dB boost from output stage you can either:
Loose 6dB inside the unit or use it together with an output fader that goes up to +6
Or use a -6dB input receiver (like that 1206 or ina137)

Thanks guys
this was a piece of geek-theorie that I didn't understand but kept ignoring, until I finally had to face my ignorance  

 

[dmills]

The reason those parts often have a 6db gain (and you want it!) is actually quite subtle.

Consider some single ended circuit block running from say +-15V, its output can cleanly swing maybe +-12V or so.
Now if we add a differential output stage with 6db of gain, then we can now swing the output +-24V differential, and if the input block has -6db gain then our output driver will clip at roughly the same point that the internal single ended electronics does and we still have unity gain overall.

Running the output hotter makes for 6db better interference rejection all things being equal (Seldom matters in a studio, sometimes is a big deal live).

Regards, Dan.