Lynx Electronically Balanced Outputs

[Steamy Williams]

I have an Lynx Aurora 16 that I will soon be integrating into my studio. The Aurora manual doesn't say anything about running the electronically balanced outputs into an unbalanced input. So I called Lynx and Paul referred me to the Lynx TWO manual, as the information about wiring outputs in that also pertains to the outputs on the Auroras. I will rarely be running the outputs unbalanced, but I'd like to be aware how to patch it into one or two devices in my studio that have an unbalanced input (API 550A's for instance) when the occasion arises.

Anyway, I'm vaguely familiar with all of the various types of outputs - transformer, quasi-floating electronically balanced, "true" electronically balanced, impedance balanced, ground compensated, etc - and how to connect them to balanced or unbalanced inputs. However, Lynx's instructions to leave the cable sheild unconnected at the destination has thrown me a bit, as it does not follow the wiring convention I use in my studio - which is to have all the grounds connected at both ends. So I'm not entirely sure whether to tie the cold to ground or to leave the cold floating at the source (i.e. the Aurora).

Therefore, my question is - from looking at the manual, can you tell if the Aurora's analogue outputs are quasi-floating or not?

Hope that makes sense.

[ulysses]

The manual says that when you connect the XLR to a TS plug, you should connect XLR pin 3 to the sleeve of the TS jack. Therefore, this must be a "ground compensated" (what you're calling "quasi-floating") output circuit.
The sleeve contact on the unbalanced 1/4" jack destination is, of course, ground. It will (or should) be connected to the chassis of the Aurora (and therefore XLR pin 1) through the safety ground on the power cables. The reason they do not recommend connecting the cable shield to the TS sleeve is because that would create a 2nd ground path. This is where you start getting into trouble with ground currents and such. However, if the gear at both ends is properly designed, then connecting the shield at both ends of the cable should not present a problem. If that's the convention you're using successfully in your studio (and I would argue that it is the correct convention - floating shields are a band-aid for other problems) then you will probably be fine with this scenario as well.
This is a complicated subject with lots of variables and lots of conventional wisdom that isn't especially wise. My suggestion is to go read this article at Rane's website and then read it again and again until it makes sense. It helped me a lot. It is really a very good discussion of the subject.

[Steamy Williams]

Originally Posted by ulysses

The manual says that when you connect the XLR to a TS plug, you should connect XLR pin 3 to the sleeve of the TS jack. Therefore, this must be a "ground compensated" (what you're calling "quasi-floating") output circuit.
The sleeve contact on the unbalanced 1/4" jack destination is, of course, ground. It will (or should) be connected to the chassis of the Aurora (and therefore XLR pin 1) through the safety ground on the power cables. The reason they do not recommend connecting the cable shield to the TS sleeve is because that would create a 2nd ground path. This is where you start getting into trouble with ground currents and such. However, if the gear at both ends is properly designed, then connecting the shield at both ends of the cable should not present a problem. If that's the convention you're using successfully in your studio (and I would argue that it is the correct convention - floating shields are a band-aid for other problems) then you will probably be fine with this scenario as well.
This is a complicated subject with lots of variables and lots of conventional wisdom that isn't especially wise. My suggestion is to go read this article at Rane's website and then read it again and again until it makes sense. It helped me a lot. It is really a very good discussion of the subject.

Thanks Justin,

I thought I had a grasp on the differences between the various types of "balanced outputs", but now it's clear to me that I don't. It was something I touched upon quite briefly in a project I did for my audio and recording tech degree, but only to ensure that various setups were wired up correctly for comparative purposes. I tend to use the following site as my main reference when trying to understand these things:

http://www.dself.dsl.pipex.com/ampin...d/balanced.htm

Most of the actual info about the output circuits is way over my head. However, that site describes "quasi-floating" and "ground-cancelling/compensated" as to two distinct and different output types, of which ground-compensated is the one I have the most difficulty understanding. Anyway, just one question before I go and have a look at the Rane website - is a ground-compensated output actually differential (unlike an impedance-balanced output, for instance)?

[ulysses]

Okay, the circuit Self refers to as "ground-cancelling" (fig. 5a) is kind of an oddball that in my experience is not at all common. It is an unbalanced output with, like it says, compensation for ground noise. That's not what I was talking about.

The "Balance line output" (fig 5b) is the typical non-compensated, electronically balanced output. This is similar to the circuit I used in my Super Stereo compressor. Its advantages are its simplicity and its short signal path. The disadvantage is that special consideration is needed when feeding an unbalanced load. Most other arrangements allow you to simply tie Pin 3 to ground, and away you go. In this circuit, which is fairly common, grounding Pin 3 makes the inverted output somewhat unhappy (though most decent op amps can handle it okay) and it can also dump a lot of highly-distorted signal current into the ground. If either the source or the load are laid out poorly in terms of their ground scheme, this can be an ugly scene. Also, you get 6dB less signal when feeding an unbalanced load. For the purposes of the Super Stereo compressor, I felt these were acceptable limitations since it's being marketed to high-end post-production applications where it will be interfacing with balanced loads and/or well-planned and well-executed interconnection schemes.

The Quasi-Floating Output (fig 6) is the one I was discussing in my previous post. This is basically a version of the balanced circuit that's been modified to eliminate the two compromises I mentioned above. The only real drawbacks to this circuit are that it's more complicated than the regular electronically-balanced circuit, with higher parts count and longer signal path. Of course it still doesn't offer the galvanic isolation of an actual transformer.

[Steamy Williams]

Thanks again Justin, I understand again now.

It was the confusion between quasi-floating and ground-compensated/cancelling that got me even more muddled. Incidently, I think the Soundcraft Ghost I used to have had the unusual "ground-cancelling" output that Self is describing. Soundcraft didn't called them balanced or unbalanced, but rather "ground-compensated" (as opposed to Self's term "ground-cancelling"). As an aside, I think it's kinda like false advertising when certain manufacturers try to pass off "impedance-balanced" outputs as real balanced outputs, even though that is inline with the AES description of a balanced output.

Anyway, I know were I am with the other two types of proper (i.e. differential) electronically balanced outputs, so that definitely helps.

[ulysses]

I actually think impedance balanced outputs are one of the best solutions where full compatibility with both balanced and unbalanced loads, regardless of wiring convention, is required. Especially on TRS jacks because the circuit doesn't even care if you use a TRS cable or a guitar cord (other than to achieve noise rejection). Of course a transformer is still best, but still less "universally compatible" than the impedance balanced circuit (if you float the cold on a transformer circuit, it doesn't work).
Remember, it's the impedance and the line noise that needs to be symmetrical for good common mode rejection, not the audio signal.