Issue with balanced to unbalanced output

Guys I'd need a bit of explaining here!

I've built two units that use balancing circuit with NE5532 inverting the signal and 22R/47R as a series resistors on the output.
It all worked well for months as I've been using those units exclusively on balanced lines. But as soon I hooked them on my studio setup, where I have my console's (unbalanced) inserts on the patchbay, and I wanted to insert them in the console - everything fell down! The sound of both units was very distorted!
So I went to search for a problem.
As I found out, my patchbay has insert points soldered in a way that pin3 (or ring part of the TRS connector) is grounded. Since insert signal travels unbalanced, ring and ground are tied together both on input and output point on the patchbay (console side). At the same time, gear has its ins and outsconnected balanced to the patchbay.
So my problem is what happens when the output from the unit has ring (pin3) connected to ground. Series resistor with it's 22R is not enough to stop ground entering the circuit. So it distorts on the output. Even the second unit with it's 47R has too small series resistance.
I solved it by attaching 150R in series on both pin2 and pin3. And it works now, both in balanced and unbalanced situation! At least in my setup...
But it's bugging me! It shouldn't be that way, right? I'm reducing current flow with increased resistance and it is reducing headroom. I haven't noticed it in practice, even though I tried boosting too much bass freqs on the already hot signal, and it didn't clip.
If I lower the series resistance to 100R it introduces distortion. I guess it's not enough to stop the ground from entering the unit.
The units in question are Nightpro EQ-3D and Pico compressor balanced with additional Audiox balancing circuit.
I know the best thing to do would be to just replace the balancing circuits with THAT or BB IC's, but this setup is already working. I just need some clarification why those series resistors are not 150R (or 180R to 200R) in the first place, and what is wrong with this higher value.

thanks!


Luka

There is a trade off as you increase the series resistance it will interact more with cable capacitance and non-linear input impedance.

In general it not good practice to short an active output to ground. Besides the increased distortion, the elevated ground current can cause crosstalk issues. I have seen semi-pro interfaces with close to 1k output resistance but they are not intended for long cable runs or professional applications.

Do what works for you, for years many professional interfaces operated at 600 ohm  impedance.

JR

Thank you John for clarification! 

I knew that if I connect them unbalanced permanently with pin3 disconnected it would be all okay.
But my problem is that I need to have my units connected to the patchbay since my converter outs are normaled there, and I want to be able to use those units in a mastering situations where I use them balanced.
I guess this is the flaw of this design...

Would this happen if there was an opamp with unity gain on both lines of the output (pin2 and pin3) followed by low series resistance (22R)?

shot said:

Thank you John for clarification! 

I knew that if I connect them unbalanced permanently with pin3 disconnected it would be all okay.
But my problem is that I need to have my units connected to the patchbay since my converter outs are normaled there, and I want to be able to use those units in a mastering situations where I use them balanced.
I guess this is the flaw of this design...

Would this happen if there was an opamp with unity gain on both lines of the output (pin2 and pin3) followed by low series resistance (22R)?

Click to expand...

Sorry I don't follow what you are suggesting.

Alternate active designs have been around for decades with cross connected NF sensing so a short on one output leg is detected and the level increases in the other output (these have their own problems). This capability is built into some of the new driver ICs.

Your hard wired dual output does not really buy much benefit over a clean impedance balanced output (one active drive and one output resistor to ground) .

Consider doing less rather than more.

JR

The straightforward foolproof answer to your problem is:

1. Wire your patch bay balanced
2. Convert you unbalanced equipment to balanced using isolating transformers.

Cheers

Ian

Why 22 Ohms? Are you driving speakers?

That's really too low for driving l-o-o-o-o-n-g lines. (60 ohms is a better send termination for lines a mile long.)

In a small studio, I'd use 330 Ohms. Treble loss is negligible past a hundred feet. '5532 can drive a shorted 300 Ohms without gross distortion or major crap-up of shared power rails. 330r:10K is 0.3dB loss, insignificant (especially on top of the 6dB change from only taking one side of the push-pull).

JohnRoberts said:

Consider doing less rather than more.

JR

Click to expand...

You're right on this!
Got me thinking to just use two patch points and have balanced and unbalanced on different locations. That way it's (almost) best of both worlds.

ruffrecords said:

The straightforward foolproof answer to your problem is:

1. Wire your patch bay balanced
2. Convert you unbalanced equipment to balanced using isolating transformers.

Cheers

Ian

Click to expand...

In an ideal world I'd just put transformers on all my gear. But sometimes it's too expensive or the circuit has already it's own balancing circuit of some sort so I'm hesitating to remove it.

I asked this question to understand more of this specific circuit and series resistors theory in general.
I'm satisfied how it works in my setup now after I've increased series resistors.



Luka

PRR said:

Why 22 Ohms? Are you driving speakers?

Click to expand...

I've etched this project:
http://groupdiy.com/index.php?topic=29333.msg355073#msg355073

It has series output resistance 22R

Another similar balancing scheme, used on Nightpro EQ-3D "Night Pro" project.
http://groupdiy.com/index.php?topic=19789.0
It uses 47R as series resistance.

PRR said:

That's really too low for driving l-o-o-o-o-n-g lines. (60 ohms is a better send termination for lines a mile long.)

In a small studio, I'd use 330 Ohms. Treble loss is negligible past a hundred feet. '5532 can drive a shorted 300 Ohms without gross distortion or major crap-up of shared power rails. 330r:10K is 0.3dB loss, insignificant (especially on top of the 6dB change from only taking one side of the push-pull).

Click to expand...

Since I had problems with 22R, I've added 150R in series with it. So now it's 172R.
According to what you say - this is perfectly fine. Right?
After I did it, I don't have any problems with interfacing into grounded pin3. But it was bugging me if what I've done is somehow wrong. That's why I started this topic - to ask if I'm doing wrong!

L

Hi, I'd go as JR said, with impedance balanced outputs, unless you really need the extra 6dB headroom, in which case you also need the balanced inputs. You only need to use transformers in the unbalanced inputs, the ones which already have balanced inputs don't need a transformer.

If you are using this in a fixed installation you could wire unbalanced output, if the input will be unbalanced anyway, there's no reason to have the balanced output, or you could have both options, or cut the wires from the ring to the ground which may be an issue if your source has an output transformer or a floating balanced stage.

I still think the impedance balanced output is the best options most of the time, first is the simplest, and works in every single condition, it the input is balanced you get the noise cancelation, if it's not you don't have any other noise or distortion and the level is the same, the only issue are the 6dB you loose there, but as I said, if you need that you also need or your balanced input or higher rails or output transformer.

JS

I should press LIKE buttons on all of your replies! But it would feel to facebookish!

Thank you all!