Ground loop breaker and unbalanced connections

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

saint gillis

Well-known member
Joined
Nov 4, 2012
Messages
871
Location
Brussels - Belgium
  Hello, suppose we want to make a device with several boards inside, and several inputs and outputs, maybe some of them balanced, some others unbalanced.
  The PSU is connected to the chassis through a ground loop breaker. What is the best way to connect the unbalanced ins/outs ground pins?

GROUNDING-RACK.jpg
 
Based on past gdiy advice given to me, unbalanced signal should sport a dedicated agnd bus/plane, eventually tied to chassis at a single point.
 
saint gillis said:
The PSU is connected to the chassis through a ground loop breaker.
Depending on specifics, that could be unsafe. If a hot wire touches the metal chassis, it could be a hair straitening device.

My friend had fridge in his basement that shocked people. We would hold hands in a line while the person on the end would put their bare foot down on the concrete more or less to see who would give up first. I ne-ne-never gave up.
 
Unbalanced connections are always difficult because they have the potential (haha) to create ground loops. Balanced ins and outs are straightforward. Screen goes to chassis, hot and cold go the electronics.

For unbalanced outputs, the cold is effectively connected to analogue 0V (which really should be connected to chassis at one point only). So to avoid multiple connections of analogue 0V to chassis you can wire unbalanced outputs to balanced connectors. Screen goes to chassis, hot goes to unbalanced out an cold goes to that output's local analogue 0V. That way, at least if you connect it to a balanced input you don't get a ground loop. You don't get all the benefits of a balanced connection either but you can improve the situation by balancing the impedance of the source. To do this you (simply) add an impedance in the cold leg equal to the output impedance of the unbalanced output. This makes what is called an impedance balanced output.

For inputs you cannot use a balanced connector because it means you will be connecting the cold from the feeding source direct to analogue 0V and many op amp based differential balanced drivers do not like this at all. Personally I would recommend using an isolated connector for the inputs so that if nothing is plugged in at least there is no additional connection form analogue 0V to chassis. There will be when you plug something into the input but that is unavoidable.

Cheers

Ian
 
saint gillis said:
  Hello, suppose we want to make a device with several boards inside, and several inputs and outputs, maybe some of them balanced, some others unbalanced.
  The PSU is connected to the chassis through a ground loop breaker. What is the best way to connect the unbalanced ins/outs ground pins?
It's a tough issue. HiFi manufacturers often solve the problem by making their equipment class 2, i.e. not earthed, which indeed breaks ground loops.
The big problem is when balanced and unbalanced equipment need to be mixed.
There are two constraints that need to be met: breaking the ground loops and making sure the ground cannot be polluted by the shields picking up EMI/RFI.
In order to fulfill the latter, one would hard connect the connector shell to chassis; that would put a constraint on the equipment connected to them to be floating (not earthed).
Others choose to have the connector shell connected to a remote-ground sensing input and RFI/EMI protection be taken care of by a capacitor. Obviously, that requires a special design of the unbalanced inputs (actually, they would be somewhat balanced).
Anyway, I would suggest that making sure the unbalanced equipment is not earthed is the best way to get rid of hum. This must be done in the respect of safety rules. You may want to isolate the chassis from earth, for example with rackmount isolators, but you should provide a direct sparate connection between the chassis and that of the unit it is connected to.
 
The possibilities begin to show up...
For instance a modular synth is a huge ground loop, that's why it works at a big level (+10/14dBu), I guess to attenuate the hum afterwards...
 
saint gillis said:
The possibilities begin to show up...
For instance a modular synth is a huge ground loop, that's why it works at a big level (+10/14dBu), I guess to attenuate the hum afterwards...
Analog synths are notoriously noisy; good operating level is necessary.
 
saint gillis said:
  Hello, suppose we want to make a device with several boards inside, and several inputs and outputs, maybe some of them balanced, some others unbalanced.
  The PSU is connected to the chassis through a ground loop breaker. What is the best way to connect the unbalanced ins/outs ground pins?
Chassis/safety ground is not the same thing as audio 0V while ultimately gets connected somewhere.

Inside product chassis signals are generally handled singled ended (because it's cheaper and easier). I consider each handoff from module to module or even between sections inside a single module as opportunity to forward and back reference different local audio 0V (I don't like calling them grounds) with differential stages.

Often typical active stages can be converted to differential by adding a couple resistors. If you look at schematics inside large consoles you will see many of these convenient differential stages.

JR


 
saint gillis said:
Something like that?

Unbalanced%20Differential%20OpAmp%20Example%204.png
Not bad, except that the pot ruins the hum rejection when it's not full CW. The pot should be at the output.

Maybe without the 10r,
You want the 10r, in case the source is floating, and a cap (0.1uF ceramic) across it for EMI/RFI rejection.

and the ground arrow being the chassis ground?
Correct.
 
saint gillis said:
Something like that?

Unbalanced%20Differential%20OpAmp%20Example%204.png


Maybe without the 10r, and the ground arrow being the chassis ground?

Yes, (lose the 10r) you have pretty much described every console fader gain stage...  As Abbey noted the pot impedance (worst at 50%) will degrade the quality of the differential cancellation.  In general worst case for a 10k fader is +2.5k effectively in series with wiper. By scaling the resistors loading the wiper >> 2.5k will minimize this effect. Of course too much R could increase noise.

The ratio of R1/R2 should track ratio of R3/R4.. BUT R2 does not need to equal R3, the ratios are what matters.

JR 
 
saint gillis said:
The point is not to have the board's ground tied to the chassis ground isn't it?
Not really. The input stage must be referenced to the rest of the circuit. If you make it floating (by omitting the 10r), the ground loop voltage is superimposed to signal.
You should reference the 10r to a clean 0V, and add a capacitor between the connector's "ground" and the chassis ground (earth).
 
abbey road d enfer said:
Not really. The input stage must be referenced to the rest of the circuit. If you make it floating (by omitting the 10r), the ground loop voltage is superimposed to signal.
You should reference the 10r to a clean 0V, and add a capacitor between the connector's "ground" and the chassis ground (earth).
+1...

The point is to realize that ground is not some magical 0V node everywhere it exists, but a conductor with resistance and different voltages due to current flowing within that conductor.

Using differential amplifiers we can ignore these small ground voltages and accurately preserve the signal integrity referencing signals between different local 0V (signal grounds). To reduce RF noise we need to bond shields to chassis ground. With proper differential handling the signal integrity can be maintained despite different ground voltages at different nodes. 

JR
 
Unbalanced%20Differential%20OpAmp%20Example%204.png


Where is the output??

Yes, I see the one dot, but you need TWO connections to pass audio. Return to input jack? Return to triangle?

You can certainly pass audio through a box with NO connection to the box. Pro gear was often done this way: transformers in and out, a jumper to bond/not-bond circuit common to chassis. IMHO the audio should be designed without the chassis. Then as a practical matter you usually must bypass connector shells to chassis for RF with a capacitor. A low-value resistor may be wanted.

As has been said, you can get good CMR without equal-values in the two legs. Look at older Mackie mixers. Many-many interfaces, both external and between boards, had a pair of 100 Ohm resistors bridging the input common to the output common, and the other legs of an opamp picked to cancel.
 
Yes I agree this output is a bit abstract... let's say it's related to the 0V potential of the board containing the components in action, which is tied to the PSU 0V. So it's going to the next stage, and maybe at the end to a balanced output...

So the winner would be the right drawing?
Unbalanced-Differential-Op-Amp-Example-5.png
 
saint gillis said:
So the winner would be the right drawing?
Yes, the one on the right, but the symbols are not what I'm accustomed to. The triangle would be the audio reference (a.k.a. "ground" or 0V), the thing that looks like a rake or a broom would be chassis or earth.
Please note the capacitor should be physically close to the connector, but the 10r can be connected almost anywhere.
 

Latest posts

Back
Top