Grounding 101

[ruffrecords]

It's true, I am trying to use a guitar amp transformer. 

On this post http://www.groupdiy.com/index.php?topic=42745.40 ruffrecords raises concerns about having the CT of the HT transformer connected to the chassis at one point while pin 1 of the XLR connects to the chassis at a different point.

Is it acceptable to run wires from pin 1 on the XLRs to the "big bolt" rather than connecting them to the chassis right near the connectors?

Thanks!

Not really. The problem is the phantom power should really be taken from a completely separate winding from the other supplies so sharing the centre tap of the mains transformer is breaking the rules. Therefore whatever you do will be something of a compromise.

Cheers

ian

 

[earthsled]

Let me back up for a moment.

In an ideal situation (when the 48V supply has a dedicated secondary winding), where should 0V of the 48V PSU be connected? Should it go the "big bolt" chassis connection described earlier, or is there a better location for the connection?

Thanks!

 

[ruffrecords]

Let me back up for a moment.

In an ideal situation (when the 48V supply has a dedicated secondary winding), where should 0V of the 48V PSU be connected? Should it go the "big bolt" chassis connection described earlier, or is there a better location for the connection?

Thanks!

That is the right question to ask. The basic rule is that the chassis and cable screens should not carry 0V. So you connect screens to the chassis and connect the 0V to the chassis at one point (the big stud) only as discussed earlier.

Trouble is, phantom power breaks these rules because its signal 0V is the screen of the mic cable which we should connect direct to the chassis. But this means we are potentially connecting a 0V to the chassis at a second point which is a recipe for ground loops.

The solution has to be a compromise and there are two choices.

1. Connect mic XLR pin1 to chassis at the XLR and connect phantom power 0V to chassis at the big stud.  The problem with this is that all the interference currents from all sources flowing in the chassis to the big stud are now flowing in our mic input 0V - we don't really want all interference currents flowing in perhaps the most sensitive 0V in the whole system.

2.  Connect mic XLR pin1 to chassis and run a separate phantom power 0V up to the mic inputs and bus it along the pin 1 connections with a thick piece of copper wire. Any interference current flowing in the mic cable screen also flows in the phantom 0V but that is all and there is nothing we can do about them - the mic input common mode rejection is supposed to take care of that sort of thing. All other interference currents flow down the chassis to the big stud and do not induce current in the phantom 0V.

So what I do is option 2. I connect all the mic XLR pin1s to the chassis at the point of entry and I connect them all together with a neavy guage copper wire. I connect the phantom power 0V to the centre of this copper wire and nowhere else.

Cheers

Ian

 

[PRR]

The bias-tap for mikes really foils a sane "grounding scheme".

The basic plan ought to be:

 

[earthsled]

The bias-tap for mikes really foils a sane "grounding scheme".

Agreed. I've changed my plan to a custom wound power transformer instead of the one with the bias tap. Nice to know Edcor makes this sort of thing affordable!



Attached is a diagram illustrating option "2" suggested by ruffrecords.

 

[CJ]

you can also put in a ground lift switch between pin 1 and the chassis.

these are handy when stringing lots of gear together in series, you can flip switches to get the lowest hum,

 

[earthsled]

Attached is a diagram in which a single transformer and rectifier are used for a dual-rail power supply. In this situation, would it be best to return each 0V to chassis, or would this cause a loop since the 0V paths are already connected at the transformer CT?

Thanks!

 

[abbey road d enfer]

Your scheme is almost correct. The xfmr center-tap and smoothing caps are closely tied together but left floating. The junction of pos and neg smoothing caps goes to a common regulator ground reference, which in turn goes to the chassis.

 

[JohnRoberts]

Why have dual grounds? By definition they will be different from each other. Which one is the real ground 0V reference?

For FW rectification, connecting the ground terminals of the two reservoir caps together gives the two charging currents a simple return path to each other. The path to CT is more of a reference that carries only differences in + and - charging current. A  common point between the two caps can become the PS GND node, to forward reference to the regulator grounds, and return circuit ground flows back to.

Note: for half wave rectification, the CT current is much higher so that path impedance matters and should be kept low.

JR

 

[earthsled]

Thanks very much for the quick replies!

Here is my best attempt at illustrating Abbey's comments. For my application, the rails are to be identical positive supplies (not sure if this makes a difference compared to pos / neg configs).

 

[earthsled]

...and this one attempts to include JR's comments....

Am I on the right track?

 

[abbey road d enfer]

Basically, JR and I say the same and both your drawings end up being the same in practice.

 

[earthsled]

Okay great! 

I find it interesting that my PCB layout software is reluctant to allow a common ground like this. Since the two PSU grounds are already electronically tied together at the CT, joining the filter caps and regulation together makes for multiple "loops" according to my software. Working around the issue is no problem by adding zero-ohm jumpers or the like. I just thought it was something worth pointing out for future DIYers who might attempt this.

Cheers!

 

[JohnRoberts]

While the difference is subtle and depends a little on the execution I favor the shorter path between the two cap grounds.

If you visualize the current flowing out of the ground terminals of the two reservoir caps, and into each other, in one schematic that current travels over the two paths to the reference ground, in the JR version that current is kept local between the two caps, and the one ground connection to the regulators only carries regulator ground current.

If all the traces are kept short this isn't a huge difference, but I prefer to look at layouts in terms of the current flowing in traces.

=====

Not sure I understand your software's loop concerns. There should be a single connection to the transformer CT and that goes to the common between the two reservoir caps, likewise only a single connection from the PS regulator ground over to the common capacitor gnd. A loop requires multiple paths and a subsequent loop area defined. 

JR

 

[abbey road d enfer]

v2 is not a correct representation of what I said. V3 is; that's because JR and I are saying the same thing, in a different manner.

 

[earthsled]

...Use hierarchical ground;  input stage gnd to second stage gnd to third stage gnd and so on.
The final stage ground should return to a single gnd of the PSU, typically the ground that serves as a reference for the regulators (for tube gear to the ground of the last cap in the chain of smoothing caps)...

Considering this, would I be correct in thinking that the best place to connect the PSU's 0V  to the audio circuit would be the primary of the output transformer (in a mic pre, for instance)?

Thanks!

 

[abbey road d enfer]

That is correct if the primary is connected to ground, but more generally it should be returned to the ground relevant to the cathode of the output stage.

 

[CJ]

<that is correct if the primary is connected to ground...>

never seen a circuit with the pri of the output trans connected to ground,

 

[JohnRoberts]

Think like the current....  8)

JR

 

[abbey road d enfer]

<that is correct if the primary is connected to ground...>

never seen a circuit with the pri of the output trans connected to ground,