Incoming ground isolation

[armdnrdy]

I've noticed in the design of some power supplies that there is an order in which the ground is routed through initial filter caps, regulators, post regulation caps, and finally out to the load.

The incoming ground is kept separate with it's own trace until it reaches the ground plane and/or load.

Now I have also noticed routing that includes all of the above power supply components and the incoming ground positioned on the same ground plane.

It seems that it would be best to take the power and ground after all of the filtering and regulation.

Any thoughts.

[R.G.]

I've noticed in the design of some power supplies that there is an order in which the ground is routed through initial filter caps, regulators, post regulation caps, and finally out to the load.
The incoming ground is kept separate with it's own trace until it reaches the ground plane and/or load.
Now I have also noticed routing that includes all of the above power supply components and the incoming ground positioned on the same ground plane.
It seems that it would be best to take the power and ground after all of the filtering and regulation.

Ah.

Lecture #37: All Wires Are Really Resistors

Remember our discussion about reference ground and sewer ground? Power supplies are particularly susceptible to ground noise because there are big pulses of current coming into the rectifers and into the filter caps.

See http://www.geofex.com/Article_Folders/Power-supplies/powersup.htm for more info on this.

The big pulses of current go through the wires from the rectifiers, into the filter caps, then back through the wires from the (-) side of the filter caps back to the transformer. These pulses cause voltages in the wires. That doesn't seem like it should be that bad, but the nature of rectifiers and capacitors mean that the current pulses charging the caps may be many, many times larger than the DC current out of the filter caps.

You are right - the incoming power should come directly to the first filter cap, and only there. Then the "ground" wire from the first filter cap should go off to the circuit, including jacks, grounds, and ground planes. If you don't do it that way, the currents coming in on the ground wires to the filter caps cause voltages that wiggle the reference ground around, and you hear it.

[armdnrdy]

Thanks R.G.

I looked at my layout and didn't like what I saw. The onboard power jack's ground pin was connected to the ground plane.
That routing would have allowed a number of "grounds" to take a path directly to the jack's ground pin.

Made the change.
Trying to cover all angles!

[Govmnt_Lacky]

I am still having a hard time wrapping my head around the whole "different grounds" opinion.....

Bottom line is that you have a built pedal. It NEEDS to have a V- and V+ input or else it will not work. So, by this process, doesn't that mean that ALL of the "grounds" in the pedal HAVE TO EVENTUALLY got routed to the POWER ground?

I mean... its not like the pedal is being fed with multiple Ground options. You have a power adapter (be it a wall wart, battery, etc) and these only have (1) V+ and (1) V- right?

I just dont get why ALL of the grounds WOULD NOT be tied together to the V- coming from the source  

You are right - the incoming power should come directly to the first filter cap, and only there.

Was this a typo/boo-boo? Reading this makes me wonder... "How is the CIRCUIT going to be powered if the incoming power supply (V+) ONLY goes to the first filter cap and ONLY there? 

[armdnrdy]

I am still having a hard time wrapping my head around the whole "different grounds" opinion.....

Bottom line is that you have a built pedal. It NEEDS to have a V- and V+ input or else it will not work. So, by this process, doesn't that mean that ALL of the "grounds" in the pedal HAVE TO EVENTUALLY got routed to the POWER ground?

I mean... its not like the pedal is being fed with multiple Ground options. You have a power adapter (be it a wall wart, battery, etc) and these only have (1) V+ and (1) V- right?

I just dont get why ALL of the grounds WOULD NOT be tied together to the V- coming from the source  

You are right - the incoming power should come directly to the first filter cap, and only there.

Was this a typo/boo-boo? Reading this makes me wonder... "How is the CIRCUIT going to be powered if the incoming power supply (V+) ONLY goes to the first filter cap and ONLY there?  

For the first part....read through some of this thread:
http://www.diystompboxes.com/smfforum/index.php?topic=104370.0

I read R.G.s response as:

Don't connect a power trace from...say the input power jack directly to an op amp power pin. The power and ground should be taken AFTER the filtering, for reasons he outlined.


"the incoming power should come directly to the first filter cap, and only there."

The power after it leaves the filter cap would be considered the "outgoing" power or "the load".

[defaced]

I am still having a hard time wrapping my head around the whole "different grounds" opinion.....

Bottom line is that you have a built pedal. It NEEDS to have a V- and V+ input or else it will not work. So, by this process, doesn't that mean that ALL of the "grounds" in the pedal HAVE TO EVENTUALLY got routed to the POWER ground?

I mean... its not like the pedal is being fed with multiple Ground options. You have a power adapter (be it a wall wart, battery, etc) and these only have (1) V+ and (1) V- right?

I just dont get why ALL of the grounds WOULD NOT be tied together to the V- coming from the source  

You are right - the incoming power should come directly to the first filter cap, and only there.

Was this a typo/boo-boo? Reading this makes me wonder... "How is the CIRCUIT going to be powered if the incoming power supply (V+) ONLY goes to the first filter cap and ONLY there? 

Check out the Amazon and Google Books preview of Tim William's Circuit Designer's Companion.  The first chapter is on grounding and provides a very good break down of why grounds need to be managed the way they do.  It is the long form explanation to R.G.'s statement "all wires are resistors".

I also found Merlin's (Valve Wizard) PDF on grounding to be helpful in seeing illustrations of different ground schemes and why some are better than others.  Yes, its geared toward tubes, but the theory of operation is the same regardless of the active device. 

I think a little reading on those two sources will help answer your questions about "why should this be tied tot he cap negative terminal".  The answer is that it's the lowest impedance point along the power rail, but that doesn't really explain why that matters which is what those two resources cover. 

[armdnrdy]

Thanks for the info Mike.

I'll take a dive in!

[Govmnt_Lacky]

- I understand that all wires are resistors as well as capacitors
- I understand that there are, in fact, different "types" of Grounds (V-)

What I dont understand is WHY anyone disputes that ALL grounds in a circuit inevitably end up being tied to POWER GROUND.

Whether it be signal ground, power ground, or sewer ground....... it all only has ONE place to go in the end in order for the circuit to work. The POWER GROUND that lies on the wall wart plug or the battery.

Otherwise.... wouldn't it just be a floating ground that is not connected to anything 

[R.G.]

Bottom line is that you have a built pedal. It NEEDS to have a V- and V+ input or else it will not work. So, by this process, doesn't that mean that ALL of the "grounds" in the pedal HAVE TO EVENTUALLY got routed to the POWER ground?
I mean... its not like the pedal is being fed with multiple Ground options. You have a power adapter (be it a wall wart, battery, etc) and these only have (1) V+ and (1) V- right?
I just dont get why ALL of the grounds WOULD NOT be tied together to the V- coming from the source  

They are eventually connected together, but exactly HOW they're connected together matters; don't get discouraged, this is a slippery subject.

First, wires are not wires as most people think of them. They are really, truly, no-fooling simply low-value resistors. Experimental superconductors aside, there are NO connections without at least some resistance. The only difference is how much resistance. Some "wires" have 1M between their ends. We use these for pulldowns. Some have 1K between their ends. We use these to limit currents in LEDs and transistors. Some of them have 10 ohms from end to end, and we use these to confuse the competition. Some wires have 100, 10, 1 or 0.01 MILLI-OHMS from end to end, and out of frustration at the difficulty at measuring the voltage drop across these last ones, we say "oh, they're all short circuits" and smile.

But the fact is, all wires are different. They all have different resistances, inductances, capacitances, and so on. So the only way there is ever 0.0000000000... volts across a wire connecting two points is when the current is equally zero. Georg Ohm said. God told him. They're all different.

So how do you EVER get two different places to have the same voltage? Only two ways: (1) you move them together so they're no longer separated by a wire/resistor or (2) you force the current between them to be zero. If the current is zero, there can be no difference in voltage between two points connected by a conductor. One really good way to make sure there's no current between two points is to connect them with one and only one wire. Electrical current flow requires a loop to flow in, and if there is no current loop made with conductors, no current can flow through the conductors, so there voltage between the points MUST be zero.

So - all wires are resistors. Any current flow makes a voltage across them. If you want two voltages to be the same, there can be no current flowing between the places the voltages are.

Power comes in at a power jack (as DC, usually, for pedals) or as AC to be rectified. If it's rectified, there are large, discontinuous pulses of currents set up to charge the filter cap, and so the wire from the power-in jack is polluted with these pulses, as is the wire from the rectifiers to the first filter cap (which is assumed to be so big that it eats most of the pulses and makes what goes out into smooth DC.) If you attach anything in your circuit that needs to be a quiet place of "ground", like the ground lead of your input jack, then it needs to attach to either the circuit itself and not to the enclosure or filter cap, or to the filter cap "common" point only. This is a reference point, and the currents are going to be very small. Your circuit may pull a lot or a little current, and that can be a continuous, non-varying current or it may wobble like the devil with signal peaks. If it wobbles, it needs to be returned to the filter cap common by a wire that does not include the input voltage reference, or the wiggles on the wire caused by the current and wire resistance will cause your input to think the wire-voltage is part of the signal, and amplify it.

The limit in the extreme is star grounding; everything that could possibly have a ground symbol on it gets run to the filter-cap common by its very own wire. Now the wires cannot cause shared voltages from the currents in the wires, because they all have their own wire, and so they cannot pollute one another. This obviously gets ugly if there are lots of points to star ground.

The limit in the opposite extreme is plane grounding. In plane grounding, one covers the entire circuit (metaphorically!) with a bedsheet of solid copper in hopes that (1) the total resistance is a lot lower because you've used so much copper and (2) that the currents will wend their merry way across the copper plane in ways that don't insert ground return voltages into the signal. This does work very well at RF, where the electromagnetic fields of the signals force the currents to flow under the signal wires going the opposite way in the loops. It works less well at DC and audio where the field component directing the current flows are less strong.

I'm just running on about some grounding issues. The central idea is that all wires are resistors, and you have to pick and choose what currents run in what wires to make things be quiet. If you are doing small circuits and low currents, the induced voltages may be so small you can ignore the grounding issues. This is how most pedals work.

These issues are why I periodically post about what a bad idea it is to use the input jack for carrying all the current from the battery negative to the circuit. The stereo jack trick works and people get away with it under the "small circuits and currents" exemption, but the higher the gain and the higher the current, the less good running all your power supply current through your input signal ground wire becomes.

Was this a typo/boo-boo? Reading this makes me wonder... "How is the CIRCUIT going to be powered if the incoming power supply (V+) ONLY goes to the first filter cap and ONLY there? 

What ANR said - one wire goes from the power in to the first filter/bypass cap. Then another wire goes to the circuits, or they all cluster there if you're star- or semi-star-grounding.

What I dont understand is WHY anyone disputes that ALL grounds in a circuit inevitably end up being tied to POWER GROUND.
Whether it be signal ground, power ground, or sewer ground....... it all only has ONE place to go in the end in order for the circuit to work. The POWER GROUND that lies on the wall wart plug or the battery.
Otherwise.... wouldn't it just be a floating ground that is not connected to anything 

No one disputes that. It's just a slippery concept to talk about with words only and not pictures. Yes, eventually everything does get connected to power ground, and all the "used electrons" get back there or nothing would work. The points being raised are just to emphasize that the PATHS the current takes have consequences, but they all do go back to the electron-pump that makes them flow.

[armdnrdy]

Great explanation R.G.

This is one for my folders!

[defaced]

What I dont understand is WHY anyone disputes that ALL grounds in a circuit inevitably end up being tied to POWER GROUND.

I think if someone is proposing that they are in error.  I think this is an issue of people getting stuck on terminology.  Yes, all grounds are connected - you are absolutely correct.  However, as R.G. explains above, the way they are connected together is significant.  This means we as humans want to differentiate between them and call them names so it's easier for us to communicate.  Power ground.  Signal ground. Earth ground. Switching ground.  LED ground - you get the point.  So one could just as easily re-write that sentence with any of the listed ground names and be equally true.  The problem is this conversion from technical communication to casual communication usually leaves artifacts and simplification that aren't always correct when run back through the "technical communication filter". 

[tubegeek]

Well said.