# Can a capacitor decouple into floating ground?

#### coinmaster

Joined Dec 24, 2015
502
Can you decouple AC into a floating ground via a capacitor?

#### nsaspook

Joined Aug 27, 2009
8,181
Describe your 'floating ground'. A circuit diagram would be nice.

#### crutschow

Joined Mar 14, 2008
26,848
Not enough info.

#### MrAl

Joined Jun 17, 2014
8,073
Can you decouple AC into a floating ground via a capacitor?
Hi,

If i understand you right then the answer is either yes or no depending on what you need to decouple.
That is because the AC part of the signal gets decoupled normally but any transients can cause large currents which can be objectionable.

More information about what you want to decouple would be a good idea, but for now i assume you have one cap to decouple the signal and one cap to decouple the ground as in the following circuit:

SIG IN o----C1-----o SIGNAL OUT
GND IN o----C2----o GND OUT

As you might expect, if the input GND is at a very different DC level than the GND OUT when power is first applied, there could be a large transient from GND in to GND out which can cause problems.

#### coinmaster

Joined Dec 24, 2015
502
I guess the short version of what I'm asking is can I bypass the voltage rating of a capacitor in a ripple filter by filtering into floating ground?

#### crutschow

Joined Mar 14, 2008
26,848
Well, we've had the short version twice and we still don't know what you mean by a "floating ground".
Floating from what?
We need the long version of what you are asking or, better yet, a schematic.

#### nsaspook

Joined Aug 27, 2009
8,181
I guess the short version of what I'm asking is can I bypass the voltage rating of a capacitor in a ripple filter by filtering into floating ground?
I still unclear about the nature of this floating ground? In order for the bypass (or decoupling capacitor) function to work there must be a complete circuit of some sort so energy can move into that alternative (shunted through) usually low impedance path instead.
http://www.designers-guide.org/Design/bypassing.pdf

You can't (safely) bypass the voltage rating of an individual capacitor but you can arrange a circuit with multiple series capacitors to extend the voltage raging of the stack while keeping each individual capacitor within it's voltage rating.

http://electronicdesign.com/compone...r-strings-overvoltage-yet-maintain-efficiency

#### coinmaster

Joined Dec 24, 2015
502
Well seeing how ltspice does't allow floating circuits I can't post a schematic. Not sure how I can explain it any more simply.
Capacitors are used in power supplies to shunt ripple currents into ground. Capacitors have maximum voltage ratings. I'm asking if I connected the capacitor to a ground that is floating relative to the rest of the circuit will the ripple currents still be passed through to this floating ground without surpassing the voltage rating of the capacitor.

I ask because I would like to use low voltage caps in high voltage supplies and it's always a challenge having to design around multi hundred dollar caps.

The nature of floating electricity still escapes me since for example DHT floating filament currents can modulate audio currents and can create a voltage drop relative to earth even though it is not earth referenced, so much doesn't make sense with floating electricity. I'm hoping maybe capacitors can pass current into a floating ground even though there is no complete circuit.

#### crutschow

Joined Mar 14, 2008
26,848
You can draw a circuit in LTspice without a ground reference, it just won't simulate properly, so please draw it and post it.

There's nothing special about ground, it's just a reference node for voltages in a circuit, noting else.

Certainly you can have a floating node from your reference ground, but then you can't use a voltage from that floating node circuit to power something that is referenced to your ground node and not back to you floating node (since then it's no longer floating).

It boils down to, all currents have to have a complete path from the source and back to the source.

#### coinmaster

Joined Dec 24, 2015
502

Ignore the values. The AC power sources signify the AC from the wall. Even though the bottom circuit is floating the current still returns to the wall through the transformer does it not?

#### nsaspook

Joined Aug 27, 2009
8,181
Well seeing how ltspice does't allow floating circuits I can't post a schematic. Not sure how I can explain it any more simply.
Capacitors are used in power supplies to shunt ripple currents into ground.
This is where IMO you are going conceptually wrong (with it's origins in the water analogy). The currents don't flow into a ground reference point like water out of a facet never to be seen again or fill up a tub. The electric current behaves more like links in a bicycle chain that move in a closed loop to transport energy from source to load. The power supply ripple currents must have a closed loop to return to the their source for the shunt to work.

#### nsaspook

Joined Aug 27, 2009
8,181
View attachment 113713
Ignore the values. The AC power sources signify the AC from the wall. Even though the bottom circuit is floating the current still returns to the wall through the transformer does it not?
The only return path for C8 current would be in series to the very small L1:L2 transformer coupling capacitance or leakage currents. The Galvanic isolation breaks the current loop 'chain'.

Last edited:

#### BR-549

Joined Sep 22, 2013
4,938
You should breadboard a low voltage proto. Then measure the voltage across the cap and also the current thru it.

#### wayneh

Joined Sep 9, 2010
17,073
The only return path for C8 current would be in series to the very small L1:L2 transformer coupling capacitance or leakage currents. The Galvanic isolation breaks the current loop 'chain'.
I agree. The L2 circuit could be at any voltage compared to the L1 side and it wouldn't matter. That's the beauty of isolation! The capacitor would change that relative voltage by a small ±∆V, and it still wouldn't matter to an ideal circuit.

#### crutschow

Joined Mar 14, 2008
26,848
To summarize, there is no significant current through C8 and it has no effect on the circuit since there is no complete current path.
You need to understand what NS said about current in post #11.
Once you do, you will see why your circuit won't work as you wanted.

#### ErnieM

Joined Apr 24, 2011
8,126
You absolutely cannot connect a cap from points A to C and expect it to filter from A to B.

#### ronv

Joined Nov 12, 2008
3,770
View attachment 113713
Ignore the values. The AC power sources signify the AC from the wall. Even though the bottom circuit is floating the current still returns to the wall through the transformer does it not?
You can model it in spice by adding a 1 gig resistor to the bottom of each side of l1, L2.

#### crutschow

Joined Mar 14, 2008
26,848
The AC power sources signify the AC from the wall. Even though the bottom circuit is floating the current still returns to the wall through the transformer does it not?
It does not.
The longer answer is, the current does not "return to the wall"; any current going through the top of the secondary also comes out the bottom of the secondary.
No significantly current goes from the secondary to the primary.
In a normal power transformer the primary and secondary are two separate turns of wire and are galvanically isolated from each other.
The only connection between the two is through magnetic coupling from the magnetic field in the core.

#### coinmaster

Joined Dec 24, 2015
502
What if I did this

That would give the transformer 100% leakage current while still maintaining floating isolation, maybe?

#### nsaspook

Joined Aug 27, 2009
8,181
What if I did this
View attachment 113744
That would give the transformer 100% leakage current while still maintaining floating isolation, maybe?
You are wasting yours and our time. The current path that actually allows C8 (or some equivalent total capacitance) to filter the ripple current from the positive side of the bridge needs to be directly connected to the negative side of the full-wave bridge to be effective.