Hi,
I'm just curious to ask this question ....couldn't find enough information.
We all now that in the most cases a thick closed Groundplane is the way to go (lets restrict us to Analog only without any interfaces to digital). e.g. thick PCB Layer or Copperclad.
What if try I to guide the return current of powertransistors by soldering a solid thick copper wire on the groundplane ,lets say in the middle, to lower the resistance for this path. My guess would be the higher currents might flow more dens in this area due to its lower resistance. So that the "dirty" current is not completly distributed over the groundplane....
Any thoughts on this? I know that this is like the opposite of sloted groundplanes ....but we have no glavanic speration and it is gradually.
Thnx in advance.
Alex
I'm just curious to ask this question ....couldn't find enough information.
We all now that in the most cases a thick closed Groundplane is the way to go (lets restrict us to Analog only without any interfaces to digital). e.g. thick PCB Layer or Copperclad.
What if try I to guide the return current of powertransistors by soldering a solid thick copper wire on the groundplane ,lets say in the middle, to lower the resistance for this path. My guess would be the higher currents might flow more dens in this area due to its lower resistance. So that the "dirty" current is not completly distributed over the groundplane....
Any thoughts on this? I know that this is like the opposite of sloted groundplanes ....but we have no glavanic speration and it is gradually.
Thnx in advance.
Alex
Last edited:
As usual, it depends. 
It might help, but you have to think about closed loops in which the current circulates; currents (return currents) use the path of least impedance, so it is often frequency dependent; at low frequencies it is mainly the path of the lowest ohmic resistance while at high frequencies it is the path of lowest inductance.
In the frequency range between these extremes it is a mix of both.
So if you lower the ohmic resistance in some parts of a ground plane it will help, but it doesn´t guarantee that all return current will take that path, it depends on the entry points and the consecutive selfestablishing current loops.
If any sensible analog circuitry shares (even only partly) the same impedance coupling occures and lowering the resistance will help only up to a certain degree.
Slots in the ground plane prevent this impedance coupling problem by guiding the path of current. Still a lot to consider in that case too.
It might help, but you have to think about closed loops in which the current circulates; currents (return currents) use the path of least impedance, so it is often frequency dependent; at low frequencies it is mainly the path of the lowest ohmic resistance while at high frequencies it is the path of lowest inductance.
In the frequency range between these extremes it is a mix of both.
So if you lower the ohmic resistance in some parts of a ground plane it will help, but it doesn´t guarantee that all return current will take that path, it depends on the entry points and the consecutive selfestablishing current loops.
If any sensible analog circuitry shares (even only partly) the same impedance coupling occures and lowering the resistance will help only up to a certain degree.
Slots in the ground plane prevent this impedance coupling problem by guiding the path of current. Still a lot to consider in that case too.
A thick groundplane is the way to go when you simply want to minimise resistance and inductance, but don't really care where the currents go. An alternative is to carefully consider where the currents go and ensure they only go where you want them to go - this leads to some sort of bus ground. Somewhere in between is the star ground - which can be considered to be a short unplanned bus.
The best way (at HF) to guide a current through a plane is to closely place a conductor carrying the return current over the intended path: for example the hot wire and GND to a load, or the collector and emitter of a transistor.
This will work down to the cut-off frequency of the transmission line thus formed.
For lower frequencies, the currents will follow laws of diffusion, and controlling the resistivities is the only mean of guiding it (using your suggested method for example)
This will work down to the cut-off frequency of the transmission line thus formed.
For lower frequencies, the currents will follow laws of diffusion, and controlling the resistivities is the only mean of guiding it (using your suggested method for example)
Current will always try to take the lowest impedance path. For DC, resistance dominates. At high frequencies, the lowest resistance path may not be the lowest impedance path. At frequency, the current will try to minimize the total loop path.
If for example, you have a supply rail trace on one edge of the ground plane, the return ground current will tend to concentrate on that side of the plane.
Same with discrete wires. If you run a supply rail wire along the metal of the chassis and use the chassis as ground return, the return for that rail will take the chassis route closest to the feeding rail wire.
Jn
Ps. And as Cliff states, the current will divide according to some rules..
If for example, you have a supply rail trace on one edge of the ground plane, the return ground current will tend to concentrate on that side of the plane.
Same with discrete wires. If you run a supply rail wire along the metal of the chassis and use the chassis as ground return, the return for that rail will take the chassis route closest to the feeding rail wire.
Jn
Ps. And as Cliff states, the current will divide according to some rules..
Could this also mean that STP is better than coax for single ended connections?
Ah.
STP allows two power conductors so can be quite better than coax for outputs. Just means isolating speaker ground so the shield can return all the load current.
Jn
For an unbalanced interconnect between two individually AC powered units, a coax is better than a STP as it probably will have less Common Impedance Coupling noise.
Yes, not ALL the current will take 'the lowest Z path'. It splits among ALL possible paths according to the impedances.
Current is like cars. In my neck of the woods you can take Rt 3, fairly straight and smooth, or Old Oak Road which is narrow and twisty; both connect the same two main points (employment and beer). While Rt 3 has a lot of cars, Oak is far from empty, and tends to run 30% of the traffic on the highway.
Or a water supply. Say you run three big sprinklers and a small water fountain. Water flows at all four paths. More on the lawn, less in your face, depending on pipe/hose/valve sizes.
Adding "a solid thick copper wire on the groundplane" *does* reduce the total impedance. If, in addition to Rt 3 and Oak Rd, we built a 6-lane 85MPH highway through my neck of the woods, it would not take an hour to get beer, maybe only 20 minutes. And less back-up on the roads. And easier for me to cross the road to my mailbox with the super-highway taking much of the traffic.
(Actually, traffic is a poor analogy. In electric stuff, the forcing voltage is constant or nearly so. All experience shows that if you build a road twice as good, three times as many cars show up.)
So far, elvee, cliff, you and I have all said this...I think we agree, no?
Edit: I think some may be confused between lowest z path and multiple paths as stated. Discussion of potential always helps.
Within a wire pair for example zip, as frequency goes up, the lowest impedance path is actually closest to the other wire. The current centroids are trying to occupy the same space, but cannot leave the confines of the conductor.
A single wire as elvee stated, if run along the chassis, will cause the chassis return current to get as close to the single wire as it can, determined by freq and impedances. Oddly enough, this will occur in the upper audio band. Bass would follow a beefier copper run.
jn
Last edited:
What should be of interest is the magnetic fields produced by the wires. (Being a magnet man, you had to realize I'd somehow shoehorn that stuff in, no?)
If the return path is augmented such that there is a frequency dependency on the return current, then the magnetic field generated would have a strange interaction with any low level circuits, the input pair for example. Bass would couple more because the path is split, hf return centroid would drift towards the rail send (or worse,two rail sends that may not be together).
Even more complex is where the current goes during four quadrant operation, like every speaker on this planet forces.
Jn
If the return path is augmented such that there is a frequency dependency on the return current, then the magnetic field generated would have a strange interaction with any low level circuits, the input pair for example. Bass would couple more because the path is split, hf return centroid would drift towards the rail send (or worse,two rail sends that may not be together).
Even more complex is where the current goes during four quadrant operation, like every speaker on this planet forces.
Jn
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.