Blue is bottom layer, red is top layer (see images' names).
I tried to keep almost all power traces on one single layer (bottom layer) to allow one undisrupted ground plane (top layer).
The bottom layer is indeed disrupted (and could not be otherwise expect going with an expensive 4-layers board). I could get rid of the bottom ground plane, but continuity is still ensured (no "dead copper" areas).
My view is this: top ground plane ensures optimal return current path (chances are that - return currents going through paths of least impedance - , return currents will not be diverted because of the disrupted bottom gnd plane) , while the bottom gnd plane can still has a beneficial shielding effect. That is why I chose to keep it.
I tried to keep almost all power traces on one single layer (bottom layer) to allow one undisrupted ground plane (top layer).
The bottom layer is indeed disrupted (and could not be otherwise expect going with an expensive 4-layers board). I could get rid of the bottom ground plane, but continuity is still ensured (no "dead copper" areas).
My view is this: top ground plane ensures optimal return current path (chances are that - return currents going through paths of least impedance - , return currents will not be diverted because of the disrupted bottom gnd plane) , while the bottom gnd plane can still has a beneficial shielding effect. That is why I chose to keep it.
Last edited:
You should have the Power Plane and the Power Ground Plane next to each other. These should be adequately decoupled often and at significant current sinks.
You can then add a signal Ground Plane next to the Signal Layer.
H.Ott suggests that 6 layers is the minimum required to get planes to work properly at HF. He even suggests that 8 and 12 layers are probably better at using the Planes.
I do not think that adopting too few layers/Planes for LF signals (audio amplifiers) achieves anything of significance.
Much more important, in my view, is to control all the circuit routes so that you do not mix signals along the same trace/route while keeping loop areas very small.
You can then add a signal Ground Plane next to the Signal Layer.
H.Ott suggests that 6 layers is the minimum required to get planes to work properly at HF. He even suggests that 8 and 12 layers are probably better at using the Planes.
I do not think that adopting too few layers/Planes for LF signals (audio amplifiers) achieves anything of significance.
Much more important, in my view, is to control all the circuit routes so that you do not mix signals along the same trace/route while keeping loop areas very small.
perfect nice layout. If you use a gnd plane on both sides you can connect the open areas of the upper plane with vias to the lower ground plane.
Sorry for my criticism ,but I would recomend to put a resistor in the baseline of Q9 and Q10 because those poor trannies are tortured with a probably lethal bias-currant pulse in certain conditions. This currant comes through the both electrolytics one to plus the other to gnd.
e.g. when power up or a short break in the raw power. 1kilo ohm or so could help to prevent damage
Ingo
i
e.g. when power up or a short break in the raw power. 1kilo ohm or so could help to prevent damage
Ingo
i
I'm a it puzzled about the pins arrangements on both Q3, Q4 and Q5, Q6 - if the transistors face each other on opposite surfaces of the heatsinks, shouldn't the pin configuration be reversed also?
If this Regulator doesn't have a 'pre-filter' (maybe a C-R-C, etc) then the diode charging pulses are going to be transferred to the ground pin of X1 input power connector and hence flood thru all those ground planes both top and bottom and infiltrate into both the voltage references and the feedback loops - not a recommended scenario (IMO, naturally!)
I would suggest you provide separate tracks for the different currents even at the expense of some of those nice ground planes and particularly if the output terminal is the point where the reference ground connection is made for the following circuit.
There are various excellent ways of designing power supply regulators and this is just another option that I usually find gives better results despite the loss of continuous ground planes and high freq noise attenuation.
... just my 2 cents.
If this Regulator doesn't have a 'pre-filter' (maybe a C-R-C, etc) then the diode charging pulses are going to be transferred to the ground pin of X1 input power connector and hence flood thru all those ground planes both top and bottom and infiltrate into both the voltage references and the feedback loops - not a recommended scenario (IMO, naturally!)
I would suggest you provide separate tracks for the different currents even at the expense of some of those nice ground planes and particularly if the output terminal is the point where the reference ground connection is made for the following circuit.
There are various excellent ways of designing power supply regulators and this is just another option that I usually find gives better results despite the loss of continuous ground planes and high freq noise attenuation.
... just my 2 cents.
In a competently designed multi-layer PCB, the ground to power planes act as capacitors.
If the pulses are VHF, the capacitance of the planes smooths the ripples/spikes.
But a two layer PCB can never approach the capacitance values of a multi-layer board where the ground planes and power planes have separations of typically 10thou (10mils, 0.25mm)
If the pulses are VHF, the capacitance of the planes smooths the ripples/spikes.
But a two layer PCB can never approach the capacitance values of a multi-layer board where the ground planes and power planes have separations of typically 10thou (10mils, 0.25mm)
- Status
- Not open for further replies.