The yet tiniest single-sided LM3886?

[carlmart]

Originally Posted by hurtz

To increase RC-filter characeristics.

You have only one RC filter, if I see it right.

So you can fill the two areas on the left and two on the right and lower copper resistance.

[hurtz]

Every trace between each capacitor has a resistance and this is a characteristic I would like to keep because it will (at least in theory) reduce ripple. I see less benefit in copper pour areas to decrease resistance in the rail-traces.

cheers,
hurtz

[AndrewT]

Yes, I agree. Gootee has simulated a bit on this and the results of the sims seem to fit with what happens on a real PSU.

[Michael Bean]

Hurtz, regarding this statement;

Every trace between each capacitor has a resistance and this is a characteristic I would like to keep because it will (at least in theory) reduce ripple. I see less benefit in copper pour areas to decrease resistance in the rail-traces.

Did you do the math? If you would, I believe you will find that the tiny little bit of additional resistance will be more than off-set by the additional inductance that will also be created. The best way to impliment a R/C filter in a known and controllable manner is to use low-inductance power resistors.

Mike

[AndrewT]

I disagree.
If you have a capacitor following an impedance, either a resistance or an inductance combined with a resistance, then you must have a Low Pass Filter.
You cannot avoid that LP filter effect. You can change the effect, but not eliminate it.

Use the LP filter to your advantage. It has a strong effect on the ratio of high frequency garbage to low frequency garbage that gets through to the amplifier.

[Michael Bean]

Did you do the math?

Mike

[AndrewT]

Not quite, but Gootee has.

[Michael Bean]

So did I, many years ago. And I found that in general, the additional trace inductance will swamp any possible benefit provided by the very tiny amount of trace resistance that using this methode will provide. If the traces are the equivalent of 24AWG and one foot long, you will get 0.02567 ohms for that. The amount of additional circuit inductance will of course depend on the physical layout, but will almost certainly nullify the (tiny) benefit the additional resistance would provide. One could allways make the PC traces even smaller for more resistance, but at some point, the traces become "fuses, ready to blow".
I'll say it agian, the best way to impliment R/C filtering in your power supply is to use actual resistors and capacitors.

Mike

[hurtz]

The PCB traces are not meant to provide proper RC-filtering, the decrease in resistance using copper pour areas is as negligible as the increased RC characteristics.
But for completeness, solder-joint resistances have to be taken into consideration as well. And the increase of the trace width i.e. pour areas will only reduce the initial inductance of a few nH by a few more.
If doing the math neither option has any significant impact on the characteristics of the whole PSU.

So, as to why I'm not using copper pour areas is to increase RC-filter characteristics and I'm only stating that this applies to the theoretical model.

cheers,
hurtz

[AndrewT]

Originally Posted by Michael Bean

......... the best way to implement R/C filtering in your power supply is to use actual resistors and capacitors.........

I prefer [L+R] C filtering. The L comes free with the inductance of the connection lengths. The damping R also comes free with the connection lengths. The C is paid for and chosen appropriately.

This is extra filtering that comes along for the ride, without adding any components.