Power Filter Placement - Does it matter?

[mattoverse]

This is probably a dumb question and I feel that I know the answer but wanted to double check with those who are more knowledgeable:

When using a simple power filter circuit, ie reverse polarity protection diode, 100 nf cap & 100uf cap ripple/filter caps, does it matter where in the circuit these components are placed between supply and ground?

My gut tells me that it shouldn't matter, as ground is ground and supply is supply, but just wanted to make sure that power and ground connections for the IC's etc.. should not all be downstream (is it even possible for something to be downstream?) from the filter?

For example, supply of 9v+ enters the circuit goes across the filter before being fed to the IC's etc?

Also, any issue with placing IC power and ground pins right on the power and ground rails, or should they be jumpered?

[mth5044]

That's a good question. I would imagine you'd want filtering as soon as the power hits the board, but like you say, is 'downstream' really a thing?

The only bit of filtering I know for sure is that you want the bypass caps for IC's to be as close to the IC pins as possible. You don't see it too much, but in a lot of other audio circuits, IC's tend to have a 0.1u cap from positive to negative rail (9V to Ground, 15V to -15V,etc). It's always stipulated that this cap should be as close to the pins as possible, even some saying you should solder the cap to the pins.

I'm going to guess in pedal world, it doesn't matter, like a lot of things that manage to just get swept under the carpet. But who knows? Perhaps the person posting below me?

[Transmogrifox]

I assume you're asking a layout question. 

The short answer to your question is for the most part it doesn't matter where the filter is located or where you tap your power from it -- especially if you use large/wide traces or wire on the power/ground bus.  A typical diode equivalent resistance is so many more times that of a good power bus wire that the noise rejection is about the same anywhere you tap power off the filter.

Now for what's really going on, in your schematic replace every wire segment with a resistor in series with an inductor.  From that you can see that the very best rejection to noise at all frequencies will be AFTER the power filter.  As mentioned in the "short answer" these resistances and inductances are so small it doesn't usually make a difference except in a few key areas.

It's a good practice to put the filtering components as close to the power jack as possible.  Also you want to put switched/switching components as close to the power jack as possible.  That way transient loop currents follow a path right out of the power jack and back in, instead of making a large path bisecting your audio grounds at noise-sensitive places (causing switching pops due to "ground bounce").  Also the ground/power paths can inadvertently cause high frequency feedback paths from high gain IC's (like op amps) and be a cause of RF oscillations that make your circuit operate mysteriously.

We can learn some things from high frequency digital circuit layout principles for application to audio circuits. In some cases the same things are at play.  The advantage with audio frequency circuits is that trace/wire inductance is a small part of the impedance so it's  much more forgiving when it looks mostly like a (very small) resistor.  In other words, at audio frequencies with ground/power currents on the order of several milli-amps a wire looks a lot more like a wire (zero ohms). 

The main thing to look out for are large and/or fast switching currents (like 100's of mA) and keep these isolated with a star connection back to the power jack.  A secondary thing is to not assume that RF noise and oscillations won't be audible.  Any kind of changing or pulsating RF noise can cause pulsating charges to ebb and flow, and it then becomes an audible problem.  (It's not a bad idea to add a good 10 nF to 100 nF cap into your power filter to reject RF noise and/or interactions).

[PRR]

> ground is ground and supply is supply

No.

Ground at my house is 1 to 5 Volts different from ground at the utility street pole. That's 0.4 Ohms of long-wire resistance against my 3 to 15 Amps typical unbalanced load. This (being AC) throws hum-bars in my telephone and cable TV. (I got it sorted now.)

Ground at your first audio stage is probably 0.01 Ohms "away" from ground at your power jack.

If your power input were pure DC, this might not matter. But then the power filter is not needed.

If you input a really crappy power source, the 100ufd cap injects much of that into "ground". If it goes to a place 0.01 Ohms away it may be injecting several MilliVolts of buzz right at the bottom of your first audio stage where signal levels are not-many milliVolts.

It does help to include several Ohms between the power jack and the first cap, to limit the peak crap current (also gives the cap better leverage against strong crap). 47 or 100 Ohms is often a good starter.

Power supply/filter layout can be CRITICAL on HIGH-power loudspeaker amplifiers (and RF stuff as mentioned). The resistance of a fat wire can not be ignored.

Pedals are MUCH easier. Any wire fat enough to survive gigging has resistance almost negligible compared to the pedal guts and typical power crap.

But still good practice to put the main water filter *right* where water comes into the house; and likewise with power filters.

The water-filter analogy can be extended. I have two water filters, coarse and fine. I tap out hose-water after the first filter, house water goes through both. The hose water does not have to be dead-clean, but the well water had major grit and was a problem even for the hose faucets, so I put it through a screen. The house water should be some cleaner so it goes through a fiber filter too. If you study larger audio systems, the Power stage gets slightly-filtered power, the low-level preamp stages get additional filtering.

[mattoverse]

Thanks so much guys, I really appreciate the responses and they are helping me get a better grasp of the issue.  However, I am still a bit confused.  Based on PRR's example:

I can understand how the power at point A is different from point B as it has crossed the low value resistor, but is point B really any different than C D or E?

For the ground points, is F any different than G or H?

Also, I thought the electrons actually flowed from ground to supply, so how does the placement of ground affect the placement of the filter.

My understanding is that in these small audio circuit shouldn't make much of a difference, so probably not worth worrying about?

Edit: just noticed the polarized cap is flipped the wrong way.., ignore that.