Generalised power filtering question

[lars-musik]

This might sound trivial, but I try to figure out what the best practice for power filtering in a guitar pedal is. My search didn't supply me with "the" answer and when it comes to basics I am still very much a novice.

Here are some examples what I found (besides some fifty other solutions) on my hard-drive:

The circuit analysis of a Crybaby on Electrosmash:


The circuit analysis of RAT on Electrosmash:


Madbean's Krankosaurus:


I see that one (sometimes two) capacitor does the magic (ranging from 33-220µF), a diode is often used as polarity protection and sometimes there's a low value resistor for whatever reason.

Say I want to built some transistor based booster (no need for polarity protection, no need for a 1/2 bias supply): What would you do? 47uF to ground and basta? A "for whatever reason resitor" series? If yes, then why?

Feel free to simply point to the threat/website that I overlooked where everything is explained for the dummy and served on a silver platter.

[Groovenut]

The cap at the input of DC is there as an additional filter but is relying on the resistance(impedance) of the power supply to form the RC filter knee with the cap (may or may not give good filtering results depending on the psu). The small resistor at the input of dc prior to the cap ala madbeans pic is there to form the RC filter knee with the following cap, this guarantees filtering at the set frequency (in this case 43.1Hz). The zener in the crybaby is a 9.1V for overvoltage protection as well as polarity.

The madbean pic is your best bet for the circuit you mentioned you were building IMO

[R.G.]

This might sound trivial, but I try to figure out what the best practice for power filtering in a guitar pedal is. My search didn't supply me with "the" answer and when it comes to basics I am still very much a novice.

One reason your search didn't find the answer is that there is no "the" answer that is correct for all situations. As a gross illustration, some opamps can ignore power supply noise by up to about 100 to 1000 times (that is, any noise on the power supplies appears on the output 100 to 1000 times smaller). On the other hand, single-ended circuits like single transistor boosters may allow noise through to the output almost intact, having nearly NO power supply noise rejection. On the other hand, opamp circuits can oscillate if the power supplies have any inductive reactance up in the MHz region, so they really, really need ceramic caps right at the chip.

What matters for power supply decoupling is the circuit's needs.

Say I want to built some transistor based booster (no need for polarity protection, no need for a 1/2 bias supply): What would you do? 47uF to ground and basta? A "for whatever reason resitor" series? If yes, then why?

Now we can talk. You need a small-value resistor, 10 to 100 ohms, a 22 to 100uF cap to ground after that, and probably a 0.1uF.

[antonis]

and probably a 0.1uF.

I'll dare to overcome "probably" with "definately"...
(and add "ceramic"..)

[lars-musik]

So a 82R in series followed by a 47uF plus a ceramic 100nF (yet to order - a film type wouldn't do??). Thanks for your input! Guess I'll have to be back if another circuit with opamps comes up.

[hymenoptera]

So a 82R in series followed by a 47uF plus a ceramic 100nF (yet to order - a film type wouldn't do??). Thanks for your input! Guess I'll have to be back if another circuit with opamps comes up.

Film would work, but ceramic is cheaper and offers better ESR and high frequency performance.

If you think about it, adding 0.1uF in parallel doesn't add much to the total capacitance. One point of multiple caps in parallel is to take advantage of the pros of various dielectric materials, so for that reason the material is important. For instance, if instead you used a 0.1uF aluminum electrolytic it would be pointless.

Ceramics are cheap and you'll use more 0.1uF ceramics in power supplies than just about any other value.

[PRR]

You can NOT judge (or design) a power input filter without knowing:

1) the load
2) the required clean-up

However--

A "C only" filter can't do anything. A strong crappy power supply WILL force the cap to follow the crap.

An "R-C" has much more leverage against strong crap, because the resistor limits how hard the crap source can force the cap.

What size resistor?? When available voltage is more than you need, let the resistor drop 10%, 20%, even 30% of the available supply. Tube hi-fi amp with 350V on the final stage, the phono input stage may only need 100V to be clean, we'd drop 30% in a heartbeat. This large resistor also lessens the size of cap needed for a specific amount of filtering (and caps used to cost real money).

When you have just 9V and that may not be generous, 5%-10% may be all the resistor drop you can stand.

Of course this implies that you know your Current, so you can use that with your Voltage Drop and figure the resistor. Most pedal workers don't have a clear idea of Current.

With a simple 1-transistor stage, find the collector resistor(s), or emitter resistor (whichever is larger), and pencil 10% of that. With two transistors, often one collector resistor is much smaller, ignore the larger one. For FuzzFace often built with 5K or 8K (plus 1K), pencil a 470 Ohm.

Experimentally, you can try a 100 Ohm resistor and see what Drop you get.

For many-many small-audio schemes, 100r is not too-much drop, and a 100uFd cap with that will give good crap rejection, a 470uFd will give very good filtering, and these are smallish cheapish caps now. Try it on bredbord or tack-solder jury-rig before you commit. (Try it with all your crappiest supplies: nearly any darn thing will play clean with a 1Spot or other quality battery replacement.)

Note that both Smash and Madbean's cited plans use *roughly* these values. (2:1 variations in audio are often more about what's laying on the benchtop than any deep thought.)

Keep the 100r on large pads with clear space. For many "oops!" situations, this resistor will burn before other parts, so easy replacement may be good. With reverse diode, this resistor (if 1/4W) WILL burn on reverse DC or on AC, giving first-order protection against most wrong-supply mistakes.

Re: 0.1u caps all over.... my opinion is that these are often like the guy who sprinkles salt on everything. There was more justification for this (but less use!) back in the 1940s when e-caps were awful. Miniaturization and improved fabrication has made e-caps wonderfully better. In typical pedal sizes they are near-perfect to far above the audio range. If you have switching (supersonic) switching supply, 0.1u film/ceramics ARE often useful. (However small e-caps are sometimes preferred because a good ceramic has so little self-loss that it can incite VHF instability against a few inches of lead.)

'741 and TL07x chips need 10uFd across the supply pins within 3 inches of the chip. (CMOS should have this much; TTL/LSTTL should have more.) Usually a >47u cap (power input filter) on the same board will do. Hotter chips may need cap closer and may need ceramic bypass. Read the sheet, and any app-notes it mentions.

[amz-fx]

A "C only" filter can't do anything. A strong crappy power supply WILL force the cap to follow the crap.

What Paul is saying here is that the output impedance of a power supply can be very low --  the output Z of a 7809 regulator is 0.5 milliohms, which is very low indeed. So any capacitance (C) you add is working with the output impedance to form the low pass filter and since the impedance is so low, the corner frequency of the low pass is WAY above the 60Hz or 120Hz that we are trying to filter. It takes some added resistance to make a low pass with the corner significantly below the hum that we are trying to filter. This is why the resistor has been added, but you get some voltage drop when adding it, and the amount of drop depends on the current being drawn through it.

The Rat filtering that you linked has a big problem if a reverse polarity power supply is connected. The current flows through the D3 diode and the 47 ohm R10, and then the resistor acts like a fuse since about 170 ma is flowing and the power is close to 1.4w. So the little 0.25w resistor turns red and melts down. I've replaced a lot of them...

Best regards, Jack

[lars-musik]

Now that I learned there's no such thing as the general optimal power filtering, might I inquire what you professionals would advice in the following case. This DOD230 noise gate seems to be a bit thin in regards to power filtering - just the R6  100R resistor? Can this be so?

[Kipper4]

Looks that way Lars.
You can add your own now you know how.
Just my 2 pence but I like the dod fx30 gate.
If you build the dod230 I'd breadboard it so you can optimize the vactrol if your rolling your own.

[lars-musik]

Just my 2 pence but I like the dod fx30 gate.

Hi Rich,

does it mean you like the DOD FX30 BETTER than the DOD230 or does it mean that the DOD230 will be a good choice?

(I tried to knock the Decimator into a 1590a shape but failed (several times). So I thought, maybe something a bit less complex would do also, so I searched for a low part count gate and found this one. From the youtube videos, the release time might be a problem, but the 1590a layout is not.)
I am not much of a breadboarder, so this is what it is going to be (power filtering still to be added):

[Kipper4]

I mean I liked the fx30 better as a gate.
Miniaturising ain't my bag so sorry can't comment.
I hope it goes well.

[amz-fx]

Since much of the old DOD stuff was expected to be powered by a battery, most of them are lacking in power filtering.

You can add a 220uF capacitor after the 100R resistor at the point labeled +9v (with negative side of cap to ground) if you are using an external power supply and it will take care of it. Even 100uF will do, if that is all that you have.

regards, Jack

[lars-musik]

Thanks very much! Yes, the lacking power filtering makes sense when only batteries are used.

[GibsonGM]

Sorry to bring in an 'extra dimension' here...Am I the only one who sees the Madbean schematic up there, and notes that the cap is BEFORE the 82R resistor...and, as R.G. mentioned, it therefore isn't part of the filter?   So, the ~41Hz filtering isn't there? 

Looks like you have PS resistance and that 47u cap working as a LPF, rather than it being involved with the 82R...

2)  In that case, do we then have a filter composed of the 82R and the parasitic capacitance of the zener??  So, you'd need to know that capacitance and see what the 82R is doing with it.  Additionally, that would mean if there is any additional capacitance just down-stream, we'd have to add that in, as well.

No?

I always just do as Jack suggests above, and know why/how it works...cap before seems more...odd...

[lars-musik]

Extra dimensions are very welcome in this - not so trivial after all - thread.
The "9V" on the right side is the point of entrance and the "9V" on the left side ist what is called a "net port" in Diptrace - I have no idea how the right term for it is, but this it the symbol used in the schematic for all 9V points after the filtering. So the cap is physically placed after the  resistor.

[PRR]

> just the R6  100R resistor?

You would really expect a cap, right?

Aside from power-crap, any wise designer will throw a few uFd across the chips to clamp their power pins against MHz instability.

As this is not a factory drawing, I don't know how far it can be trusted. (Never trust factory either.)

But there's no need for heavy filtering. The opamps have excellent supply rejection. The main Vref has C2. I can picture a little ripple in the sidechain, but the snap-action of R10 makes this a random timing issue not a audio-crap issue.