Best place to put a DC filter (noise reduction) cap

[DeusM]

Hey guys! I was thinking lately about the best place to put a filter cap to reduce noise. I think I noticed that the best place was placing the negative side of the cap next to the - "input" (of the power supply I mean) and the positive side as far as possible from the + "input". But I was testing it with a battery so I think I fooled my self. I was going to test this with a 9V transformer but I though about asking you first. My guess is that it should be in front of the + side of the power supply since all the current will flow from there first but All the schematics I saw placed it on different places (Also, shloudn it be in series instead of the - of the cap going to the negative side of the power suply since it will be in paralel with the circuit?) . I know it doesn't matter that much but I want to make sure I reduce as possible the noise. Also, can the reverse polarity protection diode reduce some noise too? Again, I was testing it but I think I fooled my self because I was using a battery. Hope to hear about your opinion and experience.
Thanks as always guys!

[antonis]

Hey Mat..

A "filter" cap is supposed to be engaged into some kind of filter..
It can't perform a "filtering" action by it's own only - even the big electros for PS ripple reduction can't form low-pass filter without a series resistor.. 
(you don't want to mess up with either transformer secondary impedance considered as series resistance or wire/track resistivity per unit of length, do you..??)

So, for effective filtering you need a series resistor..!!
(then you can calculate filter's cut-off point for known R&C values or from any combination of 2 known/desired values among R, C and f..)

Given that, best place (if such a term could exist.. ) for the junction of resistor-capacitor should be as close as possible to the start of supply splitting..
(the point from which the PS current starts to share into succeeding circuits..)

P.S.1

All the schematics I saw placed it on different places

They have placed it where it was more draw-friendly.. 

P.S.2

shloudn it be in series instead of the - of the cap going to the negative side of the power suply since it will be in paralel with the circuit?

Hey Mat, again..

A capacitor IN SERIES with DC supply is only of (discenting) usefulness  for the first 4-5 times RC constant after initial power-on..
(I let you guess what happens after that...) 

P.S.3
Due to posted emoticons and way of expession (which may be considerd as "irony" but I ensure you it isn't..) let's speak straight:
(no emoticons/icons/exclamation marks etc. - just allow me to underline some words..) 

A cap is series with pure DC is a blocking object..
(it appears infinite resistance in series with DC supply..)
A cap in pararall with pure DC is an useless object..
(it appears infinite resistance in parallel with DC supply..)
A cap in parallel with "dirty" DC is of great usefulness..
(it cleans up DC by shorting contaminating AC(*) to GND..)

(*)I use AC in the mean of not steady amplitude (ripple) and not necessarily in the mean of polarity changing..

[DeusM]

I changed the title. My English betrays me sometimes. With filter I meant "ripple" filter, or noise reduction. I guess you where talking about a hpf or lpf. So I guess the interpretation of my question changes. I was talking how to be effective in noise reduction. For instance this circuit:



The red one is supposed to be the ripple reduction filter (I guess). But, slouldnt it be in the + input so it will filter all the ripples before the current flows? (Also the ground in the - input)

PS: No offense taken by any emoticon  . Your dedication for answering right and clear surpass any irony that they may contain sir 

[antonis]

With filter I meant "ripple" filter, or noise reduction. I guess you where talking about a hpf or lpf. So I guess the interpretation of my question changes.

I guess it shouldn't change at all..
(ripple filter can be formed by LP filter 'cause it by-pass small voltage variations)
For a sole capacitor, another way to think it is like the capacitor "pushes up" low voltage variations to a level line (by discharging) and also "pulls down" high voltage variations to the same line (by charging)..

Trying to see what you've posted:
[img]https://s14.postimg.org/lp7uao2st/Sin_t_tulo.png]https://postimg.org/image/lp7uao2st/][img]https://s14.postimg.org/lp7uao2st/Sin_t_tulo.png

The red one is supposed to be the ripple reduction filter (I guess). But, slouldnt it be in the + input so it will filter all the ripples before the current flows? (Also the ground in the - input)

But it IS on the + input..!!
(R24, C4 +, D5 cathode, R6-12-17-20-23 upper leg and +9V shoud be consider bond at the very same point - same for all points connected to ground-symbol arrow..)

PS:
"Sir" is only one person in this forum..!!!

[amz-fx]

The red one is supposed to be the ripple reduction filter (I guess).

Since you said you are using a transformer, the ripple noise will be multiples of 50Hz or 60Hz.

At low frequencies like 100Hz/120Hz, that filter capacitor is going to do almost nothing to reject noise from the power supply. The reason is that the output impedance of the power supply is very low. Once you calculate the RC of the power supply's output with the filter capacitor, then you will find it is not doing much filtering at 100/120Hz. Even if you make the capacitor quite large, the noise won't go away (not enough R).

You need a series resistor as Antonius said previously. I would suggest starting values of 47 ohms in series with the +9v input and a 470uF filter capacitor. If size is an issue with the cap, you could go with 100 ohms and 220uF for much the same results.

If you were using a switching power supply where the noise pulses are at a much higher frequency, you can get by with significantly smaller R and C values. Even then, some small added resistance (like 2.2 to 10 ohms) is a good idea.

Filter Calculator: http://www.muzique.com/schem/filter.htm

Best regards, Jack

[DeusM]

But it IS on the + input..!!

Yeah! I shloud have said: shouldnt it be in front of the other components?. I guess my lack of electrical knowledge is the problem here but shlouldnt it reduce al leat a litle more noise that way?

You need a series resistor as Antonius said previously. I would suggest starting values of 47 ohms in series with the +9v input and a 470uF filter capacitor. If size is an issue with the cap, you could go with 100 ohms and 220uF for much the same results.

I'll guess Ill have to try that
By the way. Are you AMZ from the PAGE AMZ? I guess you are. Learned a lot from the page so, thanks a lot!!! The page is in my "favorites" tab on chrome.

[PRR]

> shouldnt it be in front of the other components?

For the size of typical pedal guts, exact location is probably (probably!!) not real important. A couple inches of wire/trace is very nearly "zero" Ohms.

Your porch is bouncy. You want to put a rock under it for more support. The very-best point is under the jumping load, or under the exact center. However if the porch beams are fairly thick, just a little long, almost anywhere near the middle will help a lot. (It gets different in very long bridges, or very large studio installations-- that's a different world.)

YES just the one cap won't isolate well. If you have your pedal building bench framed from that bouncy porch, a rock won't do much. You want to break the beams between porch-bounce and pedal-bench, so they are not a "good connection". In small electronics we insert a resistor, picked for small/tolerable DC drop but larger than the cap's impedance at crap-frequency. 47r and 100uFd is often a good starting point. (Gives 0.5V DC drop at 10mA, gives 3:1 reduction of 100Hz crap, 6:1 reduction of 200Hz crap.)

[antonis]

The page is in my "favorites" tab on chrome.

(with all respect to Jack..!!)

"Favorites" tab pages aren't there just for quick opening..
(author's ulterior intent is to be studied and reader's ulterior intent should be to get understood..)

Inspite of text simplicity level, there always some minimum prerequisite amount of knowledge for a satisfactory ken..
e.g. it's almost meaningless to know how to use a formula for a first order passive RC filter cut-off point calculation whereas you don't know what it actully does mean/represent/reflect..

So, you better enrich your lack of "electricity" knowledge before you mess with "electronics" issues..
(I totally agree with your de rigeur response about that nasty AC behavior with all those filthy imaginary/complex/phase bankqueters but, unfortunately till nowdays, sound is not a whit involved with DC..) 

on topic now:
I hope you've understood the main role of your posted Big Muff supply filter cap which is for voltage regulation..
It reduces voltage ripple by a factor of 0.144/RCf where C is its capacitance, f is mains frequency & R is the load..
(for no ripple at all, we should need a sole capacitor of infinite capacitance or zero load current..)

To take more advantage of that cap existenece, we usually place a series resistor to form a low-pass filter and get rid of the low frequency(twice the mains frequency) unwanted remaining small ripples..
(those ripples which we couldn't "vanish" due to inability to make a perfect zero ripple factor..)

That low-pass filter corner frequency should be set lower enough than pipple frequency, for an effective elimination..
(e.g. Jack's RC values suggestion is for a filter of -3db cut-off point at 7.2 Hz whereas full wave rectified frequency is 100/120 Hz..)

[DeusM]

"Favorites" tab pages aren't there just for quick opening..
(author's ulterior intent is to be studied and reader's ulterior intent should be to get understood..)

Yes. I know. I was just thanking the guy for having the site which from I learned a lot by reading the articles like you said.
English is not my native language so sometimes I say the wrong words and if I don't enrich your lack of electricity is because I don't have the time honesly (I study at the university and also practice guitar above 2 hours everyday). Also, I think is pretty hard to learn electricity and electronics by myself. Specially when all the good sources are in English and not being it my native language. I try my best and always read and investigate but there are lots of things and It takes time to understand the more tecnical thing for comeone like me. That's why I post so many times here. Because I know there are people that know a lot and not because I want a quick answer. Your explanations where clear I know I know a litle more. I'm still having a hard time trying to finish my first pedal which is taking me a lot more time than I though. I plan to keep studying electronics one I finish it but I just want to get it done.

[amz-fx]

I'll guess Ill have to try that
By the way. Are you AMZ from the PAGE AMZ? I guess you are. Learned a lot from the page so, thanks a lot!!! The page is in my "favorites" tab on chrome.

Many thanks for your kind remarks. 

One last note...  I suggest that the LED be powered directly from the raw +9v and not from the filtered supply after you add the series resistor (Rx). The LED may be using as much or more current as the rest of the circuit, so having it on the 9v source will reduce the voltage drop across the new resistor and reduce potential pops when switched. A little ripple on the LED will not be noticeable.



Best regards, Jack

[Frank_NH]

Question - in this article

http://www.muzique.com/news/reverse-voltage-protection/

Jack shows how to add an Shottky diode (1N5817) in series with the input resistor for additional protection.  If you do this, is the series resistor still needed?  I've seen pedal designs with just the series diode and a large filter cap, but perhaps that's not a good (or complete) design for reverse polarity protection.

I've also seen this TI document which shows how to use a MOSFET for this purpose.  Hmmmm...

http://www.ti.com/lit/an/slva139/slva139.pdf

[antonis]

Question - in this article
http://www.muzique.com/news/reverse-voltage-protection/
Jack shows how to add an Shottky diode (1N5817) in series with the input resistor for additional protection.  If you do this, is the series resistor still needed?

(on behalf of Jack - without his written permission..)

In case of series protection diode, input resistor's role isn't anymore for "fusing" purpose but it's still there to act as part of supply LPF..

[anotherjim]

...is the series resistor still needed?

Yes. There is a catch though. Your circuit's supply capacitor is going to have to be sized large enough to cover the entire AC bypass/ripple smoothing since the series diode stops any AC current finding the path back to the supply in one direction. This sounds particularly bad if it's a power amplifier which will have significant AC current. Usually, Stompboxes benefit from the supplies own capacitance, be it battery or AC adapter.

In a daisy chain DC powering, a box with not ripple cap can benefit from another box that has one, but no series diode - another boxes cap is effectively on all the other boxes supplies too, with some wire in between. Fit a series diode and that mutual support is compromised.

Another situation related to that is that some boxes, if they have proper supply filtering, can get away with the user having a 9VDC wall-wart that has no ripple capacitor, only the transformer and rectifier in its box. So they may assume that the power supply is perfect without realizing that A: Their pedals are actually getting nearer 12V (supply caps convert the supplies 9v "average" to the peak value) and B: A pedal on its own that has no on-board filter will hum and buzz like hell - and of course claim that this pedal must be faulty because the others are fine.

[Frank_NH]

Thanks Antonis and anotherjim.  Wow - things get complicated pretty fast.    I suppose there's no-one-size-fits-all approach to power supply ripple/noise filtering and reverse polarity protection, but if you can live with the voltage drop I like Jack Orman's approach.

For reverse polarity protection, I've also seen the use of one of these fuses instead of the Shottky diode:

http://www.smallbear-electronics.mybigcommerce.com/resettable-fuses-2/
http://www.littelfuse.com/products/resettable-ptcs.aspx

Perhaps this is the best approach, given that reverse polarity is a (hopefully) rare event, and with a fuse you don't have to pay the penalty of the voltage drop that you have with a diode.

 

[DeusM]

One last note...  I suggest that the LED be powered directly from the raw +9v and not from the filtered supply after you add the series resistor (Rx). The LED may be using as much or more current as the rest of the circuit, so having it on the 9v source will reduce the voltage drop across the new resistor and reduce potential pops when switched. A little ripple on the LED will not be noticeable.

Interesting. Although, I made a modification on the circuit and it seems that the pop is not so notisable.



I'm going to check later how diferent it is the way you are telling me.
I can't remember now where did I get the modification from. Also I was looking at the modification for poping in your page. I'm going to have to see tomorrow the best way to avoid noise and pop in the circuit. Is not bad but I think it can be improved. Today I'm going to buy the components I'm missing.
Also, guys. I made a mistake. Tha cap is not for filtering. Is for pop noises. I was reading about RC filters. My question is, shouln'dt it be at the ouput?

[antonis]

If you're talking about C15 above, it's placed exactly where it should be.. 
(it eliminates LED pop right into the heart of pop-source..)

[antonis]

Perhaps this is the best approach, given that reverse polarity is a (hopefully) rare event, and with a fuse you don't have to pay the penalty of the voltage drop that you have with a diode.

Perhaps only in case of fuse faster "open" than semiconductor "blow out"...
(which isn't true at most cases..)

[jsleep]

Interesting. Although, I made a modification on the circuit and it seems that the pop is not so notisable.

There will be no pop in the little schematic you included, because there is no switch to make it pop.  Is the switching on 9v+ going into the schematic?

JD

[DeusM]

Interesting. Although, I made a modification on the circuit and it seems that the pop is not so notisable.

There will be no pop in the little schematic you included, because there is no switch to make it pop.  Is the switching on 9v+ going into the schematic?

JD

I didn't included the switch in the schematic.

I'm sorry but I cannot understand what you mean with "Is the switching on 9v+ going into the schematic?". My english is not perfect sometimes.

I use the wiring from Mad Bean. I think it works good. Dont know if that answers your question

[Frank_NH]

Perhaps this is the best approach, given that reverse polarity is a (hopefully) rare event, and with a fuse you don't have to pay the penalty of the voltage drop that you have with a diode.

Perhaps only in case of fuse faster "open" than semiconductor "blow out"...
(which isn't true at most cases..)

Hmmm.  So a those particular fuses would blow slower than a diode?  You can purchase them at different current ratings, and so limit the damage you would do to your circuit.  I'll have to investigate...