[Folsom]

I am using some Bryant/Hubble ac receptacles, four (only two in box, wiring it now) Coilcraft CMCs, and 20 GE 5.0uf capacitors (400v). The capacitors are on the ac receptacle side. I figure this is most appropriate. Overall capacitor uf amount is more particular than placement, but closets to ac receptacle makes for best filter because it as far down the line as possible towards the equipment if not in it. The little boxes the wires go into for the CMC's are pure copper coated in silver, terminals.

Why I choose the capacitors and amounts... First no one could answer why choose the random uf amounts and denied any reasoning behind them when I started plugging numbers with effectiveness, based on CMC graphs. A fellow on the forum helped with the suggestion for capacitor usage, and I am sure he can comment on this topic if he wants. The one downside is they draw 5amps at all times. However  I can switch off 3 of ac receptacles, via the breaker (gone for long periods etc), but one always stays on for SqueezeBox 3or whatever other device needs to not turn off to prevent having to setup over again. (I will add a fuse to my linear power supply for SqueezeBox 3)

The more capacitance you use the more unwanted signal will be loaded down.

The impedance of the capacitor falls as frequency increases so it will be much more aggressive at loading down higher frequency noise.

Capacitors will have an equivalent series inductance (ESL) which will limit how high in frequency they will be effective at shunting noise on the power line. For this reason you will want to use multiple capacitors which will divide ESL by the number of capacitors used.

(If anyone cares here are links to my SB 3 linear power supply which I am going to modify with some better capacitors and maybe a toroidal transformer, some day, I got it for free, and it is adjustable up to 15v, sorry no inside shots! I am using a shielded cable, the shield connects to earth ground, via bypass power supply, but does not connect to SB, it is just a drain wire, also free)
http://img.photobucket.com/albums/v638/Stainlessberetta/jeremyps_003.jpg
http://img.photobucket.com/albums/v638/Stainlessberetta/jeremyps_001.jpg

[Occam]

If you actually measured the no load current at 5 amps, unplug it immediately!!!!! 
I do hope those bolts and metal bar securing the CMCs are not forming a 'shorted turn' (which might be the cause of that dangerous current draw).

You don't need anyone to tell you how much capacitance is appropriate. Take the responsibility upon yourself, read and UNDERSTAND the previously referenced coilcraft pdf and actually calculate the the Q of your filters with the known differential (leakage) inductance, the ridiculous amount of capacitance, and your anticipated minimum load. The Q should be <1. Yours isn't. You are amplifying noise at the pole frequency, not attenuating it. This is not rocket science, this is high school algebra; use it.

FWIW

EDIT - and please don't refer to THAT as a Felix, even in quotes. You're aptly named.

[Folsom]

They draw 5 amps but do not use 5 amps. Fact is that out of 20amps I will only get 15amps because of the draw, not the usage. This is something I continually tried to discuss but no one had any worth of mentioning. There is a small amount of dissipation that will always exists in use of capacitors in a shunting position, that I talked about, yet no one mentioned, even if it is almost not existent. The box has ventilation on the top as well, and the capacitors have room on purpose, just in case.

The only person who made any sense had this recommendation. I tried to use the graphs etc to no avail. Honestly I feel the person who helped me knows what he is talking about, as all his other information has been very smart, and makes sense. A lot of the information has been literally voodoo magic, so I keep asking to have it explained, but no one has yet. I will see how this works.

The CMC's just have a piece of plastic between them, besides that the bar across the top connects to a plastic, tapped, base.

Anyone is welcome to chime in about, I am very welcome to any backed up statement, any reasoning.

[Occam]

'Draw'????
I'm well aware of the difference between va and watts. I'd tried to explain to you that reactive impedances only dissipate energy to the extent of their esr. The fact that you took away from those exchanges  - 'tried to discuss but no one had any worth of mentioning', is not my problem, just confirmation that good deeds and efforts rarely go unpunished.
Now that I understand I've misunderstood your nomenclature and you're not as likely to burn down your dorm, but simply implemented a piss poor filter, I am much relieved.

[JoshK]

Let me try to cut through any festering animosity and try to paint a big picture view of what the problem entails.

Paul is trying to suggest that every component will need a different cap for near optimal filter because of the differences in current draw (the component).  He simply isn't going to do the math for every component of every person who builds one of these. 

He has nudged people to take a look at some of the datasheets and look up some of the basic guiding formulas to figure it out for themselves.  There are numerous reasons why he, or anyone else for that matter, isn't exactly going to spell it out for you.  This is the Y in DIY. 

One of former professors used to say, the difference between memorizing material and learning is owning the material.  He would say to use, "go home and make it yours.  Trained monkeys can watch me solve a problem" on the chalk board and repeat that same problem, but when the problem changes they don't know what to do.  Its only if you learn the material that you can adapt it to another problem.  If one simply takes a bit of effort to look it up, all the information has been provided.

The 20,000ft picture (with lots of hand waving) explanation is that the shunt capacitance sets the corner frequency of a filter.  ESR, leakage inductance and all that factor in, but essentially you are probably shunting a lot of the power to ground because of the corner frequency and the Q make the filter shunt all the way down to line frequency.   My naive understanding and just looking at your filter would tell me you are probably having a serious unwanted affect. 

[Folsom]

Paul you explain in a way that suits you already having the knowledge. You must forget some people know much less, so what you say is so vague it is hard to even know where to start to attempt understanding anything.

What I have gathered is that different capacitors have a different range of filtering, but it usually starts at a certain point. However it then encompasses everything after that point.

Paul if you want to be helpful your manner is well most the time you are rude instead of wanting to actually help. It is always nice to express your knowledge but I mostly get a tone that you want to yell at people you deem stupid or something. You know a lot and can be very helpful, but with your own ideals, of looking it up, most the time I wish you would not answer things I ask about because of the way you do it.

I calculated the 5amps draw by the way, it is actually more like 4.5 but who is counting.

Information I got, and so far everything I can find on it, it adds up and makes sense, is accountable for what I can find and understand. I looked into it, and it is the only bits of information I found to relate to anything I found while investigating. Like I said I tried to come up with ideas of what capacitors to use etc, but to no avail, and no answer other than "this one sounds better than this one, and more variety equals more filtering." Yes people did give formulas for the range of frequencies from those capacitors, but that was just for fun almost at this point since no matter what one you choose it had no relation to the CMC graphs "it does not work that way" as people said. Besides that point most of the values of capacitors being used in the felix seem to not match up with the graphs anyways, at all. IE if you are trying to attenuate as much as possible with the capacitors, that the CMC's do not do, why would you have one, or rather multiple that pick up in the STRONGER attenuation range of the CMC, and not well before it's lowest point of attenuation, or if not possible due to massive uf, right after. That is what I tried to calculate out and do but everyone disagreed, however continued with subjective "this sounds a little better, no joke." Perhaps the felix needs some rethinking. (also mine is working fine so far at the right voltage)

Anyways a bit more (on top of the other about ESL etc) information:

"The specific value is not critical for the capacitors in th Felix.

What they do is shunt any signal (noise) which might exist on the power line.

The more capacitance you use the more unwanted signal will be loaded down.

Of course the capacitors also draw current from the AC line so you want to keep an eye on it.

The capacitors are reactive so it wont cost you much since relatively little power is actually disipated."

then

"This is because they are reactive meaning voltage and current are out of phase.

Power is taken from the source over 90 degrees of the wave form (capacitor charging) and then for the next 90 degrees that power is returned, then it repeats.

In reality some power will be consumed mostly due to the resistance of the powerline but the capacitors do have a small amount of dissipation."

and

"The two recomended 0.47uf capacitors wont shunt noise very effectively due to their high impedance."

[mcgsxr]

I believe the tone in Paul's discourse tends to lean towards trying to ensure folk's know the danger in messing with electrical power.  I know squat, and as a result stay clear of this. 

There are implications (the nasty legal kind) of putting forward designs for others to build, so I think that too is at play here.  That might be the nature of the vagueness.

He can be a little harsh (subjective take on my part) but I would rather be barked at on a forum, than fry my gear, or worse yet self.

I see your side too 'Smiles, but just glaze over the part that rubs you the wrong way, and you will learn a ton from a guy like Paul. 

As a side bar, the older are always crustier, and often due to having learned some of the hard parts the hard way.

I think Josh hit it well, to suggest that you continue to work on it, and learn from those around you.

[JoshK]

DOS, calculate the attenuation at 60hz based on your filter. 

[Folsom]

Oh I intend to continue to learn. I have to be allowed and opened up to it if I am to start into something more complicated than simple DC circuit stuff.

I am not giving a schematic or saying anyone should do this. Anyone can look at anything and end up hurting themselves by trying to recreate it. However say the PCB boards for Felix are more in chance to get in trouble without a good disclaimer. I am just posting pictures with information. With that reasoning I do no think it is really the issue here.

I would not do anything if I felt it was dangerous. Paul makes a lot of assumptions, when the reality is a lot of things I talk about are with knowledge, theoretical, and actual in hand involvement of people that do know what they are doing.

AC voltage and amperage from the wall is quite dangerous, I know. I have had a small explosion due to a poorly wired old piece of equipment (from the factory) that had no intention of avoiding the possibility of a loop back to ground from another device.

Now back to my version of a felix, unless someone cares to explain why the reasoning behind mine is wrong, or how amplification is actually happening, I have no real reason to believe it, especially when what I am being suggested to now is what I originally went for, got shot down, and found someone who put it all into black and white.

[Occam]

........
Now back to my version of a felix, unless someone cares to explain why the reasoning behind mine is wrong, or how amplification is actually happening, I have no real reason to believe it,

I take it you still haven't computed the 'Q' of that second order filter? Nor asked why I referenced the differential (leakage) inductance of the CMC in its computation. At this point, don't bother.

...especially when what I am being suggested to now is what I originally went for, got shot down, and found someone who put it all into black and white.

Good for you. I've no idea who that would be, though I certainly know who it isn't.

[Folsom]

DOS, calculate the attenuation at 60hz based on your filter. 

I am not sure how to do that. I have been using graphs etc... However there is something I am going to point out that might be, being over looked, and that is that impedance drops with multiple capacitors, as the ESL is divided in half each time you double capacitors.

[Daryl]

"I do hope those bolts and metal bar securing the CMCs are not forming a 'shorted turn' (which might be the cause of that dangerous current draw)."

The loop formed by the hardware is perpandicular to the direction of the coils winding so it can not couple to the coil in a meaningfull way.

Also said loop is outside the magnetic circuit which means only a small amount of coupling could be achieved anyway.

And the most obvious reason why that statement is false is that the CMC's are in SERIES with the load which means even if you had a perfectly coupled shorted turn all it would do is render the CMC's ineffective, no load current could possibly be levied upon the filter by parisitic coupling to hardware.

[Daryl]

"They draw 5 amps but do not use 5 amps. Fact is that out of 20amps I will only get 15amps because of the draw, not the usage."

You will have nearly the entire current rating of the CMC's available for your load.

The reason is that the current draw of the capacitors has a phase of 90 degrees.

The current of a resistive load which has a phase of zero degrees will add to the current of the capacitors as follows...

I(total)=SQRT(Ir^2+Ic^2).

Where:
SQRT()=Square Root.
Ir=Resistive load current.
Ic=Capacitor current.

As the resistive load current gets higher the additional current due to the capacitors gets smaller and smaller.

For inductive loads the total current is actually less than the load current...

I(total)=Ic-Il

Where:
Ic=Capacitor current.
Il=Inductive load current.

Capacitive loads which are unlikely but if they occurred they would simply add to the load of the capacitors.

[Folsom]

Interesting. I think I understand that too. I did some playing with calculating my loads like my Squeezebox using say 1.5a would use a total current of 1.875 so that is .375 worth the capacitors have loaded down, yet if it was using 3a it would take a total of 3.2a and that would be .2a worth of capacitor loaded down. That would that the total current used by a Squeezebox drawing 1.5a would have a ratio of 1.25a to 1used due to capacitors, where as the 3a would have a ratio of 1.07a to 1. That meaning basically the more the resistive load the less the capacitor loads down, and I do get near 20a. It would be a logarithmic curve, in my favor. (I realize my SB load is different via my power supply, I just wanted to make sure I understood what was going on)

Very cool.

Thanks Daryl.

[Daryl]

"You don't need anyone to tell you how much capacitance is appropriate. Take the responsibility upon yourself, read and UNDERSTAND the previously referenced coilcraft pdf and actually calculate the the Q of your filters with the known differential (leakage) inductance, the ridiculous amount of capacitance, and your anticipated minimum load. The Q should be <1. Yours isn't. You are amplifying noise at the pole frequency, not attenuating it. This is not rocket science, this is high school algebra; use it."

That is excellent advice.

Not only can you share it with Smile destroyer but I hope you will apply it to yourself.

As you said already this isn't a real difficult circuit and would be a good place for you to start learning electronics (maybe after that you can start showing others how to build filters and critisizing their work).

You don't use the minimum load impedance differential leakage inductance and filter capacitance with a rigid voltage source to model the circuit (it doesn't work that way).

The most problematic situation is when the load impedance is essentially infinite which very well could be the case.

The differrential leakage inductance you speak of is a parasitic element, a perfect CMC would not have any.

As a parasitic element it will be pretty lossy and have a low Q, especially when you use greater filter capacitance to bring the resonant frequency down further where the elements Q will be even lower.

This loss will be eclipsed however by the most important element in this circuit which is the impedance of potential interference sources.

Interference sources on the electric line will have considerable impedance (they are not rigid voltage sources as a novice might use to model the circuit) both due to the characteristics of the distribution system that they must traverse to arrive at the filter and due to the various ways the signals are induced upon the line.

The large capacitance of Smile Destroyers circuit causes it to resonate at frequency where the reactive elements  have a lower impedance which in turn results in low filter Q when combined with the considerable impedance of the interference sources.

The important circuit elements then are the filter capacitance, interference source impedance, differential leakage inductance and ESR of said inductance.

The load is unknown but attenuation is increased as load is increased so it is a good thing if some load is present but don't count on it.

Smile Destroyers circuit starts attenuating interference at a frequency 100 times lower than the Felix circuit when it's 100 times higher capacitance starts tugging upon the impedance of interference sources combined with the impedance of the CMC's in series forming a voltage divider and attenuating the interference at rate increasing by 6db every octave.

When the resonance frequency of the filter is reached the attenuation rate will then increase from 6db/oct. to 12db/oct.

The Felix circuit with it's too small capacitance will be ineffective over much of the important spectrum until very high in frequency.

At frequencies where the Felix is effective Smile Destroyers circuit will have an additional 40db attenuation compared to Felix.

I believe their was mention of a piss-poor filter earlier in the thread?

Whatever you do, DON'T CALL IT FELIX! (you'll give people the wrong impression).

[Folsom]

Wow Daryl, I changed the name just for that effort Daryl.

By the way my stuff sounds amazing now. I will take pictures of the outside before too long. I have a cool vent on top with some shiny filter under it.

My lower extension is much more defined. The Who - Tommy, Pinball Wizard, I can hear the bass much more defined; by the way John Entwislte is a badass who pioneered a lot of bass playing, unfortunately on Tommy he is hard to hear because it is so low and most stuff does not reproduce it well. I listened to the album some at my local audiophile store and did not hear much for the bass, on around $20k worth of stuff... sorry for the rant, I am just excited! It is funny how much lower extension is often produced extremely poorly.

Imaging was increase a lot as well. The specialness and impact of things, voices, coming in and out, is much greater, and can be felt, physically, not be ear, substantially more.

I guess what I am getting at is that perhaps maybe mostly with digital (SqueezeBox3) power filter really makes such a difference it is mind blowing and worth your time. I will report back on Vinyl later.

[kyrill]

DOS what you hear is that compared to a standard Felix or to non filtering at all?

[Folsom]

No filtering

No wait I was using an APC brand "filter". It was just a tiny thing, I will look inside and see what filtering it had... I think it was just a few capacitors from the first time I looked in it. It did help but not like this!!!!

Inside the APC filter it has several capacitors, one X2 rated and then a few not. It has two induction devices as well. I would rate it as probably about 1/3 affective as a "Felix" based on the parts inside, and being made in China. That is however for maybe one device, it was used for multiple.

[JoshK]

Daryl,

If you goal is to test where my moderation boundaries are, you are doing well.  Debates, discussions, even heated are acceptible, but personal attacks aren't. You post is clearly trying to insult and insight.  You've been warned.

[JoshK]

You don't use the minimum load impedance differential leakage inductance and filter capacitance with a rigid voltage source to model the circuit (it doesn't work that way).

Well then, explain to us how it does work.

The differrential leakage inductance you speak of is a parasitic element, a perfect CMC would not have any.

Right, and notice we have purposefully not recommended a "perfect CMC" but rather one that has substantial leakage inductance.

Smile Destroyers circuit starts attenuating interference at a frequency 100 times lower than the Felix circuit when it's 100 times higher capacitance starts tugging upon the impedance of interference sources combined with the impedance of the CMC's in series forming a voltage divider and attenuating the interference at rate increasing by 6db every octave.

Noone said otherwise.  That effect seems obvious to even me and I have basically zero interest in this topic, which is to say I haven't done any real research on the topic.  But what happens to the PF? 

I am sure you've built both versions and have the empirical evidence to support your "better designed filter". So tell us how it sounds sonically superior?