[Occam]

For Occam and DAJ: A most common technique of reducing the effect of inductance in a capacitor is to bypass the cap with a smaller non/much lower inductive cap. Why pay big bucks for Auricap when you can bypass?
              d.b.

From my second post in this thread -

....
Typical industry practice is to use multiple different value capacitors to spread the attenuation spectrum. I've found that a triplet of .47, .1 and .01uf low cost X2 rated caps (used in concert with inductive components) can equal the performance of a single .47uf Auricap in the same circuit. While that makes tremendous sense from a manufacturer's perspective, for a DIYer, doing a oneoff, dunno.
......

Ummmm....Thanks? for the suggestion. 
My writing is admittedly a bit dense.

[Dan Banquer]

For Occam and DAJ: A most common technique of reducing the effect of inductance in a capacitor is to bypass the cap with a smaller non/much lower inductive cap. Why pay big bucks for Auricap when you can bypass?
              d.b.

From my second post in this thread -

....
Typical industry practice is to use multiple different value capacitors to spread the attenuation spectrum. I've found that a triplet of .47, .1 and .01uf low cost X2 rated caps (used in concert with inductive components) can equal the performance of a single .47uf Auricap in the same circuit. While that makes tremendous sense from a manufacturer's perspective, for a DIYer, doing a oneoff, dunno.
......

Ummmm....Thanks? for the suggestion. 
My writing is admittedly a bit dense.

Well I guess I missed that one!
A senior moment strikes again!

               d.b.

[Occam]

Hey Dan,

A missreading of a dense text is well worth the the favor you did me in real measurements of an old project. Often, an empirical smack upside the head is a great prompt to revisit some equations one took for granted. Recently, I've actually gotten access to both a 20mHz LCR bridge as well as an HP network analyzer, and I'm happy to report the high 'Q' resonance is gone with the realization that throwing more capacitance at the problem, without the appropriate inductance isn't worth it. And even with various damping networks possible, I hate to admit that for practical reasons (actually knowing the load), ideally,[but not necessarily practically, for marketing reasons] the conditioning is best integrated into the component. That being said, the typical Corcoms and the like available to us can certainly be substantially bettered.

Regards,
Paul

To everyone else who doesn't manufacture their own components, I still advocate the use of an appropriate external power conditioner, if mains conditions call for it.

[Dan Banquer]

Hey Dan,

A missreading of a dense text is well worth the the favor you did me in real measurements of an old project. Often, an empirical smack upside the head is a great prompt to revisit some equations one took for granted. Recently, I've actually gotten access to both a 20mHz LCR bridge as well as an HP network analyzer, and I'm happy to report the high 'Q' resonance is gone with the realization that throwing more capacitance at the problem, without the appropriate inductance isn't worth it. And even with various damping networks possible, I hate to admit that for practical reasons (actually knowing the load), ideally,[but not necessarily practically, for marketing reasons] the conditioning is best integrated into the component. That being said, the typical Corcoms and the like available to us can certainly be substantially bettered.

Regards,
Paul

To everyone else who doesn't manufacture their own components, I still advocate the use of an appropriate external power conditioner, if mains conditions call for it.

If one is going after low frequency attenuation in a AC line filter for linear supplies then yes it really does become a customized issue. At HF with a roll off starting at approximately 50 kHz for linear supplies the problems appear, from what I have observed, to be rather minor. Small amounts of peaking anywhere between 500kHz and 750kHz (with a very small bandwidth) is not what I call significant, especially when the peaking never goes above 0 db and is approximately 6 to 8 db above where it should be.
When you can send or post details of your new information I will be very interested.
                        d.b.

[JoshK]

This will definitely show my ignorance on the subject  , but is it possible to measure or see such resonances with an o-scope?  If not, how about a spectral analyzer with FFT capabilities?  It might be useful to sketch out a layman's understanding of how one would go about looking for such resonances. 

Thanks,

[ctviggen]

Josh, I purchased a digital oscilloscope with an FFT function.  This should be able to see certain features of a signal, although they may be overwhelmed by the massive spike at 60Hz.  In other words, the general noise in the analysis might hide the specific noise you're trying to find.  Dang, I wish I had more time to fool around with this stuff. 

[daj]

This will definitely show my ignorance on the subject  , but is it possible to measure or see such resonances with an o-scope?  If not, how about a spectral analyzer with FFT capabilities?  It might be useful to sketch out a layman's understanding of how one would go about looking for such resonances. 

Thanks,

Signal generator and either voltmeter or oscilloscope. Sweep frequency, look for voltage peaks across inductor. But at the higher RF frequencies, you're going to need special gear, not to mention special techniques to measure the DUT and not the parasitics.

[JoshK]

Thanks Daj!  I have the Stingray also, Bob. 

[ctviggen]

Cool.  You'll have to show me how to use it (mine sits, gathering dust), if I can ever get enough time to do DIY or other analysis.  I think testing is very interesting, as you really have to know what you're doing.  Any mistake will lead to incorrect conclusions. 

[JoshK]

I'm afraid that'd be the blind leading the blind. 

[Occam]

Bob,

Which specific O-scope? The easiest way to get an attenuation curve would be to feed a square wave from a signal generator at about 2-5kHz.(no one takes the pole frequency below 5k or thereabouts) You'll know the source breakdown of the fundamental and odd harmonics, so you can capture the filter's output and do that fft. Shift the frequency to capture any high 'Q' effects and to give more data points. You're going to be feeding a low voltage so you need to calculate your load to reflect your minimal current draw under 120vac, as filter behavior is dependent on load impedance. And as actual use of filters aren't into purely resistive loads, you'll possibly need some reactive terminination in the load. The PDF links that Scott provided address this as well as the different configurations you'll need to measure both common and transverse mode performance.

Or as daj mentioned, an appropriate adjustable sinewave signal generator with just the o-scope functionality should get you there. Beyond 10mHz, or thereabouts, those parasitics, interwinding capacitance of the inductance, will make you appreciate the efficacy of properly specified ferrites.

FWIW,
Paul

[ctviggen]

Paul,

Something similar to this:

http://www.dataq.com/products/hardware/oscilloscope.htm?source=googleos

I forget the exact make, manufacturer, and sampling rate, though.  (And mine isn't a function generator -- darn, that's a cool function to have.)

Thanks for the info.  That's very informative. 

[JoshK]

Paul,

This is the one I have (thanks to the rec on Hagtech's site), and the one I thought Bob had.  http://www.usb-instruments.com/oscillo_stingray.html

I also have a more traditional 5 channel Kikusui table top o-scope.

[ctviggen]

Josh, mine's a bit different, but similar enough. 

[pureAC]

Occam,

So the Hammond choke in not a common mode choke but a differential? In that case would large 16A CMC help in the filtration?

pureac

[Occam]

PureAC,

Rather than that JW Miller 8115 CMC, which is about the only Miller CMC with a reasonable current rating available at Digikey, go to Newark -
http://www.newark.com/jsp/search/advancedsearch.jsp
and enter in the part # '811' , select - starts with
in Manufacturer name, enter 'JW Miller'
and check 'In Stock'
and you'll see that Newark has in stock the larger 8116, 8117, 8118 and changing the part#, 8120 and 8121 CMCs for $9.31, which is less than Digikey offers the smaller 8115

Download the PDF.
The higher the current rating, the lower the inductance for their size (the leakage inductance inherent in their construction will also attenuate differential/transverse mode noiseand is augmented by the flanking caps).

The '-RC' suffix simply indicates RoHS compliance, and given a choice in N.A., the non RoHS compliant versions are easier to solder.

The choice of the specific model depends on the RMS and peak current draw of the components powered.
The 8120 works very well, and flanked by caps as discussed previously, can be quite effective in powering a full system, including a 100wpc class aB amp, and source components. YMMV

If you do an 'advanced search' with starts with '158x', you'll also find the CDE/Mallory X2 caps.

FWIW,
Paul