Re: EMI Filter Freq

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: "davep by way of Terry Fritz <twftesla-at-qwest-dot-net>" <davep-at-quik-dot-com>

Tesla list wrote:

> Original poster: "Jim Lux by way of Terry Fritz <twftesla-at-qwest-dot-net>"
<jimlux-at-earthlink-dot-net>

> Re parallel filters...

> > I would expect it to reduce the filtering, by roughly
> > 50%.  (6db).

> Hmm.. not entirely sure about that...

	Nor i, that's a case I never tested...

> (although in practice, mileage may vary)..  But,  if half the current
> goes through each filter, then half the "noise" also goes through each
>filter,

	Maybe.  I suggest that the power current and the 'noise'
	simply share the lines, but may not (do not) react the
	'same'.  Trivial distinction:
		Power is 'normal mode'
		'noise' is (mostly) 'common mode'.
	Put another way:
		The noise is just 'along for the ride'.
	Put yet another way:
		source and load impedance for the 'noise'
		is very different than for the power current.
	(in any case: I suspect the 'radiate around the filter' will
	be the largest effect in terms of noise as noise.  The filter,
	with luck, WILL drop out the conducted stuff.)

>While the noise and is attenuated by the requisite
>amount..., then summed back together... So you get something like:

> 1 filter:
> Signal Out = Signal In * .001                (for 30 dB)
 
> 2 filters
 
> Signal Out = 2 * ( (SignalIn/2) * .001) = Signal In * .001

	I think not.  Could be wrong.  cf above.  I 'model'
	(each) filter as a whole in a bucket.  Two holes, each
	the same size will leak twice as much.  (granted,
	this is an analogy, possibly nonapplicable.)

> However, since it is non coherent combining, etc...

	I think it may be coherent, thus:
		While the source is 'noisy', at any moment it IS
		coherent: there is one.
		The dimensions of the source (and wiring, and etc)
		are smallish, for most wavelengths involved.
	I don't see where/how coherency is lost.
	?

> I'm not sure it would work like this....

> > > this will make the .1uf caps = .2uf and the 450uH inductors
> > > = 900 uH.....??? maybe.

> > It's iffy to combine component values this way, when
> > each is part of a separate circuit.  And i expect the
> > inductors would be 'in parallel' (roughly) so the net
> > inductance would be _halved_: 225 uH.
 
> Say, though that the filter is basically a PI circuit C-L-C... hooking two
> in parallel will give you 2C - L/2 - 2C.   The "resonant" or cutoff
> frequencies should stay the same (since it is proportional to LC and 2C*L/2
> = LC...)  Parasitics might (will!) bite you probably...

	yeah.....  I went thru the same, sort of...  but the 'shape'
	of the roll of needs, i think, to be different, since L
	and C (total) are different?
 
> I suspect that most line filters also depend to a certain amount on the
> driving and load impedances... (part of those test conditions).

	100% right.
	Vendors (or purchasers, or standards bodies) define some
	test conditions.  Vendors design to them.  (knowledgable)
	users may design THEIR end of the assembly to meet them.
	Such numbers are mostly guidelines.

	In the real world of coiling we may be debating things of little
	purpose...  Especially as much will 'radiate around'.

	best
	dwp