Re: EMI filtering for the HiresEEG

From: sleeper75se (sleeper75se_at_yahoo.se)
Date: 2002-03-14 18:26:49


--- In buildcheapeeg_at_yahoogroups.com, Sar Saloth <sarsaloth_at_yahoo.com> wrote:

> Point taken, no need to get snarky, even if I do deserve it.

I'm sorry, I wasn't meaning to use more than
a hint of sarcasm. I should have learned by
now that text does not carry some emotions
very well, so I am very sorry if you were
taken aback by my comment. It was more
intended as a positive dismissal of your
apology for not suppying proof, because the
improvement you suggested will obviously work.

Anyway, I added your suggestion to the design
right away (on pure "faith" so to speak).
The following question about extra enclosures
is entirely serious.

> I wasn't actually thinking of $2.50 toroids,
> I mean isn't that close to the price of a big
> ferrite sitting on a power cable anyway? (which
> is the last minute fix that my well-intentioned
> comment was trying to eliminate). But if you
> are really winding your own inductors on $2.50
> toroids, why not just add some wire and make a
> real common-mode choke for one of the earlier
> input stages? EFT is one of the harder tests
> to pass anyway.

The toroids I was thinking of are fairly small, 7mm inner diameter,
so two wires will probably not fit. However, winding both wires on
the same toroid is of course the way to do it. Doh! I guess I wasn't
thinking there... %-).
Just using a slightly bigger core (they don't cost much extra) will
work of course.

Hmm, maybe it is best to go all the way with this - winding both
ground and +12V on all three toroids before the voltage regulator.
What do you think?

> In fact, why use a toroid? (that is a real question, I have some
> understanding of magnetics, but when it comes to fancy filtering,
> I just use a sort of shot-gun method).

Ok, think about the magnetic flux in a regular rod-shaped inductor.
The magnetic field flows in a circular pattern, first moving inside
the rod from one end to the other. Then it leaves, and flows back to
the starting point on the outside. Since a toroid is basically a rod
bent back on itself, a greater portion of the magnetic field stays
within the core. In other words, they are much more quiet, and reduce
the risk of transmitting the noise somewhere else into the circuit.
Worth the extra money I hope.

>
> Really, I was thinking of some $0.10 ferrite beads instead of $2.50
> toroids. I have never done a step-by-step test to prove which
> components are useful.
>
> From a practical point of view, even if everything I said was
> correct, (which we know it wasn't) my points are somewhat
> irrelevant. [...]

Wrong. You *are* correct, and your points are not irrelevant, far
from it. So far the points you have been raising have not been
trivial or taken time away from anything either. Trust me. :o)

Discussion makes a person look at things from new angles and can push
a project in a new and hopefully better direction.

[...]

> And in the off-chance that your suggestion was not sarcastic,
> I can tell you the following from experience. There are some
> medical devices on the market that have no shielding whatsoever.
> (The Stardust portable sleep recorder is an example, although
> it has no EEG inputs, it does have a pulse oximeter which is a
> very noise sensitive device). Many other devices have shielding
> in the box (or a shielded box) in order to pass the 3V/m radiated
> susceptibility.

Hmm, maybe if they all did it right, we would be able to use cell-
phones inside hospitals? Not shielding medical equipment sounds quite
reckless, don't you agree?

> If you wanted to be ultra paranoid, you could always lay out your
> board to take your shielded ideas as an option.

I am that paranoid. :o) Consider this: I am torn between letting the
toroids lie down (against the ground plane) or stand up. If they are
standing, they would use much less board space and I could build two
devices on a single euro-size PCB. But then I might get more EMI...?
(If you have suggestions, let me know. :-)

> My suspicion is that many devices with no shielding at all are
> improperly tested and approved. I have seen one EMI test report
> from a European notified body for a medical device that clearly
> did not understand the test specifications. [...]

Um...you mean the testing lab failed to understand how to test? Oops.
First they do a poor job and then they charge you $7000. Let us hope
that was a one-time event...

> Just trying to be helpful - Sar

Looking forward to next time...

Regards,

Andreas



This archive was generated by hypermail 2.1.4 : 2002-07-27 12:28:41 BST