[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [TCML] RFI EMI Filter question and PC survivability



Hi Jim,

That is a really nice and educational summary!

All monitors should have at the very least a sizeable Ferrite barrel clipped
around the shielded cable. I remember a fix that was sent out by let us say
Viewsonic which was a large ferrite clip on barrel. I don't know how many
VGA adapters failed or got flaky before the kits were retrofitted. Now that
was presumably without a poorly grounded Tesla coil just running near
breakout running in the neighborhood ;-)

Ground loops are an audio engineer's night mare in living color, especially
not so way back in the analog days. At least you could hear those! Studio
design is interesting from what I have read and seen.

I guess RF though gets around anything though that's not isolated in a
Faraday cage. Hole size of course even becomes an issue at the really short
wavelengths. We used to have a big, very elaborate fine copper screened one
when I worked for a science company in Illinois doing microvolt stuff. Even
the door seals are impressive.

My personal computers and servers (its what I do) are housed in a shielded
19" rack powered by a rack mount APC 3000 on a short run, dedicated 30 amp
circuit. This feeds very fast frame mounted transient protected outlets
inside the rack. Ideally, an online ups should be used but is way outside my
budget! I am only a few feet from the mains ground rod or I would have
driven one through the floor right next to the rack or inside it.

I suppose I could still kill the servers if I build that Marx generator I
have been cogitating ;-)

Jim Mora

 -----Original Message-----
From: tesla-bounces@xxxxxxxxxx [mailto:tesla-bounces@xxxxxxxxxx] On Behalf
Of Jim Lux
Sent: Wednesday, January 28, 2009 9:34 AM
To: Tesla Coil Mailing List
Subject: Re: [TCML] RFI EMI Filter question



-----Original Message-----
>From: BunnyKiller <bunnikillr@xxxxxxx>
>Sent: Jan 28, 2009 8:29 AM
>To: Tesla Coil Mailing List <tesla@xxxxxxxxxx>
>Subject: Re: [TCML] RFI EMI Filter question
>
>Hey Thomas
>
>probably should also do 
>Line=Filter=Computer
>
>Scot D
>

You need to think about where the interference is coming from and going to.
The usual line EMI filter is designed to keep noise *inside* the box it's
attached to, and the noise they're concerned about is usually fairly high
frequency stuff (digital clocks and such).

If you're trying to keep tesla coil transients out of other stuff, it's down
in the hundreds of kHz range, and the coupling mechanisms are different than
for the usual consumer electronics EMI/EMC.


First, consider the TC without breakout.  It's basically a big RF tuned
circuit, and your stuff is in the field of the capacitor.  So, you want to
make sure the RF current in that capacitor stays near that capacitor, and
isn't coupled out somewhere else. The coupling happens either by capacitance
(i.e. you've got some sort of "antenna" that couples to the electric field)
or by inductive coupling (you've got a wire that picks up the magnetic field
from the current flow)

If the current flow (from top of secondary to bottom of secondary) is via
some wire, then that wire radiates RF, which will be picked up by other
wires that are approximately parallel to it. So, the take home is: no long
wires carrying significant RF current.  Short wire from bottom of secondary
to ground plane, which is right under the coil.


Now consider sparks and breakout.   This is a bit different.  You have a
spark which connects the top load to the ground, and the rise time of the
spark is a LOT faster than the hundreds of kHz.  What you now look at is a
loop carrying a fast rise time pulse, radiating a magnetic field. (and, of
course, the field is proportional to the 1/rise time and the current being
carried).  That pulse gets picked up by other conductors, and because it's
fast rise time, it can induce bigger voltages (V is proportional to
d(field)/dt).  

But those pulses probably won't be stopped by a EMI filter.  Why? because
it's the ground wire picking up the pulse. Say you've got a PC connected to
the wall socket and then connected with a cable to a monitor, which also is
connected to the wall socket.  There's a loop formed by wall socket: "ground
wire in the power cord of PC": "chassis of pc": ground wire in video cable:
chassis of monitor: ground wire in power cord of monitor, wall socket,
ground wire between wall sockets.

That loop picks up the transient field from the TC spark (or Marx bank
spark.. a real equipment killer) and a voltage is induced around the loop.
Not only that, but the current flow in the loop can couple the transient
into parallel wires in cables.  The power cords aren't a big deal (after
all, there's a filter at the equipment end), but the monitor cable probably
isn't filtered, so now, the transient couples back into your video adapter.
Ooops, dead ICs.

And, the line filter on the PC might not have very good isolation for a 100
MHz pulse.  The fast rise time pulse from the spark (we're talking
nanoseconds, here) might couple by leakage capacitance right over the filter
into the PC.  Remember, the filter on the PC was designed to keep PC noise
inside, not 100 MHz transients out.

_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla

_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla