Gentlemen,
There are two distinct issues here: Differential Mode noise and Common
Mode noise.
Differential mode exists from one conductor to another in the same bundle
or harness. It is probable generated intentionally within the device in
question, or some artifact of that generation.
Common mode exists on ALL of the conductors in a bundle or harness with
the same voltage and phase relationship to some external reference.
With respect to AC power generation or any circuit carrying high DC
current, the filter must take this current into account in its design.
A common mode filter passes ALL of the CURRENT-CARRYING conductors through
the series element, but NOT the non current-carrying conductors (i.e.:
safety or shield ground). There must be NO SERIES IMPEDANCE in these
conductors; it defeats their purpose and may create unsafe conditions.
The higher the frequency, the more likely the noise is mostly common mode.
By 20 or 30 MHz noise is usually nearly all common mode. Below 2 or 3
MHz, differential mode becomes prevalent. By 100 kHz noise is almost all
differential.
Common mode noise is most likely picked up through the antenna input.
Common mode filters for AC lines must pass all of the phase conductors
through the single series element. In the case of a ferrite toroid, L1,
L2 and neutral contain a near zero sum of the powerline current, but
effectively choke all of the common mode current. A set of shunt elements
(bypass capacitors) on the load side of the series element now provides a
low impedance to ground, completing the L-section filter low-pass filter.
Differential mode noise is most likely to enter the device directly
through the power supply connection.
Differential mode filters must act independently on each line
independently (Although filtering just the ungrounded side usually works
as well.), so the series element must provide low impedance at the power
frequency and be large enough cross sectional area to avoid saturation
(simple if it is DC, but saturation is more of an issue) and high
impedance at the noise frequency. This requires much more careful
selection of the choke cores. The shunt elements require the opposite
characteristic in frequency domain: low impedance at the noise frequency,
high at the power frequency. The circuit configuration is the same, with
the shunt element on the load side of the series element for an L-section,
low-pass filter.
All filtering should be done at the source if possible. At the very least
it will protect all nearby devices, and it always remains with the source.
Additional filtering of differential mode noise is sometimes required at
the receiver.
Note that if the power source is three phase, L1, L2 and L3 all must pass
through the single series element in a three-wire delta system, but L1,
L2, L3 AND neutral fro a four-wire wye system.
Twisting follows the same rule: single phase: L1, L2 and neutral, three
phase: L1, L2 and L3 in a three-wire delta system, and L1, L2, L3 and
neutral fro a four-wire wye system. Do not use separate twisted pairs,
The ground wire may be included in this section of the line. and is often
the messenger or support wire in overhead systems.
Remember: No series impedance in any ground conductor!
One more note on grounding: There should be only one system ground. Codes
usually require this to be at the entrance panel and the bond to neutral
must be only here.
The NEC has recently been amended for remote generators. Newer generators
have a screw that can be removed to isolate the local neutral from the
case/frame, which still must be grounded!
In the case of inadequate (ground noise issues) receptacle grounds, a
separate driven or other ground system may be employed. This must be
permanently marked by using orange service devices (receptacles).
For reference, I am the author of the DCX two-way radio installation guide
(available from dealers as a service bulletin or from the ARRL website
under RFI) and the corporate grounding design standard. I am also a
member of the SAE subcommittee on electrostatic discharge in fuel systems.
Bill Gilmore WB8FPQ
Electromagnetic Compatibility Engineer
Core EMC, Electrical/Electronic Engineering
Powertrain Component EMC / RWD, SRT & HEV Platform EMC
DaimlerChrysler Corporation
Auburn Hills, MI
"Jim P" <jvpoll@dallas.net>
Sent by: rfi-bounces@contesting.com
03/23/2005 01:50 PM
To
"Jim Brown" <jim@audiosystemsgroup.com>, "RFI List" <rfi@contesting.com>
cc
Subject
Re: [RFI] Honda Generator RFI
You brought it up Jim; and I can see how, given
the prescription for looping the neutral and just
one phase through 'the magnetic' back at the
source where this saturation problem that you
spoke of can become a problem; I simply gave
the cirumstances under which this can occur
Quoting J. Brown from earlier:
As to cores -- they are likely to saturate under
load if you put them on individual conductors, so
plan on putting them on the paired cable so that
the total field is small (that is, the hot is cancelled
by the return). Used this way, the cores will
attenuate common mode radiation from the line.
Simply placing a 'core' on either phase plus neutral
back at the source (we _were_ talking about suppres-
sing gen noise IIRC) results in a _full_ magnetic field
contributed by the single phase, during balance, with
little or no 'countering' magnetic is produced owing to
little or no current on the neutral at/near the gen. This
is really pretty basic stuff ...
A better location looks to be placement around both
hot-conductors, such that at least partial cancellation
of the field produced by the 60 Hz line current while
suppying common mode rejection to noise coming
from the gen would be achieved.
My choice? I'd buy a commercially rated filter and
escape all this napkin-based 'ballpark engineering'.
Jim P / WB5WPA /
---- Original Message -----
From: "Jim Brown" <jim@audiosystemsgroup.com>
To: "RFI List" <rfi@contesting.com>
Sent: Wednesday, March 23, 2005 11:07 AM
Subject: Re: [RFI] Honda Generator RFI
> On Wed, 23 Mar 2005 10:19:21 -0600, Jim P wrote:
>
> >It does/it can, if you saturate the particular magnetics
> >selected, by improper application, which was what I
> >was pointing out
>
> What magnetics are being saturated, and how? If all the current-
> carrying conductors are going through the ferrite choke, the net
> field is limited to the common-mode current, which ought to be the
> noise, unless something is broken (or shorted to a conductor that
> doesn't go through the choke).
>
> I wouldn't expect a power transformer to generate RF trash upon
> saturation. What else might I be overlooking?
>
> Jim
>
>
>
>
_______________________________________________
RFI mailing list
RFI@contesting.com
http://lists.contesting.com/mailman/listinfo/rfi
_______________________________________________
RFI mailing list
RFI@contesting.com
http://lists.contesting.com/mailman/listinfo/rfi
|