line filters

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

Original poster: "Mr Gregory Peters by way of Terry Fritz <twftesla-at-qwest-dot-net>" <s371034-at-student.uq.edu.au>

Hi all, below is a series of emails between me and John at roxburgh 
electronics. Hope it helps anyone who has had line filter problems.


MY ORIGINAL EMAIL:

John,
I own several roxburgh 25 amp cylindrical line filters, part number me> 
SDC 256. 
I was wondering if these filters can be used in reverse (by swapping 
line and load connections) to prevent interference created by equipment 
from going back into the power source.
Any help would be great. 
Cheers, 
Greg Peters 


JOHN'S REPLY

Greg, Hi! 
Re: filters, part number SDC 256. The filters will work either way 
round, in fact there is the principle of reciprocity that says the 
filters work just the same either way round. In practice with the 
source and load impedances being ill-defined, frequency and time-
dependant, there is a better orientation; usually the noise sources and 
line impedances are such that the rule of thumb is to point the 'Y' 
capacitors (those connected to ground) at the noise. The beter 
orientation can be determined by comparative measurements or 
empirically by observing the effects of the filter in circuit.
What is the source of the noise in this instance? 
Best regards, 
John Taylor 


MY REPLY:

John,

Sorry to bug you again, just for clarification:
You said that the "y" capacitors should be pointed towards the noise 
source. However, the schematic on your web site says that, in a normal 
mains filtering configuration (where you want to stop RF going into the 
equipment, ie: the noise source is the power line), the "y" capacitors 
should point towards the load. This is the opposite to what you 
suggested and now I am confused.
Cheers,

Greg Peters


JOHN'S REPLY

As I indicated, there is a rule of thumb, and the difference between 
this way or that way may, in the case of a single stage (single 
inductive stage) filter, be quite small. The majority of filters that 
are intended to stop noise getting into an equipment deal with low 
energy, fast transients in the common mode. Most of these fillers are 
made up as an addition of 3 capacitors and a double-wound choke on the 
back of an IEC appliance inlet; the mechanical construction of these is 
such that the Y capacitors are most easily installed at the back of the 
metal housing rather than at the front at the mains input inside the 
[plastic] molding of the inlet. This means that for the last 30 years 
(Corcom and Schaffner were the founding fathers of this filter) the 
founders of the appliance inlet filter have successfully supplied 
filters working in the time domain with low energy transients or 
glitches. The debate as to orientation has disappeared into the mists 
of time. 
As far as the noise-producing equipments are concerned UPS, PSU, 
drives, Tesla coils etc, the noise spectrum is made up of higher energy 
pulses in both common and differential modes, here the Y capacitors are 
at the load end or the filter for performance reasons. 
In a given instance the better orientation can be decided as mentioned 
earlier by measurement, it is dangerous to be dogmatic by declaring 
that all similar installations will be best dealt with in like manner; 
the problem of unknown terminations and multiple noise paths in a real 
life instance may go some way to explaining my grey hair!
In due course I will have a look at our marketing-inspired comments on 
the web and take any action which may be appropriate.