[R-390] RE: GFCI trips!

[Drew Papanek]

Paul Galpin wrote:

>AFAIK, modern equipment with built-in mains filters does not connect to
>Earth. It relies on series chokes, and parallel capacitors across the power
>supply. Presumably, this is because of the problems caused to ELTs (GFCIs).

(snipped)

In equipment filters and prepackaged IEC cord connector/filter modules I 
have seen here in the US, capacitors from each side of the line to ground 
(earth) are used.  These capacitors are typically .0047 uF and so pass only 
a small leakage current.

The usual configuration has a capacitor of about .1 uF from hot to neutral 
followed by an inductor in both lines.  At the output is the pair of .0047's 
from hot to ground and from neutral to ground.  Some of the better filters 
precede the aforementioned arrangement with another capacitor across the 
line followed by another pair of inductors.

I have seen the schematic for some of the medical type filters; they appear 
to be like the others but without the caps from line and neutral to ground.

In all of the more modern designs that I have seen the inductors are both 
wound on one core.

I opened up a line filter from a '61 Teledyne-Amelco (filter bearing a date 
code of the same year).  Relatively easy, but dirty and smelly torch work.  
The arrangement was essentially 2 seperate filters, one in each line.  The 
incoming line (hot or neutral as applicable) was connected to a small 
toroidal inductor.  The output of that inductor went to a .082 uF cap to 
ground and to the input of a larger toroid.  The output of the larger toroid 
went to another .082 uF to ground and to the filter's output terminal.
That makes a total of 4 caps and 4 inductors, in two separate L-Pi networks.

The original filter might be restuffed/reconfigured as in the "better 
filter" configuration mentioned above.  From line to neutral would be, say, 
.1uF, both lines then go to both small toroids. The outputs of the small 
toroids would be bridged with another .1uF, then feed the inputs of both 
large toroids. The outputs of the large toroids would each have .0047 uF to 
ground and would feed the filter unit's output terminals.

I have no idea of how well that configuration would work using the original 
seperate inductors instead of the more modern ones with 2 wound on each 
core. The mutual inductance/magnetic coupling between lines in today's 
filters must have an effect.

Line filtering capacitors MUST be types specifically rated for that service. 
Mouser, Digi-Key and many other suppliers carry them.  Do not use capacitors 
which are not rated for line filtering service no matter how high their 
voltage rating might be.

The caps which are connected from line to ground or chassis and whose 
failure in a shorted mode would present a shock hazard are designated Class 
Y.  Class Y caps are designed so that the predominant failure mode is open 
(fail safe).  For across the line use where failure would not present a 
shock hazard, Class X caps are used.

Both X and Y type capacitors are designed to reduce the risk of explosion 
and fire when subjected to large transient voltages.

I have seen in equipment of 20-25 years ago another line filtering 
configuration.  That older arrangement has at at the filter's output a cap 
of .1uF or so across the lines and a larger cap (1 uF) line to ground. That 
arrangement seems to be no longer used.  A fault condition in supply wiring 
with hot and neutral swapped AND with ground not connected would make the 
chassis mighty hot.  I can only guess that is the reason for that 
arrangement's demise, and for the introduction of today's two small caps to 
ground.

Finally, one could just install one of the prepackaged connector/filter 
modules (as seems to be popular nowadays) and dispense with all the filter 
re-engineering.

Drew