[sdiy] Mains interference

[cheater00 .]

On Fri, Mar 28, 2014 at 9:43 AM, Needham, Alan
<Alan.Needham at centrica.com> wrote:
> Is it possible that too much emphasis is being placed on earth integrity here?
> My home shares a 100kVA pole transformer with about 20 houses (3 phase, so not exactly the same).
> The earth in our house is the incoming lead water pipe - less than ideal.

Lead is a good conductor, as evidenced by the decades of use on PCBs.

> Could the radiation problem be line lengths? If a mains run just happens to be tuned to some magic RF frequency?
> I wonder if a 240v isolation transformer could be borrowed to feed an offending device to break that 'tuned loop'.

Yes, but it has nothing to do with RFI (not sure if you actually meant
it does). The isolation transformer would be an impedance in series
with the offending load.

Earlier, Richie wrote:

> The trick to controlling EMC problems like this is to find the source of the
> high di/dt and then try to return the high-frequency current back to the
> source by the shortest path possible.  You can add to this by saying that
> the path taken should preferably not include going via what ever equipment
> is susceptible to RFI!  So, for example, a thermostat in a fridge should
> have a suitably rated capacitor connected across it right near the contacts
> to allow high-frequency energy to circulate and decay close to the
> thermostat.  Keeping the current loop small stops it affecting the supply to
> other things, and the small loop area also makes a terrible radiator of RF.

That is only half the truth. A capacitor will always have higher
impedance than a piece of copper. Let's not forget that the current
will be formed where the highest voltage is put through the lowest
impedance. Effectively, the output impedance of your mains feed up to
that capacitor together with the capacitor form an impedance divider
(think AC resistive divider). To make the current reluctant to form
through the mains, you need to make the mains output impedance higher.
This is done by adding a series impedance. One possibility is a
transformer, another is a series inductor, another is a resistor. The
resistor always works, the transformer might work (but then doesn't
its impedance only come from heating losses anyways?), I'm not sure
what all the disadvantages of an inductor are. When using a resistor,
make its impedance (= resistance) about 50x higher than the
capacitor's ESR at the relevant frequencies you want to block from
going up the mains. This will give you rejection numbers in a high dB
range directly calculated from the impedance divider.

As an example, you may check out my TB-303 mod thread in which I have
started with a bank of very large, C-Cell sized bypass capacitors to
bypass the LED driver, but adding a series resistor to the rail let me
reduce that to a tiny capacitor with the same audible rejection ratio.
The resistor was in the 100-150 Ohm range and the capacitor was "low
ESR", which I believe means 1.5 Ohm or less.

D.