In <9610258489.AA848933452@mail2.creo.bc.ca>, on 11/25/96 at 06:39
AM,
   Ed VanderPloeg <evanderploeg@CREO.BC.CA> said:

>>     maybe 3 cm from the power contactor).  Here's why it works:

>     - nice omron relays with 2500 VAC isolation between contacts & coils,
>     and with 10A/250VAC contact rating, oversized to reduce arcing.
>     - 1N4148's as flyback diodes on omron relays in parallel with coil
>     - simple linear power supply for circuit, but with the transformer
>     ouputs isolated, Motorola 1.5KE's across those outputs to protect from
>     overvoltage, and properly size cap's after the linear regulators to
>     protect from slower dips & surges.
>     - separate external transformer to generate the 24VAC used to power
>     the contactor coils.
>     - careful board layout: no low power DC traces cross over the AC stuff.

>     Works great, I might add.  MOV's aren't designed for constant,
>     repetitive, high-energy surge suppression, they age with every surge
>     and eventually die.  The 1.5KE's are designed for this exact task.

Actually those capacitors you mention can help a lot. Any time you
clamp a spike like this, the resulting square wave is rich in
third order harmonics. Adding the capacitor keeps the rise time in
check, so the resulting EMF field is less likely to disturb a
properly designed Vdd or I/O line.

I did field service on a Z80 based controller that had its data
and I/O bus strung over about 20 cards in two separate rack
enclosures. A varistor fixed about 90% of the problems with
switching 24VDC. An additional resistor and cap was needed to make
the circuit work consistently without occasional glitches.

The other fix we need to tame this beast was opto-isolators for
the RS-232 lines to the terminal. This was the only sure fix for
ESD produced by operators fingers when it was very dry in the
winter months.

Best regards,

Mark

//Mark Timpany           timpany@ibm.net
//Jakarta, Indonesia     US Toll Free Fax 1-888-316-2357