Help with Noise Injection into Circuit

Is your 'coil' made with nonlossy core material? If it were laminated iron, the risetimes might become more moderate. If you can add a small winding with an RC snubber load, your high-efficiency core can emulate a cheaper iron unit.

The stray inductance in series with the spark gap is also important in keeping transients in check.

Yes, of course a ground wire that carries high current at high frequency has magnetic coupling (not antenna coupling, exactly) to nearby wiring.

Thank you very much for your feedback,

I cant find any info on the ignition coil as far as materials are concerned. (destructive testing with saw to follow ) So If the coil and wire inductance are equal to X and I must maintain said length and wire size how could I compensate for this to have minimal transients?

By winding you mean a R-LC tank circuit? What could the configuration be?

Would it be somewhat of an impedance matching circuit for less reflection. Or just minimizing di/dt?

Also why are fast recovery diodes not conducting these -/+ V transients fast enough to keep the mcu in check.

The most promising avenues to follow for reducing magnetic interference, which this almost certainly is, are (1) slowing the risetimes - which you're obviously aware of; and (2) reducing loop areas. Can you, for example, twist the cables to / from the items of equipment?

Also someone posted here a few weeks ago: "Magnetic fields are dipole fields, and therefore the field strength diminishes with the cube of the distance. This means that quite modest separation increases attenuation a lot."

The coil may be generating a non uniform (in space) magnetic field, you could try rotating it 90 degrees to see if that helps.

Have you tried a ferrite bead in the earth return? Remember your ferrite components need to be rated not to saturate with the current passing through them, or they'll stop dissipating in an optimally lossy manner.

yes you can determine if it is magnetic by rotating the axis of the coil to see if that changes the interference... if so , try shielding with a copper strap like the way some power transformers have a copper strap around them...

Mark

Thank you all again for the suggestions. I tried rotating the coil with no changes. I minimized the power/signal lines as well, it reduced the transients signifigantly but they still remain around 10-20 volts. I have tried RCD snubbers with no change, and beads as well. Is there a way to mitigate these transients if I have to have this length of wire?

On 3/10/2010 5:51 AM, Johnny5 wrote: (...)

Try reducing the inductance of the wire between the coil and your driver transistor. This should turn your voltage pulse into a current pulse.

First prize: Flat copper bar. Second prize: Flat copper braid. (Salvage from old, large diameter coaxial cable).

--Winston

Toroid core coils do not notably emit. The problem is elsewhere.

Your AC input should be a twisted pair once within the chassis, OR STP (Sheilded Twisted Pair). All DC feed lines should ALSO be in twisted pairs. The feed, and the return.

interference,

which

dipole

the

attenuation

you

ferrite

manner.

Are you doing everything with twisted pairs or coax?

Or non-inpregnated solder braid.

Silver plated copper wire also makes good high current feeds for short jumps.

Thank you all for your help, absolutely the strangest case of RFI/EM enduced cmos latchup I have ever seen. No layout issues just some carefully placed bat54's.