EMI and UD2.7C

[alan sailer]

This post describes a case of electromagnetic interference that I found on my
mini DRSSTC and what I did to fix it.

While starting to test and tune my mini DRSSTC (full bridge SKM100 @ ~150kHz)
I noticed a problem at about 80 volts. As I raised the interrupter pulse width I
noticed that the spark output of the coil made a sudden jump. The collector
current trace would go from ~100usec to over 400usec.

I quickly decided interference was a problem. I have built several coils and not
had this problem but there is always a first time.

Probing the output of the hex inverters buffering the optical fiber receiver
showed this;



The red trace is the tank current and the blue is the interrupter. It seems obvious
that some signal from the coil is coupling onto the interrupter pulse and causing it
to enable the coils oscillations past the cut-off of the interrupter. The next question,
is the interference magnetic or electromagnetic? I removed the secondary to test this
question. If the interference remains, then it is probably magnetic.

The interrupter pulse with the secondary removed showed that the interference
was electromagnetic. This is good because I didn't want to buy Mu metal to shield
magnetics. Copper I have.



So now the idea was to Faraday shield the UC2.7C. Since the operating frequency
of the coil is low the wavelength of the EMI should be very large meaning that the
screen doesn't need to be all that exotic. However I didn't want to do this twice so
I went for overkill. Plus it was fun trying to build the best shield I could.

Here is the result;





The box was built using double sided copper plated epoxy board. All the edges
were soldered or covered with copper tape. The lid is press fit copper screen,
folded and soldered at the edges.

The box woks, I have seen no signs of interference. I tried to probe the interrupter
pulse by passing a wire through the holes in the screen but the probing coupled
the signals back into the UD board causing the original problem to return.

I hope this may help someone with a similar issue.

Cheers.