On Thu, 12 Jul 2007 19:37:21 +0000, Michael Black wrote:

Look in old ARRL Handbooks, or even their old Mobile Handbook.

Good afternoon, Michael.

Yeah, I gave away a stack of my old Handbooks a while ago... XYL got tired
of looking at them in the garage (hadn't finished setting up the shack
yet, and still haven't...). :-(

Figure out whether the signal is radiating from the chassis, or from the
output wire. If it's the former, it needs better shielding. If it's
the latter, add feedthrough capacitors in a box that shields the output,
and maybe put an LC circuit in series with the output line, and tune it
for a null. The primary of the inverter may also be a source of noise,
you need to figure that out first.

Well, square waves have a lot of harmonics, and perfect square waves have
infinite harmonics (but there's no such thing as a perfect square wave so
in the real world harmonics are finite). If I can file off the corners of
the stairstep wave that really should help cut down on the harmonics.

Someone emailed me and recommended a brute-force EMI filter (thanks, Dave)
so since that's easiest I'll probably try that first then take things from
there.