We did some more testing, and here are the results:
1. The power state of the equipment doesn't make a difference. (We tried all nine possible combinations) All of the below tests were done with everything off.
2. The arcing only happens between the NR1604 & BD8002. We can't reproduce it between the NR1604 and the Epson, or between the BD8002 and the Epson
3. If the HDMI cable to the Epson is disconnected from the NR1604, there is no arcing.
4. If the HDMI cable to the Epson is half disconnected (ground only), there is no arcing. (????) That would suggest it is something on one of the HDMI pins rather then a simple ground issue.
5. Measuring the voltage across where the arc is being produced with a multimeter gives fluctuating readings, up to a couple volts AC and DC. Likely, the multimeter is too slow to show the voltage transient. I don't have the right equipment to measure ESD, and I'm not going to hook my scope up to something that arcs like this...
Doing a little more research, it looks like lots of people are having problems with HDMI connections causing arcs, some even to the point of damaging equipment. We're lucky that it our problem isn't as bad as what other people have encountered:
HDMI cable backfeeding Voltage from TV to cable Box
Given that many forums recommend grounding all of the equipment, it seems really odd that the Marantz equipment doesn't appear to have a three prong cable, or a good place to ground the units together. Any suggestions in that department?
We checked the polarity of our AC socket that everything is plugged into, and no problems there.