An unintended current path wreaks havoc
On my first EE job out of college I worked on a project that involved
designing an ITU (Interface Test Unit) to interface a legacy tester
with a modern digital DUT (Device Under Test). As a part of the ITU, we
had panel-mounted an active, off-the-shelf RJ-45-to-BNC Ethernet
adapter to the front of the ITU with the BNC connector facing the
The Ethernet adapter came with a typical "wall-wart" style power
supply. Since the legacy tester and the test environment didn't have a
provision for 120VAC household power, we cut the cable from the power
supply and wired it into one of the tester's DC power supplies by way
of a terminal block.
The first part of power-on testing of the ITU included leaving the DUT
disconnected and all of the dust caps for the ITU external connectors
in place. The Ethernet adapter had a metal BNC dust cap screwed in
place that was also secured to the aluminum front panel of the ITU with
a ball-chain to prevent an operator misplacing it (like one of those
pens chained to the counter at the bank). We applied power, the
Ethernet adapter’s lights came on, and everything appeared good.
By chance, I accidentally brushed my hand against the chain dangling
between the front panel of the ITU and the dust cap covering the BNC
connector. I immediately pulled my hand away as my brain suddenly
registered severe pain on that extremity. Much to my lab assistant's
surprise, I hit the e-stop and declared that I had just been
After a bit of confusion, and a bit of nerve calming about my having
just seriously damaged some very expensive equipment, I determined that
the bubbling blister on my hand was, in fact, a burn from a
super-heated dust cap chain. We took a closer look at the setup and two
observations finally led us to the explanation: First, the BNC bulkhead
connector that was attached to the Ethernet adapter’s BNC side was
insulated from the front panel by a plastic o-ring. Second, the
negative side of the power supply was connected to the Ethernet adapter
via a terminal block that was connected to chassis ground.
Using the two prong off-the-shelf AC power supply, the Ethernet
adapter’s BNC connector shield would normally have been floating.
However, when the negative supply was grounded the shield now had a
fixed reference to ground (not zero!). With the dust cap applied, a
current path was created to chassis ground through the ball-chain
which, apparently, had a resistance high enough to prevent an over
current condition but low enough to allow it to gradually heat up. We
disconnected the negative rail from chassis ground and the problem went
I still have the (albeit tiny) scar on my right hand. Fortunately, not
all lessons in engineering come with such a permanent reminder.
is an EE in the Defense Aerospace industry. In his free
time, he dabbles in paintballing and flying model helicopters and is
enjoying life in California with his wife and three sons.