corresponding
to this event show a load of zero before (the Combox, unlike the SCP, incorrectly reports
exported power) and no more than 2900W/25A afterwards.
As a separate experiment I disconnected the DLP again and noted that the load as shown by the SCP
was 950W. I threw the upstream (AC1) breaker and the XW+6848 immediately reported a W63 warning which
could not be cleared. It did successfully sustain the ~1kW load in inverter mode, but only
with the W63 warning asserted. The Conext Combox logs corresponding
to this event (16:59:29) show a load of 800W/10A before and no more than 1800W/15A while inverting. The
PQ analyzer shows that low load as well.
I was unable to clear the W63 until the XW+6848 returned to AC pass-through mode.
The XW6048 that this XW+6848 replaced could (and frequently did) sustain inverter mode
loads close to the 6kW rated capacity, without (ever once that I remember) throwing a W63
or F63.
Something is clearly very wrong. I have extensive data now from multiple F63 events, will
have to see what Schneider tech support wants.
UPDATE 2015-02-19
I had an extended discussion with a Schneider technician. He wanted photos of the wiring and -- of
course -- a reset to factory defaults. I took a shot of the terminations inside
the XW+6848 itself, one underneath it, and one of the
Power Distribution Panel.
Those connections were tightened with torque wrench1;
something that is
hard to prove photograhically and that professional electricians laugh at me for doing2.
UPDATE 2015-02-20
I did the requested "reset to factory defaults" which consisted of throwing the breakers to bypass the
XW+6848 inverter and connect the critical loads panel directly to the grid, disabling the inverter, resetting
to factory defaults, and painstakingly re-entering all the configuration parameters again.
I was able to do that in only 16 minutes (thanks to lots of practice from the AGS saga). I then enabled the XW+6848 and
reset the bypass breakers to put the critical loads downstream of the inverter again. Since the damage from the
outage was already done I experimented with cutting the grid supply breaker (AC1) while the critical loads varied
from 1800W to 2900W. No W63 or F63 errors this time!
That is good news in one respect, but bad in another. This inverter seems to be susceptible to random software
glitches. Sustaining two power interruptions and spending 15-20 minutes re-keying configuration data isn't something
we want to do often.
During the AGS saga I consulted with three different EEs who were interested in the problem. Two of them suspected
electrical noise issues (and indeed our utility power here is "dirty"), and suggested use of an
EMC/EFI filter such as
this. I purchased one rated 55A, seen here, but since the inverter was still under warranty
I didn't want to install it without the approval of the vendor, Schneider. I asked about that several times and never received
a clear response, and in the meantime the determination was made that the AGS problem lay elsewhere.
Mysterious software glitches now seem to be an issue again, and I'd really like to find a way to prevent recurrences. This
filter will fit in the space under the XW+6848 inverter. I will try again to get a
yeah-or-nay answer from Schneider, and if they don't object I will install it and hope it will reduce the need for
factory resets" and spending quality time standing in front of the SCP3 re-keying the
complex configuration.
UPDATE 2015-02-25
After badgering Schneider tech support for several days I finally received an E-mail from them stating that they
had no objection to installation of the aforementioned EMC/EFI filter. I was able to install that
fairly neatly without an outage
by switching to generator power to kill the AC1 input circuit. I may not know for a long time if it makes a
difference.
_
1 That photo was taken at the time of the initial installation of the new XW+6848, during the
AGS ordeal. It doesn't prove I correctly used the torque wrench,
but does prove I had one.
2 With expensive equipment like this that has warranty support I try very hard to do as much as I can
to avoid finger-pointing (I use anti-static wrist straps when working on warrantied electronics, for the same reason).
3> The configuration can also be entered via a web browser and the Conext Combox, but that is both slow and
buggy. It's faster to punch it in via the SCP. It's a pity the Combox doesn't provide an option to save and restore
the configuration from disk.
