Reality versus the "last" final stats. [RE-wrenches]

Fri Apr 25 14:30:15 PDT 2003

Wrenches:

I have been out of town for a few days at, coincidentally, a DOE
inverter reliability workshop.  Sorry I was not able to chime in here
sooner. 

With all deference to Jeff, his experience is an anomaly which does not
agree statistically with the installed base of inverters in the US.  I
also disagree with his conclusions, his definitions, and his methodology
for calculating reliability.  What I do know is that it is always the
inverter's fault no matter the real root cause of the problem. 

I have not yet analyzed the data for 2003 yet but...  Other than the
Firmware V8.65 / Vac_srr issue, which I think has been largely resolved,
I do not see any patterns that would indicated that there is a
significant change in the number of failures we are seeing.  What I can
say, definitively, based on real data through the end of 2002 is: 

Our total return rate through 2002 for inverters sold in the US was
4.3%.  Of these, about 15% have no problems found after they are
returned. Not including the "no problems found" units, our actual total
failure rate for inverters is 3.8%. 

The largest number of failures to date is 1.4% for the dreaded Vac_Srr
error message with which many of you are all too familiar.  I am sure
this will also be our largest failure related issue when we do the
numbers for 2003. Unfortunately, we did not fully understand the problem
for a couple of weeks and so there were more units than I would have
liked that made it into systems.  We retested and reworked many
inverters here at SMA America and recalled all the ones we could track.
We also did our best to get the word out to try to find the others we
could not track in an effort to solve the problems as quickly and
painlessly as possible for all concerned.  

For those not aware of this problem: 2500's with version 8.32 and 8.66
firmware are fine.  Units with 8.65 firmware can trip off line at too
low a voltage and will display a characteristic error message Vac_Srr.
These units can be upgraded in the field with new firmware which we
supply at no charge. We estimate that the firmware fix will solve the
problem in about 80% of the cases but that some units will need to come
back here for replacement of two small capacitors.  Contact Jason Mills
if you have any questions or need help with Vac_Srr issues.  

This leaves a remaining failure rate of 2.37%.  Of this, the most common
failures are: 
0.50% - Unknown failure reported by the user - troubleshooting
inconclusive.  
0.51% - Vac_Bfr error message.  This related to a bad batch of ceramic
surface mount capacitors that developed micro fractures typically after
a few days or weeks of thermal cycling. We changed capacitor vendors and
the problem has disappeared.  
0.31% - Ground fault will not clear.  More than half of the "no problems
found units" were returned for ground faults.  We believe this indicates
a lack of understanding of ground faults and how to troubleshoot them.
Some of these were caused by wires that were loose or could have been
damaged during installation.  Others had dead current sensors that could
have been defective or were destroyed by replacing the 1 amp fuse with a
larger size before clearing the ground fault. 
0.39% - EEPROMD This failure occurs when the end of day data is
corrupted as it is written from the inverter's RAM into Non-volatile
ROM. There is a simple in field reset procedure that solves the problem
and once solved they do not appear to reoccur.  This cause of this issue
is not well understood but the next version of the inverter firmware
will automatically reset the error.
0.17% - Failed IGBT's/caps. Typically an over voltage failure caused by
incorrect installation.  We have seen 3 units to date that look like
failures due to lightning. 
0.10% Broken cover screws.  All units now ship with greased screws and
the case was reworked to make misalignment during initial assembly less
likely.
0.09% - DOA  usually a power supply issue.  40% of these units had no
problems found and were never powered up.  Probably caused by low PV
voltage. 

The remaining failures (0.3% total) are all below 0.1% and are
essentially random production related events.  K1 open, broken
terminals, melted ac fuse holders, over current failure, noisy, bad temp
sensor, etc. 

In much this failure analysis we did not differentiate if the failure
was production related or likely precipitated by the installation.  The
one notable exception is over voltage failures where the inverter is
essentially destroyed. Another important thing to point out is there is
a wide disparity of the failure rate depending on the installer.  We are
in a newly emerging market and the installer's level of experience with
our inverters clearly has an impact on the in-field failure rates.  Some
wrenches out there have installed many sunny boys with very few problems
others have seen a much higher failure rate on a much smaller sample
size.  Some times the failures come in clusters and are, IMHO, an
unfortunate bad luck of the draw issue.  

To close: Every inverter undergoes numerous quality tests during
manufacture In addition, every inverter is functionally tested on an
automatic burn in rack before it leaves the factory.  We take failures
seriously and take corrective action to remedy the situation as quickly
as possible.  We address problems openly and factually to the best of
our knowledge and will continue to do so.  I encourage all of you to
provide us with feed back so that we can continue to improve our
products as we move forward.  

If you have any questions or if I may be of any further assistance,
please do not hesitate to contact me.

Best Regards,

John Berdner

SMA America, Inc.
12438-C Loma Rica Road
Grass Valley, CA  95945
530.273.4595 - Tel
530.274.7271 - Fax

-----Original Message-----
From: Jeff Oldham [mailto:starpower4u at juno.com] 
Sent: Thursday, April 24, 2003 04:18 PM
To: RE-wrenches at topica.com
Subject: Re: Sunny Boy Stats "final" [RE-wrenches]

Jeff,

I counted every call back, DOA and reship. In other words "failures", it
does not matter how small or trivial to me, if I'm going back to the job
to deal with it and it was not caused by UPS, the client or acts of the
Goddess then it is an equipment failure. I would consider a broken LED a
failure unless it fixed itself! Even if the number is half of what this
group came up with it is still too high in my book and triple that of
the company line. I was really just trying to see if our experience was
unique or not, it turns out it is not.

-jeff

>From the Solar, Wind and Hydro powered office of Jeff Oldham

==^================================================================

- - - -
To send a message: RE-wrenches at topica.com

Archive of previous messages: http://www.topica.com/lists/RE-wrenches/

List rules & etiquette: http://www.mrsharkey.com/wrenches/etiquete.htm

Check out participant bios: www.mrsharkey.com/wrenches/index.html

Hosted by Home Power magazine

Moderator: michael.welch at homepower.com
==^================================================================
This email was sent to: michael_welch at sbcglobal.net

EASY UNSUBSCRIBE click here: http://topica.com/u/?bz8Qcs.bz9JC9.bWljaGFl
Or send an email to: RE-wrenches-unsubscribe at topica.com

TOPICA - Start your own email discussion group. FREE!
http://www.topica.com/partner/tag02/create/index2.html
==^================================================================