Index Home About Blog
From: John De Armond
Subject: Re: Inverter Failures
Date: Sun, 24 Jun 2007 11:18:03 -0400
Message-ID: <>

Nobody seems to have read closely enough to answer your question so let me take a

No experience with that particular inverter but lots of experience designing and
servicing UPS inverters - same thing, pretty much.

The answer is "it depends". Mainly on how heavily loaded the inverter is and how hot
the components (NOT the outside case, the actual components) get.

Heat is the mortal enemy of electronics, especially electrolytic capacitors and the
power semiconductors.  Electrolytics can last essentially forever at near room
temperature but only a couple of years if run at 90 deg C (assuming 105 degC rated
caps).  The DC input filter caps in particular, handle heavy high frequency current
and so generate heat of their own.  These should have a free flow of ambient air and
be shielded from radiant heat from other components in the box.

The power semiconductors are second most sensitive to heat.  The case temperature
should be only a few degrees above ambient, no more than perhaps 20 degC.  Chemical
reaction rates (including things like whisker growth) double for every 10 degC rise
in temperature so keeping things cool is vitally important.

There is NO REASON for fixed installation devices to be designed to run any hotter
but many designers do it to save a few pennies on heat sinks.

The first thing I'd do is take some measurements.  Measure the case temperature and
the exhaust air if it has vents, while the unit is under full load.  If the rise is
much over 20 degC over ambient, consider forced air cooling.  Only a little is needed
- a 24 volt muffin fan run on 12 volts uses practically no power but does an
excellent job of cooling.

Next, if you can get to them without significant disturbance (I'd be reluctant to
disassemble a unit that old, as internal plastic could already be getting brittle,
etc.), measure the temperature of the input filter capacitors (likely the largest
components in there other than the transformers).  Again, use that 20 degC magic
rule.  Finally, measure the power semiconductor (FETs, diodes) CASE temperature.

My favorite measuring instrument is a fairly high end infrared pyrometer.  Not one of
the $100 cheapies that you can buy just about everywhere.  Mine has adjustable
emissivity and a tightly defined measuring spot size.  I can zoom in on an individual

If you can't lay your hands on one of those, the second best is a very fine
thermocouple (26-30 gauge) with a blob of heatsink compound on the junction.  This
tiny couple can be reached in through air vents and laid atop the component of
interest.  The heatsink compound thermally couples it to the component.  Scads of
under $50 digital voltmeters are available with type K thermocouple inputs.  Get the
thermocouple from

You have to be a little careful with the thermocouple because some of the components
of interest have voltage on them.  Insulate the wire down to the junction itself with
several coats of fingernail polish or similar varnish.  The voltage won't affect the
temperature readout.  The problem is when you brush the couple leads up against the
grounded case.

One last thing, go over again your lightning protection.  One trick that I learned
from years of maintaining mountain-top 2-way radios is to coil every lead that goes
into or out of the device into a couple of loops, maybe a foot in diameter.  Upstream
of the loop (toward the lightning), provide a very close arc gap for the lightning to
divert through to ground.

My standard "gap" was to take a little nick out of the wire's insulation.  Tape a
piece of solid conductor wire parallel to the wire so that it passes over the nick.
Solidly ground the solid wire.  The arc gap is the air space the thickness of the
insulation.  Optionally, take your knife and raise some barbs on the solid copper
wire where it sits over the gap. These points make the breakdown voltage of the gap
much less.

The way this works is this.  Lightning surges are made up of very high frequency
components (do a Fourier transform on a pulse with a rise time of millions of volts
per microsecond!)  Those couple of coils of wire form a choke that presents a high
impedance to the incoming pulse.  At the same time, the arc gap presents a very low
impedance path to ground.  The lightning will take the easiest path and that is
through the gap.

This was an accidental discovery.  It is usual practice to coil excess feedline at
the base of the tower.  I started noticing burn marks on the feedline and tower right
above the coils.  After I saw several of these, it dawned on me what was going on.
Since then I don't think I've lost a single radio to lightning damage.  I coil the
feedline on the tower and again where it comes through the building wall.  All power
and control wiring gets coiled inside the building and gaps as described above
installed.  It is common practice to run a ground bus around the inside of a
transmitter building.  Usually 3-4" wide x 1/8th thick copper bus bar, grounded every
few feet to ground rods driven into the ground or if the building was properly
planned, the copper mesh embedded under the building.  I arrange the arc gap
placement so that there is no more than an inch or two of wire between the gap and
the ground bus.

Here's another suggestion for preparing for the eventual failure.  Dedicate an
inexpensive generator to the well pump by itself.  Wire it up so that the pressure
switch operates the idle control on the generator and through a time delay relay, a
contactor that energizes the pump.  Adjust the idle or economy setting on the
generator to take it to as slow an idle as possible for maximum standby fuel economy.

The operating sequence is this: The pressure switch calls for water.  The generator
is throttled up to full speed.  A few seconds later the contactor is closed by the
time delay relay and the pump is powered.  When the tank is filled, the pressure
switch opens, the pump turns off and the engine returns to idle.

If you group your water consuming activities reasonably close together you can simply
start the generator and go about your business without using too much fuel.  This is
easier on the generator and probably more fuel-efficient than starting and stopping
it each time.

I have my house set up like that.  I can run the pump either on the whole-house
generator when I'm doing other things such as cooking or running the AC or I can run
it on its dedicated small generator when nothing else needs power.  All my lighting
is powered from a UPS with a large battery bank so I can go a day or two between

Up here in the mountains we can have snow outages that can last 2 weeks so I have to
be prepared to live essentially as if I were off-grid.

I've solved the refrigeration problem rather cleverly, I think.  I have two chest
freezers and an upright refrigerator.  All are on wheels.  I simply roll them outside
on the porch.  When the temperature is in the single digits everything remains nicely
frozen.  I'll occasionally put a container of hot water in the 'fridge to keep it
from freezing.

One last thought.  Now would be a good time to start looking for a spare inverter to
keep on the shelf.  Now that you have the luxury of time to look, maybe you can find
a used one somewhere on the cheap.


On Sun, 24 Jun 2007 06:40:32 -0400, Ron Rosenfeld <> wrote:

>My system, which includes a pair of series-stacked SW5548's, has been
>running without difficulty for 6 1/2 years.  But I find myself concerned
>about dealing with inverter failure(s).
>Does anyone here have data on how long these things might last?
>Strategies to deal with failures in an off-grid system?
>I could certainly run off the backup generator, but how long does it take
>to get an SW5548 repaired or replaced?  I live in Downeast Maine.
>The only critical 240V item I have is the well pump.

Index Home About Blog