From: John De Armond
Subject: Re: Cheap inverters
Date: Thu, 20 Feb 2003 17:10:21 -0500
On Wed, 19 Feb 2003 09:41:37 -0700, Alan Balmer <firstname.lastname@example.org> wrote:
>On Tue, 18 Feb 2003 05:40:13 GMT, John S <email@example.com>
>>In article <firstname.lastname@example.org>, email@example.com
>>> The so-called car adapters contain a small inverter/converter designed
>>> for the light load of a computer. The dc input is inverted to a ac
>>> voltage which is then (via a small transformer - transformers do NOT
>>> work on dc) boosted up to a higher ac voltage and then converted to
>>> the higher dc voltage the 'puter requires - usually around 18v.
>>> Will Sill
>>Will - I have suggested before that if you want to fake being an
>>engineer, then stay out of technical conversations. Your problem is
>>that you are too clueless to know that you are clueless so when you
>>open your mouth like this, you drop the ball.
The real problem isn't that he tries to play being an engineer. It's that he
plays a really BAD one.
>Will's description is not incorrect, though it may be out of date.
>DC-DC converters have been built this way for years, and can be quite
>efficient. Because the oscillator is typically fairly high frequency,
>a low-current transformer can be very small.
He is completely wrong. Inverters have never operated in the manner he
describes. The very first inverters used radio vibrators to mechanically
switch DC alternately into one winding and then the other of a specially
designed transformer. This transformer was designed not to saturate with the
large DC component and not heat too much from the odd harmonics that accompany
the square wave switching.
A bit later, inverters used simple blocking oscillator designs in which an aux
winding supplied the correct phase signal to the transistor that switched the
DC on and off. This was a unipolar design with a DC current in the
transformer of about half the total current. The very early Tripp-Lites used
this very inefficient design.
A bit later, the design evolved into bi-polar switching where two transistors
driven by two separate "tickler" windings switched the 12vdc alternately.
This greatly reduced but did not limit the DC bias. The frequency was
determined by the transformer characteristics and the load.
Still later, the design acquired a simple 555 IC timer-based 60 hz timebase to
stabilize the frequency and to further reduce the DC flowing in the
transformer winding. This is the last style that was popular before the
advent of the switchmode inverters. The 2KW UPS I built for dixie.com's
little data center used one of these inverters from Tripp-lite. It weighed
over 100 lbs.
In all of these, low voltage, high current DC was switched periodically into
the primary of a very heavy iron cored transformer whose secondary produced
the nom 120 VAC square wave.
>As you point out, modern low-power converters generally use IC-based
>Buck conversion. I don't know about other aspects, but price is an
Modern inverters have glomed onto developments in switchmode power supplies
such as are in the PC. The architecture is quite similar. The development of
high power FETs and small but high power ferrite core transformer designs are
the enabling technologies.
The architecture is thus. The incoming 12 volts is chopped at a high
frequency - 20khz to >1mhz - and applied to a ferrite core step-up
transformer. The output is rectified and filtered and results in about 145
volts DC. In many designs, especially if it is desired for the neutral to be
ground referenced, there are two of these inverters, one for the positive
swing and one for the negative. This high DC voltage is switched at a 60 hz
rate to produce the output. In most low end inverters, the switching is
simply on-off and produces a square wave. There is a zero crossing dead time
so that the wave more resembles a sine wave. An output low pass filter will
further improve the waveform. Some slightly more expensive inverters will
switch two different voltages, one after the other, to produce a multi-step
waveform that is even closer to a sine. With some low pass filtering, the THD
is typically <5%.
The high end units such as the true sine wave inverters and the inverters
built into variable speed generators like the Honda EU series produce the sine
wave by synthesizing it using a pulse width modulation scheme to switch the
high DC voltage. The switching rate is typically 20khz or greater. Since
each half cycle of the output is made up of hundreds of modulated pulses that
closely approximate a sine and because there is invariably some low pass
filtering on the output, the THD is very low. Honda claims less than 1% and
from I've seen on my instruments, I'd believe it.
From: John De Armond
Subject: Re: Costco Inverters [and campground power]
Date: Sat, 27 Sep 2003 17:43:35 -0400
Jim, I've been hearing this rumor about printers at least since the original
HP laserjet, what? maybe 1985 or so? Back then the issue was with UPS's, as
the cheap inverters didn't exist. Here's my actual experience:
In about '85 I built an uninterruptable power supply for my computer lab, not
liking the prices of commercially built units. It consisted of a 14 volt, 200
amp linear regulated power supply, a large bank of marine batteries and a
Tripp-lite 1000 watt square wave inverter. Not pseudo sine or modified sine.
Plain old square wave. With one exception everything I tried ran just fine on
this inverter. That included the computers (an original IBM AT for a file
server, monitors, modems (I expected trouble there but didn't have any) and an
original HP laserjet and then a Laserjet II.
The only thing that I tried that didn't work was a microwave oven. The reason
is the voltage doubler that drives the magnetron depends on the line voltage's
peak voltage (about 170 volts.) The square wave inverter output a square wave
with a peak voltage of 120 volts.
The original pseudo-sine inverters used two different high voltages, switched
in sequence to partially synthesize the wave. It would usually involve an
approx 120 volt signal and a higher voltage one. The sequential switching
generated a stepped waveform. Things like microwave ovens that require the
high peak voltage might or might not work, depending on what the high voltage
The current generation of super-cheap inverters use a single, typically 145
volt DC source that is switched as a square wave but with some zero crossing
delay. That is, the positive voltage is turned off and then some time elapses
before the negative voltage is turned on. About the only things that don't
work on these are the transformerless battery chargers such as many cordless
drills use. some audio equipment will buzz a bit but most don't.
On Sat, 27 Sep 2003 12:06:06 GMT, "Jim Walker"
>I agree that the warning to use true sine wave inverters is overdone. I
>have used the cheap one for years on a laptop mainly because I didn't hear
>the warning before I began. I now know that some laptops have built in
>protection that takes care of the problem. My Thinkpad is one of those with
>the protection. A thread a while back went into this and a couple of
>computers did not work well with the non true sine wave inverters. I am
>not sure of the brands now, sorry. A Life On Wheels seminar went into this
>and they concluded that some computer printers would not work with non true
>sine wave inverters. This is practical experience from some full timers.
>Again, it could be only some printers and not all. My conclusion is that
>the appliances are generally protecting themselves. I will not go for the
>true sine wave inverter unless I absolutely have to.
From: John De Armond
Subject: Mendelson's surplus 1000 watt inverter
Date: Tue, 11 Nov 2003 23:28:56 -0500
I mentioned last week that Mendelson's had a decent price on 1000 watt
inverters and that I'd ordered a couple. They came in. Someone asked me
about the waveform. I've done some testing. The results are as follows.
The inverter is smaller than the Vector unit. One of the smaller 1000 watt
units I've seen. My first reaction upon removing the cover was "holy sh*t,
where are all the parts!?!" I saw the usual 4 high frequency transformers, a
few transistors along each side of the board and some filter capacitors but
nothing else. Then I noticed a small daughter board. Upon casual examination
it appeared to have only one IC. Then I looked underneath. A high density
surface mount board. Ahhhh, that's where all the works are. It looks like
whatever chicom engineer designed this thing intended for one daughterboard to
serve a whole line of inverters.
On the main board of one inverter there were signs of the blue smoke leaking
out. On the other, none. Both have had power transistors changed out. The
original used some sort of polymer heat sink sheet but the repair tech used
thermal grease. Obvious which ones had been reworked.
Photos for the following discussion are here:
To evaluate the unit I hooked it via short jumpers to a fairly large AGM
battery* I had on hand. I loaded the inverter with a 600 watt ceramic heater.
I used a Fluke 97 ScopeMeter to observe the output.
Upon powering the unit, the fans run for a few seconds and turn off. The unit
is totally silent. The photo of the no-load waveform shows the usual
stair-step square wave with dead time at the zero crossing. Same as all the
others I've looked at. When the 600 watt load is applied, the fans start
within seconds. The photo of the waveform shows a bit of sag at the peak of
the square wave but not enough to affect the RMS value very much.
The rise time of the square wave provides a good indication of the potential
for buzz in audio equipment and magnetic devices such as fan motors. The
photos show first the no-load rise time and then the 600 watt load rise time.
The 1.2 uS rise is fairly fast but is also fairly typical of the breed. The
rise time stretches out to a leisurely 3.5 microseconds with a 600 watt load.
This is slow enough not to buzz most devices.
Of most concern to me is the fairly low RMS voltage, 112 volts, and the fairly
low frequency, 55 hz. Of the two I'd worry more about the lower voltage
affecting voltage sensitive devices like microwave ovens. The low frequency
will cause fans and line-operated digital clocks to operate more slowly than
I connected the second inverter and got 120 volts and 61 hz. There are
several pots on the daughterboard so perhaps this variability can be adjusted
One of the things I like about these inverters is that it looks fairly easy to
parallel them, with all the logic being on that daughterboard. Another thing
I like is that the DC power is distributed to the 2 sets of 2 internal
inverters that run in parallel via heavy bent copper bus bars. The Vector,
along with every other inverter I've seen in this class rely on a combination
of circuit board traces and stamped sheet metal bus bars. This is probably
the core cause of the Vector's significant sag in output voltage vs load.
These numbers seem bad but compared to other inverters I've looked at, they're
not. One inverter that I'd used for years without problems yielded 50 hz when
measured! The waveform stayed clean when subjected to an inductive load (fan)
which is good. Another inverter that I've used quite successfully on motor
loads showed a drop out (double peak) on each half cycle on the scope. As my
old Tripp-Lite square wave (not pseudo or rounded but bone-jarring square
wave) inverter demonstrated, most devices will run on almost anything that
resembles an approximately 60 hz bipolar wave.
The Costco inverter is probably a better deal at $100 (anyone near a Costco
care to buy one for me if I send you the money?) but for those of us far away
from Costco stores, this Mendelson's unit is a pretty good deal. I like mine
enough that I'm going to remove the Vector unit from my MH and install this
one instead. with the smaller voltage sag, it should provide a few more volts
to the microwave oven.
*New, code date 2002, 80 amp-hour AGM battery that I purchased at the
Lawrenceville, Ga hamfest for $25. Actually I bought several. Typical of
what one can find at a hamfest. I bought a similar set last year to go on my
From: John De Armond
Subject: CostCo Inverter first impressions
Date: Sat, 22 Nov 2003 01:35:54 -0500
Peter Crowl was kind enough to answer my request to buy and ship me a CostCo
1000 watt inverter for evaluation. Thanks again, Peter.
I will not have the time before Thanksgiving to wring this thing out but what
I saw in my initial observations are important enough to mention now. Below
is my last note to Peter I sent a little while ago.
Executive Summary: It appears that you get what you pay for and that this
Xantrex inverter isn't very good. More info after Thanksgiving.
On Sat, 22 Nov 2003 00:18:51 -0500, "firstname.lastname@example.org"
>> My first impression is that
>>Xantrex made a special version for costco with every penny of cost squeezed
>>out. For instance, they soldered the main fuses to the PCB instead of
>>spending a couple of pennys on spade lug sockets.
> Wouldn't be the first time that was done. This business about the fuses
>being soldered doed not impress me - and sure won't come replacement time!
> I look forward to the review on RORT!
My words were prophetic. I hooked the thing up and it was dead as a doornail.
those *(^&^ soldered-in fuses were blown. 3 40 amp ATC fuses. I quickly
realized that I'd have to remove more than a dozen screws holding the
transistor clamps in place to get to the bottom of the board to unsolder those
fuses. I cheated and bridged the gaps with solder after prying the fuse
bodies off. The inverter then worked and appeared undamaged. I bet someone
hooked it up backwards, blew the fuses and then 'returned' it for a refund.
Anyway, I'm not terribly impressed with this unit. It is of the design that
floats both the hot and neutral above earth ground. In this case about 55
volts. Some surge arresters and switching power supplies will behave badly.
One of my first tests is to plug in a "100 watt equiv" Lights of America CF
lamp just to see if everything is working. On this inverter the lamp strobes
badly, probably indicating the output is not symmetrical. This is the first
time I've encountered strobing so I'm going to have to investigate further
One of the important things that has been de-contented is input filter
capacitors. These prevent ripple from backing out the DC side by supplying
the surge current to the high voltage inverter on each half cycle. My Vector
has perhaps a dozen large electrolytics scattered around the board. The
Mendelson's surplus unit has probably half a dozen larger ones. This one has
2. The result is that with a 600 watt load on the inverter, there is almost
half a volt of 60 hz ripple on the AGM battery terminals. I'm going to
postulate that this much ripple would cause 12vdc fluorescents to strobe and
would cause hum on audio appliances. Not good at all.
This is all the result of perhaps 30 minutes of poking around with the scope
and voltmeter so consider my comments preliminary. I'm thinking that this is
a case of one getting what he pays for.
I have enough different inverters in hand now that I think I'm going to do a
formal comparison test after Thanksgiving. IT'll be a lot of work but I think
worth it. I wish I had one of those inverters that Flying J sells under the
Shakespeare brand but I'm not willing to pay the inflated price for one. I
think that what I have will be sufficient.
From: John De Armond
Subject: Re: 12-volt Fluorescent Lighting
Date: Wed, 25 Feb 2004 01:25:48 -0500
You are quite simply wrong, Ralph. I have tested numerous modern inverters -
and I bet I know more about testing them than you do - in recent times and
I've yet to find one that is as bad as 90%. The efficiency is almost
independent of load when the load is >10% or so.
The reasons are many. The modern architecture. The modern CMOS components.
Modern design techniques.
Let's take a look at the Vector 1kw inverter that I currently have on my
bench, testing for long term durability. The architecture is thus:
(12v to 180vdc inverter) -> (high voltage storage) - > (power FET 60 hz driven
choppers [several parallel stages]) -> LC high pass filter -> 120VAC outlet.
First the inverter. Where the heavy lifting is done. It uses multiple
parallel power MOS-FETs to reduce the on resistance to minimize loss. It also
uses more FETs as synchronous rectifiers. SRs avoid the 0.2 to 1 volt drop
see across conventional rectifiers. The switching frequency is high - 40khz -
to further minimize losses and reduce the size of the components.
The inverter's overall efficiency is remarkably improved at less than full
load with a simple feature - the DC/DC inverter does not run all the time. It
is switched on and off, with fairly high hysteresis, to maintain a setpoint
voltage on the energy storage caps. All easy to do, given processors that cost
35 cents a pop in quantity.
The high voltage storage consists of multiple aluminum electrolytic caps.
This is actually a weak point with this particular design. The caps
apparently have a fairly high ESR and/or dissipation factor, as they get very
warm during full load operation.
The output stage is again several power FETs in parallel for each polarity of
the 60 hz output. The standard stair-step waveform is used to synthesize to
some degree, a sine wave. The "sustain" vs "peak" duration ratio is varied
according to load to keep the RMS value near 120 volts and to maximize
Because of the DC/DC inverter's switching on and off at a fairly slow rate,
probably 2-5 hz, measuring the efficiency of one of these critters is a bit
more complicated. Driving the inverter from a DC power supply or
battery/charger combo requires data logging and averaging to determine the
true input power. I used that method in the beginning, using my National
After I acquired a 7 farad, 15 volt capacitor, the testing got much simpler.
The cap is large enough to smooth the input draw, at least for 1000 watt class
inverters. Simple DC instruments can then serve.
On the output I use only bench grade (5.5 and 6.5 digit Fluke and Keithley)
meters with data logging (IEEE-488) and lab quality F.W. Bell transducers.
The results agree quite favorably with old fashioned analog (iron vane or
electrodynamic) lab instruments that I own. With datalogging, I don't have to
stand there and write down the numbers. The results also agree very well with
that of an old-fashioned rotating disc power meter.
Without taking a walk to my lab to fetch my notebook, something I won't do for
a pissin' contest like this, I can recall some interesting numbers. The
efficiency remains >90% until the load drops to <20 watts. With no load
applied, the inverter draws an AVERAGE of about 15 ma. The Vector spec is 30
ma, if I recall correctly. That it does, but only when the DC/DC inverter is
topping off the energy storage caps.
Another remarkable little inverter is the Vector 70 watt unit built into an
oversized cigarette lighter plug. Here:
I have probably half a dozen of these little inverters. When I first ran one
through my lab I wasn't terribly impressed. Full load efficiency <80%. Then
I realized why. The crappy cigarette lighter plug was getting quite warm,
indicating a good deal of wasted energy. I opened the unit up and soldered
wires directly to the inverter board. The efficiency was then what I
expected, >95%. So little heat produced that the case barely gets warm.
Back to the original question, my suggestion would be similar to "Q". Use
compact fluorescent lamps where practical and conventional fluorescents,
preferably with electronic ballasts, elsewhere. Instead of one inverter, I'd
use numerous of those little vector inverters, one on each switched circuit.
That way there is zero no-load loss.
Awhile back I bought one of these Thin-lights to evaluate:
I'll have to say that I was impressed. The inverter design is extremely
clever and appears to be of conservative design. I had a couple of problems
that kept me from buying any more. One is the cost. I just can't see $40 for
a 20 watt fixture (they claim 30 watts but that is incorrect.) Two is the
color temperature of the light. They only come in cool white which has a blue
cast and to me, at least, makes the RV interior feel like a hospital room.
Warm white, or even better, incandescent white (about 2000degK color
temperature) tubes are available but not from Thin-light.
One of the main reasons CFs have been so well accepted is that the phosphor
used duplicates the color spectrum of incandescent lights. The light is
familiar and warm. People give no thought that they're now using a
"fluorescent lamp". CFs work equally well in an RV where there is space to use
a conventional fixture.
The cost advantage is large. That little 70 watt inverter is $20 from
Northern. I've seen them for $16 and change at WallyWorld. It can easily
drive 4 15 watt CFs. That would be $4 per lamp. I can get CF lights at Sam's
for $19 for 5 of them. That's about $4 each. That would be $8 per lamp
total. Plus the cost of a fixture if necessary.
On Mon, 23 Feb 2004 17:11:42 -0800, Ralph Lindberg <email@example.com> wrote:
>In article <firstname.lastname@example.org>,
> Alan Balmer <email@example.com> wrote:
>> Nevertheless, the very inexpensive Vector VEC024 (400W continuous)
>> inverter sold by Sam's Club is speced at 90% efficiency. This is
>> probably at maximum load. No-load draw is 0.4A.
>While that is the spec, have you measured one?
>The last time we did, it wasn't nearly as impressive. If you drop the
>load off 100%, the efficiency drops way off. Let's say you are loading
>it at 20W (typical lighting), the efficiency is typically less then 50%
From: John De Armond
Subject: Re: Inexpensive 24 or 48 volt inverters
Date: Sat, 09 Aug 2008 21:33:28 -0400
On 9 Aug 2008 23:41:59 GMT, firstname.lastname@example.org wrote:
>| for 48 to 120, there are telecom inverters. Lucent and Major Power make
>| them, and they're very expensive but available in that power range.
>And they expensive because?
>1. 48 volts is harder to deal with than 12 volts.
Actually 48 volts is EASIER to deal with at a given wattage than 12 volts. For
power FETS, what almost every inverter uses, the conduction losses are
proportional to I^2R. A 48 volt inverter draws but a quarter the current of a
12 volt inverter of the same size. All else being equal, that equals 1/4 the
number of FETs and 1/16th the thermal losses. It's rare to find a power FET
with a voltage rating lower than 60 volts anymore so the same device will
probably be used at both 12, 24 and 48 volts. The lower voltage simply
requires more of 'em.
The economy of scale of 12 volt inverter manufacturer is more than offset by
the reduction in power devices and the reduced heat that has to be disposed
of. That's why, with the help of some folks who actually know what they're
talking about, I've been able to find high voltage inverters in the same price
range as 12 volt one.