[TowerTalk] OT: Generators and UPS's

WA3GIN @ Arlington County, VA wa3gin at erols.com
Fri Aug 25 14:42:48 EDT 2006


A common characteristic of generators is the normal output voltage 
distortion when supplying power to nonlinear loads such as computers. 
This output voltage distortion can be interpreted by the UPS as 
unacceptable power quality, forcing the UPS to transfer to battery 
operation. When the load is transferred to the battery, naturally the 
generator distortion will be reduced or disappear, leading the UPS to 
attempt to transfer back to line operation. When the load is reapplied 
to the generator, the distortion will return, leading the UPS to once 
again transfer to battery. This cycle may repeat indefinitely at 
intervals of approximately 4 seconds. The answer in this case is to 
choose a generator which will not distort, when the nonlinear computer 
load is applied. In general, the generator should be 3-5x the size of 
the total attached load.

You can purchase a UPS that doesn't care about the noise or attempt to 
filter the noise using MOVs or other technology or up grade the 
generator but there is no painless solution.

73,
dave
wa3gin

Jim Lux wrote:

> At 08:03 AM 8/25/2006, WA3GIN @ Arlington County, VA wrote:
>
>> Perhaps when not receiving a true sine wave from the generator you
>> should reduce the load on the UPS to 50%
>
>
>
> A standard (non-inverter style) puts out a very nice sinewave, albeit 
> of varying frequency as the engine speed changes.  I have seen 
> generators that put out bad waveforms in weird load situations (enough 
> magnetic flux to saturate the iron, for instance, or if the current 
> waveform is very peaky), but by and large, the physics of how 
> generators work (spinning magnet inside a coil) inherently generates a 
> clean sine wave.
>
> I'd look to load characteristics, if it's not a governor control law 
> issue (voltage and/or frequency stability).  In particular, high 
> transient currents can overload the generator in an instantaneous 
> sense, without noticeably affecting the average power output or 
> voltage.  You get a funky sinewave with a little dip in the top.
>
>
>



More information about the TowerTalk mailing list