[rescue] WTB/Advice: UPSs

[Larry Snyder]

Scott Newell <newell at cei.net> wrote:
> At 05:24 PM 4/6/2002 -0500, Larry Snyder wrote:
> >Sharkfin regulators depend on the slope of that sine to work.  Give 'em
> 
> What's a sharkfin regulator?
> 
> 
> >a step function and they'll puke all over themselves and whatever
> >they're powering.  If it conducts for the first 45 deg after zero-
> >crossing and cuts off, the peak it hits is 70% of the sine peak.  With
> >a square wave it's 100%.  That's nearly 50% above the design center
> >value, and definitely bad juju.
> 
> This description sounds (to me) like a 60Hz phase controlled triac circuit,
> such as a light dimmer.

It's similar.  In my former life we used SMPS in some of our test
equipment.  In the then-newer stuff (back 16 years ago) we used a
sharkfin as a preregulator, taking feedback from the raw output of the
switcher, to provide a reasonable range of input to the
linear regulators that followed.

Even if you're not doing a preregulator, the voltage across the HV
DC rails (once loaded) is the rms of the input.  Square waves are
going to do the same thing to the rms that they do with a sharkfin.
The DC voltage is well above what the design expects.  Some supplies
may put up with it, some may not.

> 
> 
> I still don't understand why a square wave input would be bad for most
> SMPS.  Don't they usually rectify and filter the incoming AC to generate an
> internal high voltage rail, then do all the switch mode goodness to step
> down, regulate, and isolate?  (In which case a square wave or DC input
> would actually be better than a sine wave!)  Is it the PFC circuit that
> causes a problem with non-sine inputs?  (Most of the circuits I've studied
> are older, fairly simple, and omitted the PFC circuitry.)
> 
> 
> newell
> _______________________________________________
> rescue list - http://www.sunhelp.org/mailman/listinfo/rescue