OT: UPS for FreeBSD

[RW]

On Sun, 30 Nov 2014 21:16:13 -0800
Darren Pilgrim wrote:

> On 11/30/2014 1:03 PM, RW wrote:
> > On Sun, 30 Nov 2014 11:31:31 -0800 Darren Pilgrim wrote:
> >> An unloaded, ideal, single-phase FWR produces Vdc = 0.900Vrms.  The
> >> factor is 2sqrt(2)/pi, to be precise.  A half-wave rectifier
> >> produces half that.
> >
> > That's the mean voltage, which is of limited relevance.
> 
> The mean voltage is the output DC voltage.

Defining the DC component of a pure rectified sine wave as the output DC
voltage isn't really useful, unless the design smooths to that value.

> >>   From where did you get your figures?
> >
> > He's referring to a rectifier charging a smoothing capacitor; the
> > voltage will rise to sqrt(2)*Vrms.
> 
> No, the voltage across the capacitor will rise to 0.900*Vrms.  Look
> at it this way:
> 
> The output of the diodes is a time-varying (AC) signal with a DC
> offset. The cap provides a low-impedance path for the AC signal,
> turning the voltage cycle into a current cycle that accumulates a
> charge in the cap. Once the cap is fully charged, it fully negates
> the AC component and you're left with just the DC offset of
> 0.900*Vrms.

When you connect a smoothing capacitor it charges to the
peak rectified voltage; when the rectified voltage move off-peak, the
capacitor can't discharge through the diodes. 

Most linear regulators operate with a smoothing capacitor kept close
to the peak voltage to minimise the headroom needed by the regulator,
which in turn reduces the heat generated. This is the most substantial
reason why a square wave is problematic, it can't have both the correct
RMS and peak voltage.

In practice a computer power-supply would have active circuitry
between the rectifier and the capacitor for power-factor correction. In
that case the voltage on the capacitor is arbitrary.

> > He's added on an unnecessary  factor
> > of two - possibly because he's mixing it up with a bridge-rectifier
> > in a centre-tap configuration.
> 
> 
> A bridge rectifier doesn't use a center-tapped transformer.  You 
> wouldn't see 2*Vpeak in a two-diode rectifier either.  The voltage 
> between the center tap and either diode tap is at most Vpeak.
> 

Bridge rectifiers are commonly used with a grounded centre-tap to
provide positive and negative rails. In that case the smoothed voltage
on the bridge output is ~ 2*sqrt(2)*Vrms, but only because Vrms is
specified per secondary winding.