[Nut-upsdev] Voltage transfer points.

[Peter Selinger]

Kjell Claesson wrote:
> 
> s=F6n 2006-02-26 klockan 18:19 -0500 skrev Charles Lepple:
> > On 2/26/06, Kjell Claesson <kjell.claesson at telia.com> wrote:
> > > As I find it, there is only one point where the boost or buck (trim)
> > > kicks in.
> >=20
> > Right, but the input voltage at which it exits the boost or trim mode
> > might be slightly different than where it enters that mode.
> >=20
> Yes you are right about that. But this is due to the fact that if it=20
> would use the same voltage, it could start oscillating setting buck
> (boost) on and off if the voltage is on the transfer point.
> 
> It is not so difficult to make the on and off voltage to be the same
> by electronic. You don't get the same hysteresis as you do on a room=20
> termostat.

This discussion is purely hypothetical, since in fact, none of our
actual, existing drivers or supported devices have different settable
transfer points for entering and exiting boost or trim mode of the
kind that Charles is suggesting. This is a purely hypothetical
functionality that we do not need to support. If such a mechanism
indeed existed to prevent oscillation, for sure it would not be user
configurable.

As a look at the source shows, the few drivers that implement
input.transfer.boost.low or input.transfer.trim.high do in fact treat
them as synonyms for input.transfer.low and input.transfer.high.  This
is confusing and redundant; I am all in favor of eliminating
input.transfer.boost.low and input.transfer.trim.high. It would then
also make sense to rename input.transfer.boost.high and
input.transfer.trim.low as input.transfer.boost and
input.transfer.trim, respectively.

-- Peter