In article ,
Tim S writes:
Bob Minchin coughed up some electrons that declared:

I can't see the value for logging the data from Current Cost units apart
from the initial technical challenge. With so many domestic loads being
non resistive, and these units not taking a phase reference, they are
bound to be inaccurate. Yes they will give a relative indication to help
track down wasted power but that is about all they are good for. It
would be so easy to have a smarter clamp head that also plugged in to a
mains socket and have the pic calculate true RMS power.
Does anyone know of a unit that does this properly?

I made one about 25 years ago, but it only generates a power (Watts)
output, and not any long-term integration (total energy use).

Might be able to modify one of the plug-in ones to use the sense
coil/clamp. I did try opening one of my plug-in ones to see how
it did the sensing, but even with the case screws removed and the
two pieces of the case slightly separated, something was still
holding it together which seemed like it needed enough force it
would break, to I chickened out of opening it at that point, not
wanting to break it.

This is true - but don't most PSUs at least try to have a power factor of 1?

I don't think the consumer quality ones do, but I haven't bought
one recently. Industrial rack-mount ones do (at least, quality
ones), as it's rather important if you have 1000 of them in a
data centre.

That leaves motors and old style tube lights.

and CFLs, and wall-warts.

I expect the problem was that many people don't have a 13A socket next to
the meter.

However, seeing as the base unit plugs in, if they used an unregulated AC
wallwart, perhaps with some extreme cleverness, they could achieve high
accuracy time synchronisation between the clamp unit and the base unit,
then the clamp could send a current sample set with timestamps and the base
could take a high resolution sample of the voltage profile.

It would be needed at the sensor end.

It would be an interesting exercise...

--
Andrew Gabriel
[email address is not usable -- followup in the newsgroup]