On Wed, Aug 20, 2003 at 11:19:31PM -0600, Paul E Condon wrote: > On Thu, Aug 21, 2003 at 04:11:02AM +0100, Pigeon wrote: > > On Wed, Aug 20, 2003 at 12:33:12PM -0600, Paul E Condon wrote: > > > > > > > Also, Radio Shack sells a "current transformer" which is an add-on device > > > for their digital multimeter. With the multimeter and the current transformer > > > you can measure current and voltage on the AC supply input to the computer. > > > Or to any other electric device about which you are curious. > > > > ...but this arrangement will only give you a useful reading if the > > current waveform is sinusoidal. The current drawn by a PC is most > > definitely not sinusoidal; the rectifiers in the PSU only conduct on > > the peaks of the voltage waveform, so the current waveform looks like > > a series of narrow spikes of alternating polarity. The > > common-or-garden multimeter does not perform a true RMS conversion, > > and won't give a sensible result on this sort of waveform. > > > > (Also, you have to separate the live and neutral conductors in the > > power lead so the current transformer can be fitted round one > > conductor only, which is a pain.) > > > The power consumption values from the current transformer method are > very similar to those from watt meters such as Kill a Watt. To me this > indicates that either both are correct, or both are incorrect, possibly > because of the impulsive current wave form.
Interesting... Cynically, I would guess that both devices are using the same kind of averaging-calibrated-to-read-RMS-with-sinewaves method and so both give similar errors... OTOH, maybe you have a good enough multimeter to do true RMS conversion, and if I was designing a 'Kill a Watt' type device I'd use a PIC microcontroller to sample the voltage and current waveforms and do RMS conversion in software (cf. your next para), which would keep the hardware very simple. So maybe both are correct... > The right way, I think, is > to examine both current and voltage wave forms with an oscilloscope and > compute the time average of the product of the two wave forms. And be > careful to handle the phase relation between to two correctly! Agreed... and it wouldn't be too hard to knock up an isolating interface to feed the voltage and current waveforms into a sound card, and do this all in software. > (Also, the current transformer includes a gizmo for separating the > hot and neutral wires safely.) Oh, very good. Radio Shack used to trade as Tandy in the UK and were generally overpriced (especially for components) and of indifferent quality. I was expecting them to leave their customers to either rip their mains leads apart or to measure ground loop currents in blissful ignorance... :-) -- Pigeon Be kind to pigeons Get my GPG key here: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x21C61F7F
pgp00000.pgp
Description: PGP signature
