Now accurate are domestic electricity meters typically?

[Andy Wade]

Eh, what what is?

[Ian Stirling]

I think Tony means to deteriorate in accuracy up or down.

[richard]

Maybe it is the leccy co's timeswitch that is out and you are using your appliances at night, and getting the cheap rate in the day. Not a likely issue if it's a radio teleswitch. But have you actually checked the meter to see when the cheap rate is kicking in? and do your timers coincide?

[tony sayer]

Accuracy.. over;( or under reading:)

[Dave Liquorice]

And me, we cook by electric (spit) and only use about 23kwh/day total.

If you are not space heating by electicity it might be better to drop the E7. You need to use roughly 20% of your power in the night period to break even on the increased standing charge(*) and day unit rate.

Check that your E7 timeswitch and those used to control your other appliances agree about when the E7 period is. Ours is currently most of the afternoon...

Lots of small loads add up surprisingly quickly. Your computers are using over 4/units per day, the 500W heater on for four hours (just

10min each hour) 2 more, 6hrs of tank light at 250W 1.5 units. 8 units without really trying...

(*) Even if you have a "standing charge free" tarrif, on these the first X units per quarter at Y more expensive, it is odd that X * Y = standing charge...

[Andy Wade]

OIC what you mean: do they read high or low? I don't know really. You'd expect a rotating-disc meter to under-read at low power levels due to the friction in the bearings, but they have some sort of compensation for that, I think, with various tweaky-screws inside the meter. In principle power factor shouldn't make a difference because the accelerating torque on the disc at any moment is proportional to the instantaneous v*i product. If that's so then over a whole number of cycles the disc's movement will be proportional to the net energy flow - IOW the meter should respond to the true mean power. Errors, I guess, could then be in either direction. There are doubtless particular trends for actual practical meter designs, but I wouldn't like to speculate on that. Are there any meter specialists reading?

As to the electronic ones I have _absolutely_ no idea. How do they work? Do they sample and digitise v and i first and then compute energy flow in the digital domain - in which case accuracy will be down to the performance of the CT and VT and the ADCs - or do they use an analogue front-end with a 4-quadrant multiplier and integrator? - in which case the scope for errors would be rather greater...

[Capitol]

How do they

The last one's I worked on were operating AIRI, by using resistive current sampling, with an integrated amplifier and a/d convertor + a direct sample of V which was digitised. All the amplification and sampling took place in the same instrumentation, error corrected IC, with an external resistor for current and an external potential divider network for volts. The accuracy was very good, provided the current connections were sound. IIRC some units were trying out Hall effect current sensors, but the >10e5 current level variations could cause a few problems. The current input was of necessity integrated before amplifying and sampling, in order to cope with the problem of high speed transients from power factor corrected PSU's. The known problems were current resistor and semiconductor drift. The potential dividers were generally trouble free. Todays technology should undoubtedly be better than 15 years ago.

Regards Capitol

[Andy Wade]

Interesting - thanks for that. Having mentioned CT and VT I realised, after posting, that that would be too expensive and that simple resistive coupling could be used, there being no need for isolation from the mains.

Which, presumably, just amounts to the anti-alias filtering you'd need in any A to D system. What sampling rate was used? - thinking about it if you're only trying to measure kWh (as opposed to kVARh) you're not interested in anything above the 50Hz fundamental - so the sampling rate could be very pedestrian...

[Capitol]

Andy Wade wrote: What sampling rate was used?

100KHZ at the time. See some present practice.

Regards Capitol