floating stuff on neutral

It's done. My little Hakko 808 desoldering gun runs its heater and controller right off the mains. The heart of the electronics is an A2804

Made it rather annoying to troubleshoot when the controller went on the fritz.

I did a power meter some time ago and it was all neutral referred. At that time it complied with the relevant international standards.

Depending on the application and market there may be additional constraints on performance. If the meters are used for charging customers they have specific performance requirements. Mine were used for sub metering in shopping centres.

Most of those meters work that way. BTW you could save yourself the trouble of designing it and drop me an e-mail :-)

I think IEC Class II provisions would apply here.

"John Larkin" kirjoitti viestissä: snipped-for-privacy@4ax.com...

Why not? Many devices (coffee makers, motion detectors etc.) I have disassembled run from capacitor divider connected directly to 230 Vac. All buttons/potmeter shafts/lcd displays are of course properly insulated. (We have unpolarised plugs here)

-ek

I am confused, maybe missed something. I thought the "Kill a Watt" measured real power. (doesn't it?). This mod wires directly into the voltage and current sensor (bypassing the Kill a Watt logic) and as far as I can tell, nothing in the code to correct for Power Factor. Am I missing something?

Kill-A-Watts measure "instantaneous" current and voltage and calculate everything from there. Power Factor is calculated the same way W and VA are, if that's what you mean by "code to correct for". Yes, Kill-A-Watts "measure" real power, in the sense that they do report real power. I haven't seen a 240V Kill-A-Watt, though. I know someone who made one by rewiring a unit to put the current sensor on the high leg and then doubling the results.

There's not a lot to design these days. A voltage divider scales down the line voltage and a neutral-side shunt picks up the current. Digitize and do the math. Adding a few lsb's of noise onto the current signal zaps DNL and quantization errors.

Besides, I like to design things.

John

It is done. You just have to make sure it's all legit so you can pass UL. X- and Y-caps (if needed) with the correct certs, fuse, clearance/iso for LCD, switches and the like. Then things like whether or not water or condensation can get inside, or what happens to the smoke that the neutral shunt may let off upon a hard short and the subsequent *PHOOMP". Sometimes it's better to use a current transformer, depends on the cost constraints.

You couldn't be more wrong. There really is a lot to design into these sort of gadgets to get people to actually buy them for a profitable price. The measurement hardware and software is the least interesting part of the design.

I'm just spit ballin' here, but I take it this is a standard US home 2 hot 1 neutral power panel. If so, you will need to measure the current in the hots, not the neutral. A "220" device doesn't use the neutral, or do I not fully understand what you are doing?

One of the reasons I don't design electric meters any more (the present exception is a charity case... I'm charging $1) is that there's not much money in it, and 60 Hz is boring. I can design a really cool VME module that gets connected to a stadium-sized laser, or a jet engine, and charge $9K for it.

I did design the Synergistics C180 end-use survey meter, which was profitable.

It did 16 channels of power metering, plus temperatures, pulse counts, flow, BTUs, stuff like that, and stored time-series records. Thousands of these were sold before the utilities lost interest in end-use load studies and flooded the market with used gear.

I did design a residential meter that was to be manufactured in India by a subsidiary of Niagra Mohawk. The BOM cost target was absurd. That one never got beyond the demo stage.

I also did 1720 apartment submeter units for Battery Park City. Again, there's not a lot of money in submetering.

So it's not as if I'm inexperienced in electric metering. If anything, I've done too much of it. It's like fire and security systems: everybody wants the cheapest gear so the margins are tiny.

What sorts of meters have you done?

John

The current application is single-phase, so a neutral-side shunt is the cheapest way to go. More phases would require isolated current sensors, CTs maybe.

A shunt plus a signal-level transformer is interesting. Much less copper and iron, better linearity. Utility-grade metering turns out to need a pretty good current transformer.

John

My brain was thinking whole house metering or some such. Not a per device meter.

Jim

Not much but I'm not alone on this project although much of the measuring part comes from me. I can't get into too much details (I'd like to keep my job) but the project I'm working on involves a home control, tele-care, communication, multi-media ecosystem in which equipment submetering is just a small part. Because of that we can make others give the central control unit away for free and sell the metering / home control units as a cheap add-on. And this is not some sort of sales department pipe-dream. People are buying this stuff from us. Actually our units are not very price competitive but the added value makes people buy them.

Careful if this is for actual revenue metering and in a developing country where they don't have much in terms of GFCI. Some smart kids will find out a trick: Connect neutral to PE instead and your meter will always read zero.

Much better. When using a regular small line transformer with all the ratings you can hang that onto the hot side. Just make sure there's protection in place in case the shunt goes *PHUT*.

The never-manufactured India meter I designed had a few anti-cheat provisions. Stealing power seems to be the national sport of India. I was told that all ex-military folks think they have a lifetime right to free electricity.

John

The politicians seem to have a lifetime right to other things:

Best regards, Spehro Pefhany