(OT ?) Homeplug oddity

[Jethro_uk]

Due to the vagaries of brick walls, and WiFi, I use a Homeplug network in my house.

The setup is Router (DHCP server) in Front room, Server in utility and (other side of the house) network media player (which is off during the day).

The Homeplugs have a "power save mode" where if no traffic is detected, they "sleep". They have 3 lights on them: power, homeplug network detected, data activity.

Last night, settled down to watch TV via media player. There was no way I could get the media player to find the server. Diagnostics seemed to suggest it hadn't been assigned an IP address. However, other PCs wireless connected to the router could see the server.

Checking the router, found no REQUEST.

After 30 minutes of restarting router, server, and media player with no success, we watched a DVD. Before I went to bed I left the router unplugged.

Today, exactly the same thing.

Breaking things down further, I turned off the Router Homeplug, and the Server home plug, and the media player homeplug. Then, for no reason I can think of, I turned the media player homeplug on.

To my surprise, it displayed as if it were connected to a network !!!!!

Now my understanding of these devices is they need to share a ring via the neutral wire.

Does this mean I have a neighbour, using homeplugs, that I share a neutral with ? And if so, is this to be expected ?

There's a button on the side of the homeplug ... "Reset/Encrytion". I pressed this, and the media player homeplug just showed power on. Switching on the router and the server homeplugs bought it "into" the network, and all was well.

Which brings me to another point ... these things are supposed to be encrypted to their own network. So how the hell did my one join another network (unless there's a subtle bug in the firmware around sleeping).

Anyway, that's an hour and a half I'll never get back :( But I did learn something :) Just not sure what ;)

[Mike Barnes]

AIUI the signals are greatly attenuated when going through a meter, so neighbour problems shouldn't arise.

[Jethro_uk]

Curiouser and curioiser :)

[Capitol]

This system is prone to losing connections depending on the impedance of the 240V wiring. It can also pick up a burst of interference and reset itself. Having worked on these systems I am not surprised that your units can display strange problems. I have known situations where the signal will not travel 6ft one day and spreads all over a group of flats the next. It is very enlightening to do a SPICE model of the wiring and see what can go wrong. Unless the encryption is 3 leg checked, noise will appear as a valid signal corrupting the system. That's one of the reasons I wire everything even if it is an aggravation to do.

[Davey]

The instructions tell you that a meter will block the signals. No explanation given.

[Tim Watts]

Current sense coil in the meter has inductive impedance which varies between naff all at 50Hz to a very big number of Ohms for the harmonics used by the homeplug signal :)

[dennis]

Maybe some of the newer ones use hall effect sensors and don't have coils?

[Brian Gaff]

Homeplugs are crap, is the lesson I suspect. Not only are they anti social for anyone trying to use a short wave radio, but I'd expect in a normal street that every other hous on one side of the road might end up getting the signal. Of course this assumes it is mains phases which are shared, and this can be tested I'd have thought. I don't buy this its using the neutral conductor stuff, the coupling inside the mains wiring will put it everywhere I'd have thought. You cannot beat a bit of network cable to permanent sites in my view.

Brian

[Brian Gaff]

Ha! Dream on..... Brian

[Brian Gaff]

Maybe you need a Chinese meter then. I'll have to suggest this. Brian

[Brian Gaff]

Its already coupled pretty well to other lines in the mains though, and so I'd not expect too much attenuation, far too much other coupling going on. Brian

[Tim Watts]

I would imagine they still have a couple of turns at least to get the accuracy up. But I've never had one apart :)

[dennis]

My DC clamp on meter doesn't need any extra turns and it will measure a few mA up to 400A DC or AC. Its accuracy is determined more by the size of the jaws and how far out you can get the wire from the centre than anything else. This shouldn't be a problem with a meter.

[jrwalliker]

I've visited a factory where they are made. The ones I looked at used a very chunky shunt resistor (Manganin I think) with a low-offset op-amp for the current sensing.

Such a design would have negligible attenuation for RF up to hundreds of MHz.

John

[Tim Watts]

Yes I know.

But is it accurate enough for billing purposes?

[Tim Watts]

That's interesting - would make hacking with an AC magnet non viable.

[Capitol]

I believe there are some dedicated Hall effect sensor ICs designed for metering purposes so only a single sensing chip is required.

[dennis]

It wouldn't be difficult to calibrate a digital meter to be accurate even using components accurate to a few percent.