Electricity meter question.

Bob Minchin wrote:

I know it is bad form to reply to your own posts but it might help someone.

I have found the answer here
http://universalmeterservices.co.uk/store/images/5193.doc

The red light either flashes every 1 watt-hr consumed or stays on
continuously when the power being consumed is below a preset level.

The display flashing rEd is showing "reverse energy detection" which
would normally mean fraudulent use but in the case of power export from
microgeneration schemes (PV, wind etc) is legitimate.

In my case, when the light is on continuously it means that I am
generating more power than I'm using - this is confirmed by the display
reading now being identical to the one I took 8 hours ago.

According to the manual, the meter has a separate register measuring
exported power although the reading is not displayed.

Bob

On Apr 20, 5:10 pm, Bob Minchin


All the meters data is recorded in a Ferroelectric Random Access
Memory (FRAM) under the control of the microprocessor.  All the kWh
registers are stored in the FRAM and are updated every 1/100 th of a
kWh.  The FRAM is guaranteed for a minimum of 10,000,000,000 write
cycles


WTF?  Core memory?!  8-)

Andy Dingley wrote:

herehttp://universalmeterservices.co.uk/store/images/5193.doc

Yes I found that difficult to credit but I suppose it maybe used to meet
a possible regulatory requirement for N years retention with no power
connected??

Having looked further, I think it is the 10,000,000,000 write cycles
that makes it ideal for this application to cope with the update rate.

Bob

On Apr 22, 4:00 am, Bob Minchin

Not quite core, but not completely unrelated and quite interesting
http://en.wikipedia.org/wiki/Ferroelectric_RAM

It's been around for quite a while. FLASH and EEPROM are mostly
useless for this type of application. Given it's an electric meter
they could just use low power SRAM with a super cap backup supply.

MBQ

Man at B&Q wrote:

herehttp://universalmeterservices.co.uk/store/images/5193.doc

interestinghttp://en.wikipedia.org/wiki/Ferroelectric_RAM

Too easily corrupted. Fe RAM has indefinite data retention after removal
of power, better than 10^^15 write cycles, and a glitch during writing
won't corrupt the rest of the data (According to the makers...)

--
Tciao for Now!

John.

wrote:

Easily designed around.


You would struggle to "corrupt the rest of the data" with any
technology.

MBQ

Man at B&Q wrote:

herehttp://universalmeterservices.co.uk/store/images/5193.doc

interestinghttp://en.wikipedia.org/wiki/Ferroelectric_RAM

To use one example meter I know of, in what was an empty shop for three
years, how would your idea cope with a couple of years of disconnection?


I must remember to tell the makers of my cameras that. They tell me
under no circumstances must I take the batteries out while it's writing
to the card, as that may render the card useless.

--
Tciao for Now!

John.

wrote:

OK, so go back to using FRAM.


My point stands.

The warning, "may", is due to potential corruption of the card's file
system that isn't designed to be robust enough to withstand corruption
of a single critical memory location. The data is still there if you
have the right tools to recover it.

MBQ

Man at B&Q wrote:

Not always, IME, but you may have been luckier. Luckily, the camera with
a problem was one used by a member of a group, so they just swapped
pictures.

--
Tciao for Now!

John.

FAT filesystems can't be made safe in this way because they require
multiple sector updates to be made simultenously, to remain self
consistent. They have a common delayed action corruption mode where
you can read all the data OK after an interrupted operation, but latent
inconsistent spacemap update results in data corruption sometime
later on when the filesystem is written to again. Also, because there
are so many different implementations of FAT filesystem drivers which
do things in different orders and may not use the clean shutdown bit,
it's not easy to spot this without doing something like a full chkdisk
operation, which appliance devices usually don't implement, nor have
the time to do.

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

On Tue, 17 May 2011 12:55:31 +0000, Andrew Gabriel wrote:


Can you explain? If they are built with one FAT, I can't see why that
would be so, although there might be some free block loss.

--
Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

*lightning protection* - a w_tom conductor

You never worked with bubble memory did you?

It was actively worked on where I used to work, but not personally,
no.

I was thinking of more modern, mainstream, technologies, not dead
ends.

MBQ

It had some interesting faults.
I killed the development of bubble memory cards on SystemX by designing
something far better to replace them.
Cheaper, more reliable and five times as fast tends to kill the competition
rather well.
Its still in use now and its more than a decade (or two) old.

[sorry for incorrect threading and I hope I got the original attribution
correct, I've only got half the thread visible]

Andy Dingley wrote:


Build your own 32bit core store for an arduino

http://www.corememoryshield.com/report.html

On Sun, 15 May 2011 08:57:08 +0100, Andy Burns wrote:


Yup, saw that on sparetimegizmos recently. I do actually have a piece of
antique core - 512m bytes from an Atlas machine!




--
Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

*lightning protection* - a w_tom conductor

Bob Eager wrote:


If you fancy 'knitting' some more £25 for 45,000 ferrite rings

http://cgi.ebay.com/300556183278

On Sun, 15 May 2011 10:28:48 +0100, Andy Burns wrote:


That would be bad enough if I had two eyes! As it is....

--
Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

*lightning protection* - a w_tom conductor