One noteworthy item: I recently switched my gas and electricity supplier to Scottish Hydro Electric, who supply all their electricity from 100% renewables, ie. hydroelectric in Scotland.  As part of their “better plan“, they offer you discounts if you reduce your energy consumption, and to help you do this they send you a free… CurrentCost meter!  So now I have 3 of the things. :-/

Given the inherent inaccuracy of the CC, I’m more inclined now to do some hacking of kill-a-watt style meters which monitor individual appliances… it occured to me that if I can say that the CC’s inaccuracy is purely down to the inductive load in my fridge, then I could monitor the fridge independently and supply a correction factor to the arduino so it can calculate true overall consumption before sending the result to Pachube.

I’m also interested in building a DIY gas monitor, which since my meter has a reflective spot on the least significant digit of the dial, should be pretty simple using an infra-red reflection sensor.

Current cost says: Last 30 days 74kWh, 7 days 16kWh, 1 day 3kWh.

Electricity meter: 196kWh

Current cost readings: Last 30 days 70kWh, last 7 days 15kWh, 1 day 2kWh.

This is absolutely miles out, under-reading by 126kWh, an error of 64%.

On the other blog I had a comment recently from Eric Eisenhandler, a physicist who’s involved with the Blewbury Energy Initiative in his Oxfordshire village.  I’ll quote the comment in full here:

People using the Current Cost and similar monitor types (Electrisave/Owl, Efergy, Wattson, etc.) – which pick up their signals using current transformer clamps around the live mains feed – should be aware that these have a fundamental inaccuracy if the electrical load is not simply resistive. This is because they work on current, not power, and over-read if the voltage and current are out of phase.

This is the case with some motors (e.g. fridges and freezers), and with low-energy and fluorescent lights. For example, some of my lighting shows up on the Current Cost at roughly twice the correct wattage! There are also problems if you have solar photovoltaic panels connected to the grid, since these monitors can’t tell the difference between outgoing and incoming power. These monitors are useful, but this limitation should always be remembered.

So basically there’s always going to be some inaccuracy measuring purely current, if we use devices that don’t necessarily have their voltage and current in phase.

This error is by far the biggest under-reading I’ve seen yet, but it could be that most of my increased useage last month was due to devices that have this kind of phase relationship.

As you can see from the graph on the left, I now have an arduino-based mini-server logging my Current Cost data to a Pachube feed.

I’m using an Arduino Diecimila plus Lady Ada’s ethernet shield with a Wiznet 811MJ module.

The code is a hacking together of Usman Haque’s pachube ethernet example with Alexis Birkill’s Current Cost parsing code. Using the software serial and ethernet libraries together uses a lot of memory — 13700 bytes which is getting pretty close to the 14336 byte limit on current Arduinos.

The black cable is a hacked ethernet cable connecting to the Current Cost’s serial port, which on my unit uses an RJ45 connector.  The stand made of lego allows the cable in underneath.  The blue ethernet cable goes directly into my broadband router.  The third cable is from a 9V power supply.

It’s the last day of the month, and my electricity meter reading tells me I’ve used 179kWh since 30th November.

According to Current Cost, my averages are 30 days: 139kWh, 7 days: 16kWh and 1 day: 5kWh.

I’m still not sure whether the CC is under-estimating or whether the inaccuracy is something to do with the way it calculates its averages. I’m hopeful to get the CC connected to Pachube very soon and then I can start to take a closer look at the data.

I recently came across this hint on eBay, where Current Cost were selling one of the sensor jaws on its own:

“Best results will be obtained if the sensor is perpendicular to the cable and (if practical) with the cable nearest the hinge end.”

As a result of reading this, today I modified the clamp’s position by packing the cable with cardboard as shown in the picture above. The cable is now at 90 degrees to the clamp and held at the hinge end.

I just had a knock at my door. A young electrician working in the flat across the landing stood there sheepishly holding out a mains plug to me.

“Would you mind if we borrowed your electric, mate?”

He’s working on the “board” in the flat, presumably replacing the meter or fusebox, and so had to turn off the supply. He needs some electricity for lights and power tools.

So I obligingly passed the plug through my letterbox and used a socket in my hall.

Coming back into my kitchen, I see the CurrentCost registering 2.3kW and telling me that if this goes on all day it will cost me £4.98. I don’t expect him to be using that much power for the whole day so I’m not worried. The meter’s now gone down to 280W which is probably ambient level for being at home today with the heating on. When I hear the powerdrill start up next door, I can see that that it’s using 500W. Later, another powertool kicks in but has no effect – huh? Must be cordless. Strange to have this kind of insight into what’s happening next door.

I don’t mind helping out with my neighbours with a bit of electricity (makes a change from borrowing a cup of sugar), but this brings up interesting questions about sharing. If I were to make my own electricity, I could distribute it to the neighbours and have a meter for each to see how much they were using. Maybe I could set limits. Interesting to think of having control over your neighbours power, and by extension their appliances and acvitities.

My electricity meter reading for this month tells me I’ve used 118kWh, but both my CC units (one first gen, the other second gen) are telling me that my last 30 days reading is 82kWh. The “last 7 days” value is 27kWh, which would give a 4-week total of 108kWh, while the “last 1 days” is 4kWh, which would give 120kWh.

Maybe there’s something weird in the 30 days calculation (or my understanding of it), rather than an inaccuracy…? Perhaps once I get this thing uploading data onto Pachube — or at least logging it somewhere in real time — I’ll be able to see more clearly what’s going on.

It would’ve been great to be able to chat to fellow CC-users at Homecamp yesterday, but I wasn’t able to get down to London this weekend unfortunately. Apparently they’re all going to get a new CC128 device that’s coming out very soon.

OK so it’s the end of October.  Time to look at the meter, and the Current Cost.

The good news is that — according to my electricity meter — consumption this month (to 31/10) was 124kWh, down from 150 last month.

The bad news is that both my Current Cost displays are showing 80kWh for the last 30 days.  Quite a difference.  I’m surprised at this, because it seemed reasonably accurate last month.

One odd thing that happened — on 8th October in the evening, I noticed that both units were oscillating between showing “energy now” as zero for a few seconds, and then back to the actual figure.  Over and over.  I tried resetting and re-pairing each display unit, and in the end had to do them both at the same time to get things working again.  So maybe during this process they lost some data about averages or totals.

I’ll be keeping an eye on them.  Once I manage to get one of the units logging to a database it will be easier to spot any anomalies.

Anyway, bottom line is that October’s consumption was 17% lower than September, despite it being darker and colder, and so using more heating and lighting.  Switching off the freezer seems to have made quite a difference.

I’ve just had my first electricity bill, from Scottish Power, and they’re charging 12.747p/kWh.  So looking at it simplistically, at this rate that’s saved £3.30.  According to their published fuel mix data, that’s a saving of about 16kg of CO2.  Of course I’ve only been monitoring the meters for 2 months now, and each month is likely to be different in terms of heating and lighting, so I would need to compare with the same month last year.   But it’s a start.

A few weeks back (er, 21st Sept according to flickr!) I tried out one of the newer Current Cost units I bought on eBay, where they were being sold without the current sensor clamp.  This newer unit outputs more detailed data, including averages, and uses 9600 baud. Since the serial cable enters underneath, and I’m using a hacked ethernet cable for the RJ45 connector, I had to build a small stand out of Lego so the cable could enter properly.

I connected it up to an Arduino pretty easily, using the example code by Alexis Birkhill.

The next stage is to use an ethernet shield with it so it can put data online without the need for a host computer.   This isn’t quite as simple as it sounds at first, since I need 2 serial ports. The Adafruit shield I have, with a Lantronix XPort, uses software serial to communicate with the Xport, and although it’s not documented anywhere I could find, it looks like you can only have one software serial device at a time.  I tried using the hardware serial port to talk to the current cost, but couldn’t get it to work in the tiny bit of free time I had.

(This is a shortcoming of Arduino in general I think… I’ve been wondering if adding a couple of extra hardware serial ports using an I2C UART might be useful.)

Need to come back to this.  I’ve got an account on Pachube that I’m keen to start using with the Current Cost.

I’d been waiting for a while to get an idea of the default consumption of the flat, but now I’m going to start changing things.

The main change I’ve made is to switch off the freezer.  Living here on my own it would be hard to justify having a separate freezer, even if I did cook loads of stuff in bulk and freeze it.  According to the CurrentCost, the freezer used around 130W.

An interesting side-effect of losing the freezer is that I can appreciate how efficient its neighbour the fridge is.  It’s really quiet, and I hardly ever notice the pump coming on — I often work at the kitchen table and it’s easy to notice noise in there.  I guess it’s possible that the lack of heat coming out the back of the freezer is making life a bit easier for the fridge.

I’ve also been round replacing some of the light-bulbs, and since there were no energy-saving bulbs at all in the flat, this was easy.   I took out a 60W and 3 x 100W incandescent bulbs and put in 11W (60W equivalent) CFLs.   I also have another two 11W CFLs in lamps.   Next job is to replace the bathroom lights, comprising 4 GU10-style halogen bulbs, with low-energy equivalents, which are harder to find, but they’re out there.  The CurrentCost says the existing halogens use around 160W, which would come down to 44W with LED replacements.

Of course lights are one thing but they don’t use a huge amount of energy, and I’m pretty good at switching them off when I don’t need them.   On the other hand, the electric oven uses about 2kW and is often on for 15-30 minutes just to warm up a pizza or quiche.  More on this later, but again it seems that all this is more about changing habits than about changing technology.