DIY solar diverter and smart meter

[Matthew Marks]

I got an electricity meter with Modbus interface from eBay for a good price. I use an Arduino to interrogate this once per second to pulse width modulate the element of a vacuum flask kettle (1.8kW) to heat water with otherwise exported energy. This works so well that I decided to do the same with the 3kW immersion heater of my hot water cylinder. The heater only warms the water to a certain depth, so when its thermostat opens (as sensed by a current transformer) I operate a small 12V pump that takes water from the bottom of the cylinder to the top until the thermostat closes again (or times out). This gives nice stratification allowing you to get usable temperature hot water even if there isn't enough solar power for the full cylinder. A few enhancements such as detection of draw-off to prevent the pump coming on and giving you a cold shower, and I haven't used any gas since the Spring when I installed it.

Anyway, I just got myself smart meters and I discover that,

a) The impulse LED on the new electricity meter behaves differently from the old one. On the old one it glowed solidly when exporting but on the new one it indicates import and export impulses, thus making it impossible to determine whether the meter thinks it's importing or exporting. I had a photodiode on this LED connected to another Arduino which would bleep furiously at me if importing; I can no longer use this. I have said I expect them to configure the meter so that the LED behaves in the way expected, but I don't even know if that's possible.

b) More worryingly, there appears to be less "elasticity" in this meter than the old one, in that the in-home display often claims that I am consuming power due to the solar diverter. My bleepy box was very useful in that it would alert me instantly to any problems, and I very occasionally got an impulse (1Wh) with the old meter, but this one is definitely racking up the cost more quickly and of course I don't have the bleeps to help me any more.

Anyone have any experience of how the characteristics of (UK) smart meters might be different? Yes, they can measure export as well as import, whereas the old meter was just aware that export was occurring as a "ratchet", but I would not expect their "ballistics" to be any different because they should also be designed to simulate how mechanical meters behave in order to give comparable readings.

 

[Battman-1701]

I can't answer your exact questions, but this may be helpful.

A colleague of mine discovered that 'burst mode' diversion where you let almost 1Wh export, then switch on the load to reimport it before the meter notices, was not working with his new smart meter in New Zealand.

On a weekend when his family was away, he was able to switch everything except the water heater off and evaluate the meter's characteristics through empirical experimentation. He found that the meter was measuring power across a 10 mains cycle window and that PWM diversion worked well.

He & I are using Arduinos to monitor the power, courtesy of the work done by the good people at open energy monitor.
The load is controlled by a solid state relay with zero crossing.

Be aware that utility operators are really touchy about DC injection, so NEVER do half cycle control.

On a 50Hz system, 1 cycle is 20ms and represents 10% of the 10 cycle 200ms measurement window.
Running the load for 1 cycle in 10 diverts 10% of the load power.
My arduino uses these PWM on times repeated every 200ms
Power Time Cycles
0% 0ms 0
10% 15ms 1 Remember the zero crossing relay runs to the end of the cycle
20% 35ms 2
30% 55ms 3
40% 75ms 4
50% 95ms 5
60% 115ms 6
70% 135ms 7
80% 155ms 8
90% 175ms 9
100% 200ms 10