Archived Forum

Hi everyone,

Some of you may be familiar with the old "Fusion" sketch which I wrote some time ago. For those of you who haven't read about it my idea was to take Robin's MKII sketch and fuse it with the energy monitor and to send the results to an emonGLCD display. 

The original version had a number of correlation issues with some of the other methods and, as a consequence, I rewrote it and achieved a much better correlation. The sketches associated with this new version were published some time ago.

In the past few weeks I have been working upon the sketches once again. I wanted to extend the capability and be able to read the infra red pulse output from the Elster AC100 meter. Sadly, I didn't get that working. Instead, I decided to add additional temperature monitoring capability and to also add a feature whereby the results could be sent to the web.

The picture below shows the output on the GLCD display. The bold numbering is the grid power, the smaller numbering is the solar power (on a wet Monday afternoon) and the four temperature outputs are shown on the right hand side of the display. "3600" is the energy bucket and the black spot shows a net import of energy. This changes to a heating coil when the PV dump is running and a white ring when the tank is full of hot water.

All of the old features are retained: LED's that turn from red (importing) to green (net export) and a combination of the two when heating the water). The LED's turn off under low light conditions but maintain a heart beat each time data is received from the emonTX.

I have added a Nanode RF SMT to the hardware and have modified an existing sketch so as to take the emonTX data and route it through to emoncms. I have been working upon a dual transmitter to emoncms and to cosm but have yet to get this operational. Rather than continue I wanted to publish what I know works OK.

My feed is directed to emoncms and I have configured a dashboard as shown below. It's a promising New Years Day morning and my wife has been cooking breakfast. We can see the griddle on the oven pulsing on and off while the kettle is simultaneously boiling (my contribution to breakfast is making the tea). The displays at the top show real time visualisations of grid and solar power together with the energy bucket and the top and bottom temperatures of the water tank. The probes on the tank are not invasive so give only an indication of temperature rather than what will come out of the shower. Lets hope that the PV dumper increases the temperature of the water tank. 3 degrees C is the outside temperature on the front of the house.

I also have dials which show the grid and solar power for the day. These normally reset at midnight.

The bottom of the display shows a zoom of grid and solar power on the left hand side and bar charts day by day on the right.

The cms display was written using the web site tool. If anyone is interested, there are probably ways to share the code or the framework of the tool.

Shown below is the schematic. There is a minor modification from the previous one in that the IO lines for the red and greed LED's are switched around. The rest is as before.

Should you wish to add temperature sensors beyond the one in the GLCD, you can add them and simply plug them into the temperature sensor port. The software will look for up to three of them.

The emonTX sketch has undergone a huge amount of change: as I mentioned before, the measurement algorithm has been reworked so as to comply with the OEM published versions. It runs over a six second window with a minor jump out to sum the results and send them into the wireless network. The MKII sketch runs in the background and in tandem with the energy monitor.

I did some trials to speed up the code. I took it about as far as it would go without resorting to integer mathematics. If I am honest, I found that speeding the code (taking more samples per mains cycle) didn't make very much difference beyond a certain point. Just to be sure, I ran some trials in VBA using various amounts of under sampling of waveforms with varying amounts of distortion and my simulations yielded the same conclusion: it doesn't make that much difference, to my mind at least.

So, what we have is reasonably slim code which draws upon some of what Robin has done together with what else is needed to measure solar channels and the odd temperature here and there. We can turn off un required features such as LED displays, apparent power calculation, transmission to the GLCD, power displayed on the serial monitor, etc. if we want to make some bits quicker. Best to play around a bit and see what suits.

What else is in there?

This version of code includes a simulator whereby the user can generate voltage and solar and grid currents with any amplitude and phase offset. A useful tool to debug code without needing transformers or a solar energy system. 

There is also a feature to sweep phasecal independently of the grid and solar channels. This is a useful feature to allow you to characterise the phase lead/lag or each channel and work out which one correlates best with your reference.

If you use the emoncms software there is a mode where fixed data can be sent from the emonTX to the base station (and on to the web) so as to allow you to unravel which node id is which.

I've also added a feature which allows the solar offset error to be removed from the transmitted data. It's really tough to get accurate power measurements on each channel across a wide spectrum of power. What is accurate at one end may be off at the other. With this feature, power which is repeatable but below a certain threshold is considered an offset error and a solar power value of zero is transmitted in the RF.... a kind of additional switchable high pass filter.

There's a load of other stuff which helps with the debugging as well: seeing all samples captured, seeing the energy bucket content periodically and the raw power calculation results. Most of this probably wont be needed... but is there just in case.

In terms of accuracy, I have run this against the usual loops of wire

and plug in energy monitor. It stands up pretty well against both of these for resistive loads and equally well when the power factor drops with some garage machines. It isn't quite as good when clamped around the power cables on the consumer unit as it's still a little off compared to the enviR monitors. Perhaps a little more calibration is in order... another day!

If somebody can show me how to make my cms dashboard public I will gladly share it!

Anyway, feel free to having a play, code should be included at the end...