Archived Forum

Hi all,

A few of us on the midnite solar forum have been working on the above project. Initially it was a way of monitoring midnite off grid charge controllers, but its now growing into a project more akin to yours.

First of all, hats off to everything youve done, its fantastic. You are much further on than we are. Reading here, our projects while applied to grid v off grid have approximately the same goals, so i guess im hoping that we can cooperate in some way. We certainly dont want to duplicate things that you are doing, thats pointless.  However i cant immediately see, just yet, how your building blocks will work with our needs.

The two main differences between our projects is DC v AC and the fact that with off grid an internet connection cant be assumed. A typical off grid installation and its monitoring needs are these.

1. PV array in the range 1-3kWp, with occasional support from wind and microhydro.

2. one or two 80 amp charge controllers charging lead acid banks in the 10-30kWh range.

3. 2-6kW off grid inverter supplying AC needs.

The charge controllers and inverters both tend to produce adequate monitoring data, usually by ethernet or serial, but using various semi-proprietary protocols, modbus, zanbus etc. And basically this is what we are doing to date, collecting this data and storing it on low power local servers.

As well as battery voltage and temp, and charge controller input and output data, the midnite family now also includes a battery current sensor, which allows us to monitor DC loads and battery state of charge (SOC). It is connected via 1 wire to a 500A 50mV deltec shunt. So this data is good and nothing much more is required. However where we became interested in your sensor node concept is the following:

- the desire to add reference solar insolation data, to better know what surplus power exists, to inform opportunity loads. The 'waste not' implementations in the charge controller are fairly blunt instruments, and we are hoping to get better data. Load diversion is typically to domestic hot water, occasionally well pumps, HVAC and such. This is done using three methods, battery voltage DC, PV voltage DC, and AC. All three have pros and cons, and we need support for each.  Our plan for this is a small PV in the 2W range on the same plane as the array, and used as a reference panel (having the same glass and PV material as the main array). Basically measuring its Isc with an ADC, and a wireless connection to the monitoring system.

-This method is superior than theoretical models of insolation, (which we are also working on using sandia's PV-LIB, however collecting back of panel temperature, humidity, windspeed, and air pressure would allow us to monitor actual PV performance against the sandia models, which is a way of checking for system malfunctions or design issues. Having gone that far, we may as well get rainfall, to complete a full weather dataset.

-we also have a subproject underway which will monitor the cell balance, measuring the voltage of each cell. This uses a cascading set of PIC controllers each measuring 3 cells.

-the flexibility for each installation to define and add their own unique sensing requirements.

In terms of hardware we like the beaglebone and cubieboard. These baords have just enough resources to run a full webserver implementation in order to serve monitoring data to LAN connected computers and mobile devices. This is our primary and preferred way of presenting data. A cheap android tablet serves for those that want something like your emonglcd.

Learning from what youve done, it appears that we need an arduino/atmega communications board to talk to the remote sensors, which is connected to the ARM board vie uart/i2c/spi. It should also allow direct connection from sensors in the electrical room. Then a flexible radio connected sensor node design, with a variety of plug in sensors, using some kind of  standardized port connector.

The beaglebone in particular does  have fairly good connectivity options, several uarts, 12bit ADCs etc, but we think in best to outsource the 'realtime' activitys to another board, as the ARM board has its work cut out talking to the RE gear, and rendering webpages.

Thats about it. Wed be interested in any pointers to material you might already have, or any ideas at all how we might work together. We do have a guy cabable of designing and building PCBs, and my own focus is on the web application.

The project page is http://code.google.com/p/theblackboxproject/

As a guide my own off grid system is documented here http://zoneblue.org/cms/page.php?view=off-grid-solar