I really have no excuse, aside from just being slack with the updates! So on with the show.
After the power went on everything seemed to be working fine - the Sunny Boy panel inverter producing power, and the Sunny Island charging up the batteries.
Everything however was not quite right.
The Sunny Boy is a grid-tied panel inverter, which in 99% of cases is used in installations which have access to the electricity grid, and the power generated by the panels is fed into the grid. In this mode there's no throttling - every watt of energy produced is fed into the grid and this is how the inverter was configured out of the box.
In our installation we have no access to the grid, so we have to balance our energy production against our energy use. Power we don't use directly is used to charge the batteries, but once they're fully charged we simply have to stop generating electricity as we have no "dump loads" (basically heating elements which turn excess power into heat). The Sunny Island inverter/charger controls our "grid", and signals to the Sunny Boy that it should throttle back by raising the frequency of the grid slightly from its nominal 50Hz. Conversely, when it needs solar power to serve loads or charge the battery it drops the frequency slightly.
Because I hadn't configured the Sunny Boy for off-grid operation, it was blindly ignoring the "hints" from the Sunny Island until the frequency exceeded 55Hz, at which point it disconnected with a "grid fault" error. This served to effectively modulate the power produced, but was extremely coarse and if left, would likely cause the batteries to be over-charged.
Once I configured the Sunny Boy in "island" mode, the two inverters talk politely and the grid frequency remains between about 49.5Hz (when solar power is required) and 52.0Hz, when it's not. Perfect.
Except it's not.
When Nenad was onsite during the week to diagnose an earth fault causing one of the RCDs (safety switch) to trip, he noticed that occasionally the power would drop out completely for a second, then come straight back on. This seemed to happen in concert with the panels receiving full sun for a little while, then being shaded by cloud. I spent half the day last Friday poring over the settings of both inverters to try and sort this out, to no avail. Sure enough, in solid overcast the system worked reliably but when the panels lost full sun, the power would drop out for about a second and then return.
Even more curiously, this only happened when I had about 1kW of load present. Odd.
Unable to solve this myself I ended up calling SMA Australia for assistance. While talking them through the problem and describing the symptoms, the power dropped out again and I noticed that the word "search" flashed up on the remote control very briefly at the time the power went out. This proved to be a strong clue to the cause - the Sunny Island has a power saving "search" mode whereby it detects how much load is present on the inverter and if it's lower than a configurable threshold, it goes into "sleep" mode to save power. What was happening was that although there was plenty of load (1kW) present, when the Sunny Boy fed in a couple of kW as it got sun on the panels, the system transitioned between the Sunny Island supplying the loads from the battery, and the Sunny Boy taking over. When the sun disappeared again, we got the same transition but it seems the Sunny Island detected this as a zero nett load for a short period and decided to go into sleep mode. Once the feed in from the Sunny Boy dropped below that of the loads, the Sunny Island detected the loads again and powered back up.
SMA think this is a bug in the software (I agree) and are working with Germany to sort it out. In the meantime, I've got the system stable by simply turning off this "search" power-saving mode.
In other news, Peter and I began setting up the little concrete slab which will cover the void in front of and above the shipping container, and underneath the large window of the unit. This will be a nice spot for a shade sail and outdoor furniture :)
The aim is to pour the concrete here on the 6th of September.
No comments:
Post a Comment