I ran two more experiments yesterday. I don't remember what they were, but I recall them being somewhat incomplete, and more boring, but I do recall at least one question answered. Let's decipher what I did:
I do remember turning off the utility connection early in the morning, and not turning it on until the evening. No usage shutdowns were experienced anywhere.
Here's my first graph:
This was the more boring one. Apparently, I used energy from batteries and sun. Since the batteries were between almost depleted and almost full, they were fully flexible, able to soak up the sunlight when excessive, and able to provide energy when sun couldn't provide everything, and the sun was fully utilized. Loads were less than sun+battery. Pretty simple test, pretty boring, and pretty well known outcome.
Here's my second test:
Ok, I remember this one. I wasn't home 1PM-5PM, so couldn't do more interesting tests. This did by itself set up for a slightly less interesting test, more informative than the first one of the day. Notice what happens when the battery gets almost full. Notice the battery does not let itself get fully full. It turns off sun input completely. Considering home use was still 600W, that is unfortunate; to me, using 600W from the SolarEdge would be superior. But the reaction is binary. I'll zoom in to this time period:
Ok, that zoom in looks sized well. So, let's see with frequency information:
Looks like home frequency sort of follows battery, so I'll take that line out to make it less cluttered, as well as grid, since that was off the whole time:
I hadn't realized until now that I was recording Solar frequency as well; apparently, the home frequency covered it up:
Let's declutter some more:
Alright, that clearly shows that SolarEdge and Tesla do not talk to each other as corporations, AND the ramp-down is immediate and the ramp-up is not. Tesla's system clearly doesn't attempt to ramp down the solar inverter output in preparation for matching it to home use; I consider that a fail. Tesla's system also clearly attempts to ramp up solar output, and the solar system does not, because by then, the SolarEdge is in a sleep mode and ignoring everything. This is a total lack of corporate coordination and a lack of planning on the part of the battery to ramp down before it is at its full point. Tesla should ramp down solar output to near home input as the battery nears its full point, giving enough battery buffer to deal with the variations that will happen.
The rest of the evening, I attempted to run Test 3, which I forget what it was since it's morning and my brain isn't all the way at capacity yet. But I do recall what happened: I came home, put the clothes in the dryer, and the battery level never got to near full again. I'll try again today, if I remember what the test was going to be.
Edit: I noticed frequency coming into that zone is not 60Hz; I'll back up in time:
Very interesting! Looks like the SolarEdge is trying to say something to the Tesla, and the Tesla is trying to say something to the SolarEdge, but they both speak different languages; notice how the SolarEdge has spikes increasing in frequency (every so many minutes), whereas the Tesla has a steady decline (from ~60.055Hz on the left to ~60.046Hz on the right). I should pan left (go back in time) some more:
Ok, here's back in time, and indeed raises more questions:
They were dancing at some point during the early dryer events 12:00-13:10 or so; not sure if they managed to hook up. I'll try to zoom into that a bit.
I need my coffee. That looks like a bunch of squiggles to me right now. Let me try harder.
Dryer heater element turns on and off to get to a thermostat temp. It starts off in the graph off. The frequency is near 60.04Hz. The dryer heater element then turns on, and frequency drops to 59.94Hz, rising slowly to almost 59.96Hz. I wonder if there's a relationship between use, state of charge, etc., and the frequency.
That's why I think playing in-band communication games (with frequency as the info carrier) doesn't carry enough information for a proper energy backup and management system, especially if there's more than two components as systems age. For instance, I want to add solar and batteries to my home; they might not all be the same brands.
I looked at SunSpec(?), and it seemed both incomplete and cluttered; they should work on reducing the clutter and making it complete for proper control systems. Well, what I'm really saying is we need control systems that are simple enough but have enough info and controls.
Of course, there's a whole industry of electronics controls; they don't need an entire factory team to install a simple home battery and solar with a possible multiple of brands and of each component. That needs to be worked out.