Björn Stenberg

Yesss! A brief power cut resets the charge timer and lets me charge as long as needed:

It then went out mowing for ~1,5 hours and came back with practically identical battery voltage. This shows another firmware-imposed limit: It doesn't deduce remaining battery capacity from voltage, instead measuring the energy spent and comparing that with what it thinks the battery capacity should be. There is however an undocumented battery capacity calibration process. I'll run that next and see what result it gives.

New firmware limit found: There is a max charge time of 5 hours. Since the firmware also imposes a max charge current of 3A, that gives us a maximum of 15Ah per charging session, i.e less than half the pack capacity.

Problem solved, the first all-night mowing is completed! It turns out there is a product called Robonect, which essentially middle-mans the communication between the mower's UI and mainboard. By doing this, it can control the mower in all the ways the UI can. It even contains a premade feature to extend mowing time by putting the mower in manual mode for as long as you like, and then restoring it to auto mode when it's time to go home and charge. Elegant! Just what I needed.

Gosh. It ran for a whopping 28.5 hours on a single charge! That's... a lot more than before. 😂 Now to see if I can convince it to not abort charging after five hours. I'll try simulating a power outage after 4.5 hours and see how it responds to that. And then I have to do this 28-hour calibration run two more times. This is going to take a while.

Calibration is still ongoing. It is done by running it in the "secondary area" mode until the battery is emtpy and it stops out on the lawn. Supposedly, after doing this three times in sequence, the measured capacity is adopted as the new battery size. I'm still on the first of those three runs, 15 hours in and still running! It's at least clear the new battery is plenty large enough for my intended use case.

We can see at the end of the graph that it finally hit the end of the constant-current phase and entered constant-voltage, signalling the battery is finally full. So we know it will actually charge it fully, but we may have to do some tricks to get around the 5-hour limit.

@gufo@mastodon.nu Elflyg har haft samma syfte som vätgasbilar: En kuliss att peka på för att kunna argumentera att saker är på gång, så att inga beteendeförändringar eller miljökrav görs. En bluff för att säkra kortsiktiga vinster.

However I personally don't run the mower all day every day. I typically run it for 6-8 hours overnight. And then I have the whole rest of the day for charging, and can potentially/hopefully run the entire mowing session in one go, without interrupts for charging.

@bagder @dtemme 👋 😄

@Jeremiah It can be disabled, I think. They have like 30 different AI-related preferences in your account profile and I spent the time to go through and disable every one. I don't get any AI nags in my meetings now.

@digimint @TechConnectify Second, to address your specific needs, it appears the "GPIO Integration" available in HACS supports reading a DHT22 sensor directly from the rpi header. https://github.com/rosenkolev/home-assistant-gpio-integration

Calibration completed. At 28h runtime + 15h charging per cycle, running three cycles takes some time. Unfortunately it didn't work. The mower still assumes factory-default capacity and so 93% of the new battery sits unused when running in auto mode. I might try calibrating again. I had some issues with charging during this run that might have reset the counter. But I'm not very optimistic. As I understand it the calibration process exists to handle _reduced_ battery capacity, not increased.

"But Björn, if you can't increase the charging current, what is the point of a bigger battery anyway?" You're right. If we run the mower 24/7, then a bigger battery doesn't mean more time mowing per day, it just means fewer and longer charge cycles. But that has value too. Each charge cycle adds time spent for the mower to find its' way back to the charger. It does this in several ways, but one fallback is to follow the edge line. If this is done often, it will cause visible ruts there.

@digimint @TechConnectify Home Assistant does not require wifi, or any other specific networking protocol. You can use whatever technology you prefer, including wires and relays. The point is that you are in control of your home and can decide exactly how it works, rather than being at the mercy of whatever some device manufacturer decided.

@digimint @TechConnectify No, you don't need any more networking than you want to have. You can connect sensors and actuators in any way your chosen controller supports, such as gpio on a Rpi or USB on a PC. Neither ethernet nor wifi is required if you don't want it. You select a solution that matches your requirements. If you want wired communication, you choose a controller with wired capabilities.

@digimint @TechConnectify I cannot speak for other people or their comments. Our dialog started with me replying to your message saying that HA is a poor choice for HVAC control because it requires wifi. I absolutely watched the video and my takeaway from it and earlier ones is that Alec has a keen interest in how things work, and a desire to influence how they work. That type of personality/interest typically matches well with the flexibility and power that Home Assistant offers.

@digimint @TechConnectify I am not a proxy for those youtube commenters. 😃 I will not argue on their behalf.

@TechConnectify I strongly recommend looking into @homeassistant. It makes these kind of DIY customizations _so_ much easier, since you can combine off-the-shelf generic sensors and relays and don't have to reverse-engineer expensive proprietary single-purpose products with poor documentation. (And, unlike proprietary smart thermostats, Home Assistant is open source and fully local: No network required.)

@nixCraft@mastodon.social "AI profits"?