4 ms·

Very nice work overall, and I admit I haven't read the entire project, but it seems the entire authority to operate both pumps has been delegated to a microcont

Very nice work overall, and I admit I haven't read the entire project, but it seems the entire authority to operate both pumps has been delegated to a microcontroller that becomes a single point of failure for the entire system. Given the cost of the system and chance for failure I think it would make sense to either add a second system or enable some sort of automatic override (possibly via mechanical float switch wired directly to large pump and set at a higher water level than the other two switches) So that there will be enough time between a potential large leak occuring and detection of controller failure(via lack of daily email) to prevent it from sinking the boat.

rkangel 5y ago
That wasn't my read. My read was that they go off independently based on their own conditions (the large pump has a float switch that presumably isn't set off until after the small pump should have taken care of it) and the monitor just records these things.
- sathackr 5y ago
  Based on what was wrote I don't think that's the case. He included some special conditions like: >The monitor knows if I take the boat out when I turn on the main boat power and switches the small pump to auto and otherwise stops monitoring. But when the main power is back on, the pump will initially run every half hour and then every hour and then after about a day it will only be running once a day" Though reading closer there are only two relays, and the author talks about controlling both the small pump and the fan with relays, and there is language around the large pump that while isn't conclusive, may indicate it is indeed on a separate float switch and system and the ESP32 is only involved in the monitoring of it. > I have a large pump that only goes off if the small pump fails or the boat is trying to sink. If the float switch lifts, the monitor wakes up and sends an email alerting me.
vvanders 5y ago
Yeah, I automated some stuff out in our greenhouse, I ended up going with a PLC[1] which is a bit more robust and designed for reliability and industrial environments. They're very much like (expensive!) lego bricks that you can easily throw together + program. The PLC sends me a mail each morning with how much water it used, the max temperature of the water pump(I repurposed a RV water pump with a PID PWM + SSR driven from the PLC) all running from a 24v solar system. They're very much designed to be in-the-loop on systems that need to be robust with pretty extensive error report/alarms and the like. [1] https://www.automationdirect.com/productivity/p1000 https://www.automationdirect.com/productivity/p1000
analog31 5y ago
I've built a lot of control systems, and I've learned the hard way that switches and relays (including solid state) can fail. And they can fail in both the "off" and "on" positions. It would probably be beneficial to monitor the status of the control outputs, to see if they actually agree with the inputs. Also, of course the microcontroller and its code are points of failure like you say.
- allene222 5y ago
  Just some clarification about my pumps. The large pump is indeed on a float switch and will operate independent of the monitor. The small pump is an auto pump that normally goes off every 2 1/2 minutes and senses when it is done pumping. I have modified the code so that the last thing that happens before sleep is the pump power is turned off so that if the software hangs, the pump will still operate in its normal auto mode. One reason to avoid having the pump always in auto mode is they fail after about 4 years. I am hoping that by only running it once a day that it will last longer. If the small pump does fail, the large one keeps the boat from sinking. I am not sure if I specifically mentioned it in the article but this is a wood boat and it leaks. The pump will run about 25 seconds a day to remove the day's water. I have the sequence after running the engine because a hard sail can cause the leaking to increase until things swell back up. There have also been several changes to make the monitor more robust since I wrote the article. One main one is that a lot of the sample code on the web for email monitoring has "while" loops. These need to have counters to exit if the WiFi is down for example.

rkangel 5y ago

That wasn't my read. My read was that they go off independently based on their own conditions (the large pump has a float switch that presumably isn't set off until after the small pump should have taken care of it) and the monitor just records these things.

sathackr 5y ago
Based on what was wrote I don't think that's the case. He included some special conditions like: >The monitor knows if I take the boat out when I turn on the main boat power and switches the small pump to auto and otherwise stops monitoring. But when the main power is back on, the pump will initially run every half hour and then every hour and then after about a day it will only be running once a day" Though reading closer there are only two relays, and the author talks about controlling both the small pump and the fan with relays, and there is language around the large pump that while isn't conclusive, may indicate it is indeed on a separate float switch and system and the ESP32 is only involved in the monitoring of it. > I have a large pump that only goes off if the small pump fails or the boat is trying to sink. If the float switch lifts, the monitor wakes up and sends an email alerting me.

vvanders 5y ago

Yeah, I automated some stuff out in our greenhouse, I ended up going with a PLC[1] which is a bit more robust and designed for reliability and industrial environments. They're very much like (expensive!) lego bricks that you can easily throw together + program. The PLC sends me a mail each morning with how much water it used, the max temperature of the water pump(I repurposed a RV water pump with a PID PWM + SSR driven from the PLC) all running from a 24v solar system. They're very much designed to be in-the-loop on systems that need to be robust with pretty extensive error report/alarms and the like. [1] https://www.automationdirect.com/productivity/p1000 https://www.automationdirect.com/productivity/p1000

analog31 5y ago

I've built a lot of control systems, and I've learned the hard way that switches and relays (including solid state) can fail. And they can fail in both the "off" and "on" positions. It would probably be beneficial to monitor the status of the control outputs, to see if they actually agree with the inputs. Also, of course the microcontroller and its code are points of failure like you say.

allene222 5y ago
Just some clarification about my pumps. The large pump is indeed on a float switch and will operate independent of the monitor. The small pump is an auto pump that normally goes off every 2 1/2 minutes and senses when it is done pumping. I have modified the code so that the last thing that happens before sleep is the pump power is turned off so that if the software hangs, the pump will still operate in its normal auto mode. One reason to avoid having the pump always in auto mode is they fail after about 4 years. I am hoping that by only running it once a day that it will last longer. If the small pump does fail, the large one keeps the boat from sinking. I am not sure if I specifically mentioned it in the article but this is a wood boat and it leaks. The pump will run about 25 seconds a day to remove the day's water. I have the sequence after running the engine because a hard sail can cause the leaking to increase until things swell back up. There have also been several changes to make the monitor more robust since I wrote the article. One main one is that a lot of the sample code on the web for email monitoring has "while" loops. These need to have counters to exit if the WiFi is down for example.