Off-Grid Barndominium inverter suggestions

BB. said:

Modern Integrated charge controllers/AC inverter/and even some BMS equipped Li Ion batteries do that--And are even Internet connected (data, configuration, alarms).

Another option is that some charge controllers (Midnite with their Wizbang Jr. remote battery current sense and possibly Outback with their "Flexnet" system).

As you get more complex--There is more to configure, more functions, and possibly a bit more to debug if/when things go wrong.

-Bill

Those are interesting options and I'll be sure to look into them. Between this and what Mike95490 suggested (Logging Voltmeter), it gave me an idea that evolved as I typed this out. An Arduino and appropriate resistors (voltage divider) would make for a tidy means of accomplishing this, with the appropriate code. Last time I played around with an Arduino was nearly 10 years ago, I built a gauge cluster calibration tester. Puts out all the signals that one might realistically see going to a car's gauges, so you can calibrate each one in case of modification, repair, engine/transmission swap, etc.

Found some discussion on the subject in these links, including one with a custom PCB and suitable specs for measuring a 48v system
https://forum.arduino.cc/t/measure-solar-battery-voltage-max-48v-5a/193600
https://openenergymonitor.org/forum-archive/node/11011.html

For this, each 48v bank gets fed into the Arduino on a dedicated I/O pin (after being reduced to 0-5v via a voltage divider), then the data is output on another I/O pin to the appropriate 7-segment RGB displays (https://www.sparkfun.com/products/13999) where it gets scaled back up and displayed as numeric voltage, in different colors based on approximate SOC. For example:
White: Charging (50.4v or more)
Blue: 85-100% (49.6-50.4v)
Green: 65-85% (49.2-49.6v)
Yellow: 50-65% (48.8-49.2v)
Red: 50% or less (48.8 or less)
Flashing Red: Problem (48.0v or less)

Plus, the nice thing about Arduinos is they're cheap and completely standardized. Around $13 shipped on Ebay for an Arduino Mega2560 and $5-7 for a SD card datalogging shield with real-time clock functionality that drops right onto the Arduino. I figure logging the data every hour should be plenty for diagnostic purposes. SD cards are cheap, it's logging maybe 100 kb per day if done in a text file, and one 8gb card would last for decades. It should be able to write the data pretty quick, just the timestamp and the voltage of each bank with a tab space between them. And with a little more work, it COULD be set up to upload the data at set intervals via GSM or Lan/Wifi, with the appropriate shield boards.

The rest is writing the code to make this sorcery work. Now if only I didn't suck at soldering and writing code...

Right as I was clicking post, I found a Hall Effect sensor for measuring current.  In short, it sits at 2.5v when the system is neutral. While charging, it goes above 2.5v, proportional to charge rate and under 2.5v when discharging (current moving opposite direction). Theoretically, it shouldn't be too hard to add the right code to operate and log from these too. Arduino Mega2560s have a TON of programmable I/O available, so the hardest part would be scaling to suit the 0-254 and 256-1023 bit ranges for discharge and charge respectively.

I'm fairly familiar with using a shunt to measure current. I have the same setup in my jumper cables and bench 12v power supply, comes in really handy when testing things like lights, switches, etc. The only issue I can see is determining current flow direction. I want to have the monitoring system at the battery bank (before the inverter) so it measures exactly what the batteries are seeing, and the display colors reflecting what is going on. White for charging, Blue for when between 85 and 100% and not charging, etc.

Here's the link that referenced the Hall Effect Sensor on a 48v system, post #9:
https://forum.arduino.cc/t/measuring-48v-dc-battery-system-and-high-amps/328419/8


Looks like the manufacturer, LEM, makes them in several ranges, from 50A to 600A. For a projects like this, 100A or 200A should be plenty of range and still have good resolution for logging. Plus, it works natively with the Arduino's 0-5v input too. Definitely worth an experiment.

EDIT: I incorrectly described the Arduino bit ranges above. They would be 0-510 for battery discharging and 513-1023 for battery charging. 511-512 would be neutral (not charging or discharging batteries). The HASS 50-S sensor can measure 150A in both directions, so I think it would do the job quite well. Each step is roughly 0.29A if using the entire 0-1023 bit range. The code's logic for this would be summarized as follows:

Execute "Operation 2": (Read Pin 1 State (0-5v), Execute "Current" Output Equation, Output to Display)Repeat for next Parallel string of Battery Bank

30 seconds should be more than enough time for the code to run for every 48v string of batteries, and it would log the data to the SD card every 15 minutes.

mike95490 said:

Hall Effect Sensors are only good for gross measurements.   So it could make the decision of Charging or Discharging.  But you would still need a shunt ( a stable low value resistor that develops a precise voltage + or -  ( charging/Discharge ) for the measuring system.
Hall Effect Sensors drift quite a bit, daily !

A shunt is fine, lower voltages are easier for an Arduino to understand and translate back into current measurements. Pretty sure there are a few arduino current meter code demos floating around the net. Now comes to the fun part:

Since the panel array is sized to meet average daily (as in a full 24 hours) usage in 3 hours, how exactly could the battery bank (sized for 10 days of average-load usage with zero sun) sulphate so quickly when the solar panel array is keeping it at or near full charge under almost all conditions?