power box

I have build a small prototype system entirely contained in an Ikea box (12x16x10 inch).
This system is in production since May 2017 and provides all of the power for my home office.





Pictures:

https://goo.gl/photos/ASZPmW9Ww5wAQVjX6

Source code:

1) 18650 Cell discharger: https://github.com/jens18/li_ion_cell_discharger
2) EPsolar Tracer4215BN and Tracer2210A modbus library (for 'golang'):
https://github.com/jens18/gotracer

Specifications:

Solar panel: Phono Solar 260W, 60 cells, efficiency 16%, Impp = 8.46A, Vmpp = 30.8V
Solar charge controller: EpSolar/EpEver Tracer2210A, 20A rated charge/discharge current, 'custom battery' configuration
Inverter: Microsolar INV-300PS (300W, 12V, 10.5V cutoff, USB)
Battery: 12V nominal, 3S36P (108 18650 cells), 2000 mAh average cell capacity, 72Ah, 864 Wh
Battery voltage sensor: Arduino Nano
Solar charge controller + batterymetric capture: Raspberry PI 3
Cloud monitoring: AWS ECS + Graphite

Very interesting! First haven't realized that the battery is under the the inverter and charge controller, neat little package. Also interesting to see that you use a 3S setup, don't you have issues with the inverter cutting off a bit early?

DarkRaven said:

Very interesting! First haven't realized that the battery is under the the inverter and charge controller, neat little package. Also interesting to see that you use a 3S setup, don't you have issues with the inverter cutting off a bit early?

Click to expand...

Yes, the battery makes upthe first layer. The second layer contains: Fuses, on/off swiches,
solar charge controller, solar charge controllermonitor display, inverter, additional 12Vusb adapters.

The only connections between the first and second layer are:
a) balance leads (JST-XH 3S)
b) main battery wires (XT60 connector)

18650 cells can easily be removed from their cell holders - no soldering required. I only use the cell pairs.

The daily voltage range is:

11.1V early in the morning (60% capacity) to 12.4V (95% capacity)

The inverter has never cut off because of low voltage (limit is10.5V). At the 10.5V level (under load)- my guess is
that the battery has only about5% of the max. capacity left.

-jens

You're giving your cells a really easy life as it seems. It is not a bad thing but you are using only a small part of their capacity. I think your guess is a bit off to be honest. 3S at ~5% is more like ~9.0V idle. 10.5V loaded could easily be 50%. That's why I was wondering and why many people don't use 3S or 4S as a 12V system. 3S is too low, 4S is too high.

But sure, it is fine if the usuable capacity is more than your daily need. You only need to remember it when your circumstances change.

Nice work m8!

Like it! using the 30A version of that charge controller as well as a 10A variant as well. Are you monitoring the individual series voltages or the pack as a whole?

Using grafana here for monitoring with some php scripts, would be cool to compare how you've achieved yours.

Chickey said:

Like it! using the 30A version of that charge controller as well as a 10A variant as well. Are you monitoring the individual series voltages or the pack as a whole?

Using grafana here for monitoring with some php scripts, would be cool to compare how you've achieved yours.

Click to expand...

I have build a small circuit with 3 voltage dividers to measure individual series voltages with an Arduino Nano.
The Arduino Nano is connected to a Raspberry PI 3 (in addition to the EpEver Tracer 2210A). Both inputs
(series voltage measurements + Tracer 2210A statistics (via 'gotracer' library)) areuploaded every 5 seconds
into Graphite running inside an AWS ECS hosted Docker instance.

I plan to use Grafana in the future to replace the more basic Graphite rendering.

If you have a rasp then just get the graphite/grafana into that. Grafana can use graphite as datasource so its easy to migrate