Jer's DIY Powerwall Build

Just Starting out:

Phase 1 - Collection / Solar
Hitting a few e-waste places and smaller computer shops. I have 1 solid connection for a small e-waste place, and 1 computer store. I'm working on 2 other computer stores.





1.1 - Sorting and charging

Built a 5V power supply from an old ATX PC power supply. added 8 220ohm 1/4 watt resistors and a 12v 10watt light I had kicking around to give me stable voltage on this. added some connectors so I can quickly connect and disconnect things.







Using those ebay charger boards powered by my new power supply, I kept it nice and neat. Next phase of this small project will be hooking up an arduino with DS18B20 under each cell. I'll probably just throw a RED or maybe an LCD that will display max temp so I don't have to come back each time and check if it got hot. I still have toorder those.





Reviving batteries that are more than 1.5v. I charge 4 at a time (match starting voltages) at 62.5mauntil it reaches ~3.0 -3.2volts (depending when I come back to check on it). bump it to 750ma for a while until 3.8-4.0 volts.When I return, I check temperature.Ifgood, Iremove them and put them into a ready for a normal charge bin.





1.2 - Solar requirements

I have a limited space for solar panels, and no south facing roof. they will have to be placed on the ground which I am surrounded by houses.so I need to figure out if there's a maximum output /panel based on location and when the house starts to shade the entire backyard, figure out maxpanels I can reasonably fit without the wife complaining,(thinking 4x 260 watt - have2 now)I'll build my battery to match theoretical capacity needed.

good luck with your project

Few realizations so far after playing with 2 solar panels, location , 2000w inverter:
1) I will not generate more from PV than I use at this time. I have a limited amount of space and the wife says ground mount is ugly and doesn't want it.
2) therefore I can only place panels more east than south and fit a max of 8 on the roof. West there are trees that will block most of the sun is it's probably not worth it. Phase 2 is adding risers so the panels will angle more SE.
3) if we build a deck, there will be a solar roof designed into it for more PV production.
4) My hydro meter metal dial appears to be moving backwards at times so I'm assuming there's no immediate need for a powerwall until they make me change my meter. (pushing power back into the grid.

My powerwall will be used as a backup battery for the house in the event of power failure. I'm assuming I will have to put a main switch before the panel so I can swap between the two. more research is required.

while not exciting powerwall updates. My first phase of solar is up:

4x260watt panels. they costed $1/watt after tax.

While the 18650 collection continues. I've started on a web enabledmonitoring system for the packs. I'll upload the data to a server and probably use Grafana.

16temp Sensors will be connected using a single data wire. 14 for the pack, one for the enclosureandone for reference temp. here I'm just testing 4 for now.




Nextwill be voltage Monitoring. I will build a test pack 8s1p and connect monitoring wires. I will scale up to 14s for the final version.

I'm using an ESP32 since it has the capability to connect to wifi and has enough ADCs to support 14S packs and still have pins left over for temp sensors and oled screens.

last phase is the balancing which I still need to figure out if I can dump multiple packs toto 1 or 2large resistors instead of a resistor for each cell. I'm sure I can but just haven't figured it out yet.

The good thing about this project is that it's much smaller than my other electronic projects so I might be able to finish this one within a year :-/

Looks like the same card reader as I have

Good start!

I'm thinking also in the way I will do that, the main problem with an ESP32 or similar is how to manage the readings and the same how do you balance, you can not connect at the same time to 2 batteries of the series too much voltage, so you have manage some switches to rotatethe readings, but for top balancing with that same system is a problem, you can onlydischarge one block at same time with only oneresistor, you must disconnect one to connect other.

Maybe a little more expensive and less elegant but one ES8266 in every block then you can control voltage individually and use a resistor for evey block to top balance.

If you want to do it all with one only device then you must use some dedicated chip for cell balancing as BQ76940 and control it with the ESP32.

Regards

If I can get the voltage readings sorted. I will most likely just use a few i2c multiplexer like this: http://www.ti.com/product/tca9548a/datasheet to just trigger a mosfet on a separate discharging circuit board that sits on each battery.

I'm thinking this separate board will have JST or similar connectors to make connections simple between this board and the ESP32.

A little more progress after a few long nights/early mornings. I'm not a programmer, so i'm sure it took longer than it should have.
I was able to test a voltage range from 9-12.6V (3S) with the use of voltage dividers and lots of math to provide better accuracy. it was close to 0.1v off at 12.6. The multimeter was reading 12.61 and I was getting 12.61 - 12.62.
I'll be calibrating for each Series pack (3.0v - 4.2v, 6.0v - 8.4v, 9.0v - 12.6v, etc.):
- Voltage Reference
- Voltage Divider Resistors
- Offset and Gain.

Voltage Calculation for each series will be: Cell4 (4S) -Cell 3 (3S) to get individual voltages.

Looking for some suggestions for pack design.
I was originally looking at a single MCU, but I'm now thinking of building each pack with its own smaller MCU and it would communicate via serial (might do wireless) to a central MCU that will handle reporting/wifi etc along with a main shutdown circuit, and I should be able to send commands to the remote MCU's for better balancing algorithm during if one pack is out by more than X volts.

Each pack would have:
- up to 8 temp sensors
- top balancing circuit (never go past 4.1v as an example)
- voltage meter
- shutdown circuit
- serial communication
-Optional oled screen to see Voltage

the advantage of this is when I parallel another set of packs, a single pack could shut it self off and everything would still function, just at a lower capacity

Sounds like the common BMS design out there. Either wireless or you need to go with optocoupled communication. Dont forget the issue by powering the mcu on each pack. You need one that handle low voltage properly.

[quote pid='9461' dateline='1500403982']
Sounds like the common BMS design out there. Either wireless or you need to go with optocoupled communication. Dont forget the issue by powering the mcu on each pack. You need one that handle low voltage properly.
[/quote]

I have a bunch of NRF24L01 chips just sitting here.

I was thinking 3.3v arduino nano's. power would come from the pack itself. packs won't be discharged past 3.2v which the nano should be able to handle just fine.


nano: $3.50
Temp Sensors: $0.80/ea x 8 = $6.40
NRF24: $1.50
Misc: $2.50
1amp resistor: $1
5v Fan: $4
Mosfet: $1.50
oled: $6.25
3d printed parts: $5
heat sink: $2.50
Cost per pack (1S32P)without cells: $34.15/pack

If I remove unnecessary items (reduce temp sensors, oled, heat sink): $20.60

Interesting. Why the 5v fan? and the heat sink?

I'd make the oled part of a plug in option rather part of the end product. along with a control panel of buttons for menu driving

Korishan said:

Interesting. Why the 5v fan? and the heat sink?

I'd make the oled part of a plug in option rather part of the end product. along with a control panel of buttons for menu driving

Click to expand...

just want to keep the discharge resistor cool enough

If this is for per series/pack level, then your discharge current should be really low as you are only burning off per cell to keep it balanced. So your current would only be around 500mA or less. Batrium uses the same thing to burn off excess to keep balance and they dont use fans. I think they actually use several discharge resistors to distribute the heat build up.

By doing it that way, you could save some more cost, and space, too.

Korishan said:

If this is for per series/pack level, then your discharge current should be really low as you are only burning off per cell to keep it balanced. So your current would only be around 500mA or less. Batrium uses the same thing to burn off excess to keep balance and they dont use fans. I think they actually use several discharge resistors to distribute the heat build up.

By doing it that way, you could save some more cost, and space, too.

Click to expand...

I could add multipleless powerful resistors. A1S32P should need at least an 1 amp (that's 31ma/cell)? If you were to balance a single cell, how many amps is a good amount to effectively balance. too little and the system might not work.

If you need to run constantly 31mA to balance a cell that cell need to be removed.

In general 1A for 100Ah is more than enough. Frankly you can do with 1A on 300Ah packs without issues. It will take time to do first balance yes but thats only the initial balance. On my 1500Ah bank i dont balance many seconds per day any longer

daromer said:

If you need to run constantly 31mA to balance a cell that cell need to be removed.

In general 1A for 100Ah is more than enough. Frankly you can do with 1A on 300Ah packs without issues. It will take time to do first balance yes but thats only the initial balance. On my 1500Ah bank i dont balance many seconds per day any longer

Click to expand...

OK. I'm not sure how much heat/power I should use so it's not hot?

4 x 15ohm resistors in parallel. that's .26amps x 4 at 1.1watts each.
8 x 30 ohm is about half so .13amps x 8 at .55watts each.

I'll probably buy 2 watt resistors.

A tip is to have a temperature sensor in middle of them so you can arrange the discharge based on temperature on the board. PWM is also something that can be used to arrange the discharge.

Another option, though a bit trickier, is to feed the excess power from one cell into the low value of another cell. But I dont know if you could do this in the same series setup, though.