DIY Powerwall Concept

macskyver

New Member
Joined
May 27, 2021
Messages
8
Hello All,

I am looking into embarking on a DIY powerwall project and want to get some opinions/input from people with experience/knowledge. The rough outline of what I have in my head at the moment is the following:
  • PCB based
  • 48V system using LiFePO4 cells in 16s1or2p, or lithium ion 14s1or2p, leaning toward 26650 16s2p LiFePO4
  • each board with built in BMS using something like this IC: BQ769142PFBR
  • temperature for each cell using something like this IC: TMP1075DSGR
  • using a low power consumption microcontroller like this: CC2640R2LRHBR with zigbee or similar communication or using a simpler arduino compatible microcontroller that you can check/configure with a USB connection to a computer.
  • onboard current sensing/power monitoring per module
  • onboard configurable current limiting for charge/discharge
Basically, each board would be a self contained module. I envision them sliding individually into something like a server rack then having one molex style power connector that goes to busbars. You would be able to unplug and remove a module without taking the system down. You could also add capacity really easily. With 16s2p LiFePO4, each module would be close to 400wh so it wouldn't take too many modules to make a useful system.

I am a little concerned about the connection resistance of the snap in battery holders, but since it will more than likely use fairly low C rates, that may not be a big issue. Also, I am thinking the price might not be super cheap. I am ballparking $100-150 per module with cells. The features/cool factor/fun factor might be worth it though. I do like the idea of being able to identify/replace misbehaving cells really easily. I see these giant battery packs online with tons of cells in parallel with at best one temperature sensor for the group. Good luck identifying a bad cell in the mix and have fun replacing it.

Well... enough babbling. Let me know if you guys have any ideas or suggestions. Thanks.
 

snoobler

Solar Honey Badger
Joined
Jul 10, 2020
Messages
8,430
Location
HBR, AZ
Welcome to the forum.

Personally, sound like high effort, low payoff; however, I like the concept.

Definitely LFP.

Anderson for the plugs.
 

macskyver

New Member
Joined
May 27, 2021
Messages
8
Welcome to the forum.

Personally, sound like high effort, low payoff; however, I like the concept.

Definitely LFP.

Anderson for the plugs.
You might be right about the high effort low payoff... Why Anderson plugs?
 

macskyver

New Member
Joined
May 27, 2021
Messages
8
Don't like the idea of Molex in terms of current carrying capacity.
Each module would be max 2p, so relatively low current (probably <20A max). I was looking at the Molex Super Sabre series connectors which have 30A rating, but after looking into it a bit, I am liking the Anderson Powerpole 15/45 better. It looks like a winner.
 

Mainstream818

New Member
Joined
Jun 27, 2021
Messages
12
Hello All,

I am looking into embarking on a DIY powerwall project and want to get some opinions/input from people with experience/knowledge. The rough outline of what I have in my head at the moment is the following:
  • PCB based
  • 48V system using LiFePO4 cells in 16s1or2p, or lithium ion 14s1or2p, leaning toward 26650 16s2p LiFePO4
  • each board with built in BMS using something like this IC: BQ769142PFBR
  • temperature for each cell using something like this IC: TMP1075DSGR
  • using a low power consumption microcontroller like this: CC2640R2LRHBR with zigbee or similar communication or using a simpler arduino compatible microcontroller that you can check/configure with a USB connection to a computer.
  • onboard current sensing/power monitoring per module
  • onboard configurable current limiting for charge/discharge
Basically, each board would be a self contained module. I envision them sliding individually into something like a server rack then having one molex style power connector that goes to busbars. You would be able to unplug and remove a module without taking the system down. You could also add capacity really easily. With 16s2p LiFePO4, each module would be close to 400wh so it wouldn't take too many modules to make a useful system.

I am a little concerned about the connection resistance of the snap in battery holders, but since it will more than likely use fairly low C rates, that may not be a big issue. Also, I am thinking the price might not be super cheap. I am ballparking $100-150 per module with cells. The features/cool factor/fun factor might be worth it though. I do like the idea of being able to identify/replace misbehaving cells really easily. I see these giant battery packs online with tons of cells in parallel with at best one temperature sensor for the group. Good luck identifying a bad cell in the mix and have fun replacing it.

Well... enough babbling. Let me know if you guys have any ideas or suggestions. Thanks.
I like that idea. I'm actually in the process of releasing my custom bms to GitHub in which I designed it to be self sustaining per 8S#P module and you can increase your capacity over time and disconnect modules without effecting the system. My current design is is limited to 30-40A only because of thermal. I'm reworking it to include a better heatsink and handle 100A at 8S.

Using an ideal diode was the key for me to prevent other battery packs from charging the other packs when the protection mosfet is triggered.

Here's the link to the ideal diode I created.
 

bladetoyshop

Solar Enthusiast
Joined
Jan 24, 2020
Messages
37
Each module would be max 2p, so relatively low current (probably <20A max). I was looking at the Molex Super Sabre series connectors which have 30A rating, but after looking into it a bit, I am liking the Anderson Powerpole 15/45 better. It looks like a winner.

Anderson Powerpoles are an excellent idea; I make heavy use of them when I build cables for use with my Amateur Radio gear.
 

macskyver

New Member
Joined
May 27, 2021
Messages
8
I like that idea. I'm actually in the process of releasing my custom bms to GitHub in which I designed it to be self sustaining per 8S#P module and you can increase your capacity over time and disconnect modules without effecting the system. My current design is is limited to 30-40A only because of thermal. I'm reworking it to include a better heatsink and handle 100A at 8S.

Using an ideal diode was the key for me to prevent other battery packs from charging the other packs when the protection mosfet is triggered.

Here's the link to the ideal diode I created.
Nice! I would be interested in seeing your BMS. Without the ideal diode, wouldn't the modules just balance each other out and then track together? Or do you have some charge/discharge intelligence that allows the parallel modules to be at different voltages?
 

Mainstream818

New Member
Joined
Jun 27, 2021
Messages
12
I'm hoping to post all the files within the next couple of weeks. So right now I have a discharge path and charge path so that if I over voltage, the charge disconnects and it doesn't effect my loads. And vise versa for the discharge path. Now if one of the modules let's say over voltages because of over charging, without the ideal diode then that battery will still continue to charge that cell via the body diode of the mosfet so then you will really have a bad day if your pack isn't completely balanced. I've already had instances where one of my cells in my pack triggered cell level over voltage and when I didn't have the diode it still charge past 3.7V almost to 4V so not a good day
 

macskyver

New Member
Joined
May 27, 2021
Messages
8
I decided to tweak my plans a bit. I am building a more traditional battery bank using 60180 cells. It will be 16s2p 110ah. I am still thinking of rolling my own BMS, but to get things up and running quickly, I got a Daly BMS ordered. I am now trying to decide the best BMS configuration. I like the idea of it basically being a cell monitor/balancer with configurable outputs that can drive relays/solenoids/or even directly hook up to an inverter switch.. ie no high currents on board.
 

Mainstream818

New Member
Joined
Jun 27, 2021
Messages
12
Oh ok nice. The cell monitor is pretty straight forward as its just a matter of getting two 8 channel adc or four 4 channel adcs. The balancer is slightly more complex unless you find an integrated solution which I believe Texas instruments has a couple options
 

macskyver

New Member
Joined
May 27, 2021
Messages
8
Exactly. I was looking at BQ769142PFBR, which has all the voltage measurements and passive balancing control for 16s.. and it can talk to a microcontroller with I2C...
 

Mainstream818

New Member
Joined
Jun 27, 2021
Messages
12
Oh sweet. In case you're interested, I started to upload my Arduino based bms that I am calling the mBMS or modular bms. I don't have everything completely uploaded but I'm in the process of doing so but I have enough so people can have an idea of my goals with the project.

mBMS open source
 

macskyver

New Member
Joined
May 27, 2021
Messages
8
That's a really cool project you have going on. I like it.. How well does the hall effect current sensor work? Also, how much power does your system consume when idle?
 

Mainstream818

New Member
Joined
Jun 27, 2021
Messages
12
The current sensor I am using is the allegro acs780 which can handle 100A continuous and 150A peak.

I have it coupled with a 16 bit adc and the sensitivity of the sensor is 13mV/A so i have high confidence of 1A resolution accuracy.

I have not measured the quiescent power but I can estimate worst case it'll draw around 3-5W. That's assuming the buck converter on board is pushing the full 0.6A which it's not but that should give you an idea
 

derekisastro

Solar Enthusiast
Joined
Jan 27, 2021
Messages
47
No disrespect intended, have you looked into the BIYBMS v4? Seems like you're re-inventing the wheel to some extent? No?
 

Mainstream818

New Member
Joined
Jun 27, 2021
Messages
12
Do you mean Stuart diyBMS?
No worries I don't get offended when questions are asked.
Yes and no. I've been developing this system for about 2 1/2 years on and off since I do this on the side.
My system is intended to work with fully built standalone packs so for example an 8S100P pack. Or another way of looking at it an open source version of a Daly bms.
The diyBMS requires a BMS.module per 1S100P pack like an open source version of batrium.

All great BMS systems don't get me wrong but I built mine for a specific need and feel it has some advantages over the rest.
That's why I posted the project on GitHub ( still need to finish) and if anyone was interested they can build it. Not planning on selling it anytime soon at least not this version. Currently working on the next version with more Horsepower
 

macskyver

New Member
Joined
May 27, 2021
Messages
8
No disrespect intended, have you looked into the BIYBMS v4? Seems like you're re-inventing the wheel to some extent? No?
I've looked at and considered (and am still considering) that one. It's a valid option but not quite what I was looking for. My game plan has been evolving though, so I may revisit it at some point. It seems like it would be a pain to expand upon as you would probably have to redo the packs when adding parallel capacity.
 

Mainstream818

New Member
Joined
Jun 27, 2021
Messages
12
I've looked at and considered (and am still considering) that one. It's a valid option but not quite what I was looking for. My game plan has been evolving though, so I may revisit it at some point. It seems like it would be a pain to expand upon as you would probably have to redo the packs when adding parallel capacity.
Yea that's one of the downsides to the diyBMs and batrium alike. I designed my system to have essentially an infinity parallel upgrade capacity when needed. I currently have 4 24V packs each have a slightly different AH but they are individually protected and can reset faults on each one without having to turn off the system
 
Top