New 18650 Pack design

glennswest

1 µW
Joined
Sep 11, 2017
Messages
3
I've got the bug. Want to buld a 20Kwh design.
Looking thru the threads on welding/soldering etc, and the china tech, yes
I could build packs easy enough.

I just want something better.

Here's what Im thinking:

48V + (14S Config Pack)

Build a couple of custom pcb (I have the tools to do this, and its cheap to fab).
1. Top board has a MC33771 bms chip, and a Wemos D1 Micro NodeJS Wifi controller plus battery terminals (Spring Loaded and static pins)
2. Bottom Board has spring and static pins)
3. All batteries have a proper fuse in between (Proper automotive fuses)
4. Power and Ground are fused for max power (With proper automotive fuses)
5. Pack issolation is FET controlled for both legs - Pack is disconnected by default
6. Microcontroller and fets are controllled from separate fused path from busbar.
7. A 60V in 100ma regulator powers the bms and mcu from busbar

I will integrate the battery mount in 3d print, so whole thing is nice and smooth and looks fully commercial
No soldering or welding needed to assemble.

Pack will screw directly into drilled and tapped busbar, (Thinking brass thumb screws maybe) With large surface area.

SO mount your bus bars on wall, as much as u want.
Screw in your packs.
Bring them up and charge and balance them.
Bring them online.

Any thoughts?

I will most likely use very heavy copper for pcb's and maybe nickel on top.

Everything Opensource of course.

Anyone have feedback? Like dislikes?
 
any particular cell in mind ?..
what pack config 14S, OK, .. but 120+P in one unit ??
or smaller parallel blocks ?
what output current peak are you intending to use.
I guess this will be a fun project for you, but it would be much simpler and probably not much more costly to just buy 4 Tesla modules.
 
The concept is to place bus bars in a pattern that control spacing of the unit.

So need to work out the math, but may 8 down for a 6 foot wall with spacing at top and bottom. The bus bars at the top I'd put a contactor as well, so a column could be issolated.
Then you could put the width of these across, so you can build very large systems.

Everything is digitally controlled and monitored.

Cells are something like this:
https://www.aliexpress.com/item/YCDC-4pcs-lot-Original-18650-Li-ion-3000-mAh-3-7V-Lithium-Battery-18650-Rechargeable-Battery/32808008215.html?spm=a2g0s.13010208.99999999.271.pFFx8p

18650 3000mah (target is $2 plus a cell.

Im really good at sourcing, so considering Im buying in wholesale volumes, this works out to huge cost saving compared to a 4 pack of tesla size.

The other lovely part is everything is pretty reusable and serviceable.

Engineering cost is zero (Im software and hardware).

would look at stackable inverters etc to make it very scalable.

The overall management could be something as simple as a rasberry pi, or even a cloud vm.

The control mcu are designed to start up with just bus power, so even initial charge of battery etc can be controlled.

Its so cheap to get pcb made, and I was looking at the price of bms, its quite high, so using at chip level is cool.

I've got access to both a low volume assembly house and a smt oven.

This would make a DIY powerwall so much better.

And its really scalable.

3 x powerwall is $17,200 + installation.
 
You have not said, but i guess this is for an off grid storage battery ?
If you are good at sourcing, you should recognise a con when you see one.
If not, then i suggest you brush up on cell specs and quality before you buy many of those !
At the very minimum , get some samples and test them fully.
 
Oh yes, I seen every variation of fraud over the years.
I am just getting a hand full to start and test.
I may even pay more in the end, but there plenty of suppliers.
And with the metrics etc in the packs Im planning, not much place to hide.
 
Dont know how you spot fraud bur the cells you list is right of the bat one. No 3000mah cell weighs 36g and has 1000 cycles. You sound like its easy, bms this, wifi that, fuse here, conractor there but its not. Teams develop bms'es let alone complete solutions. I would go on this - dreaming but wish you good luck
 
Back
Top