My First 48V 6.5KWh Li-Ion Powerwall šŸ„°

Announcement - Help us fight the BOTS! Please report all spam including stuff in your inbox!

italianuser

Member
Joined
Feb 25, 2020
Messages
335
I am trying to PWM dim about 8A of light strips running off 24vdc nominal so about 29vdc max. I figured design it for 10A for some buffer. I think part of my problem is I was trying to just upgrade the MOSFET on a burned dimmer I have running off a 555 timer. I was able to read the full 28vdc on the gate which seams problematic and I don't know the switching frequency or gate charge capacity and was reading a significantly higher ohm resistance across the MOSFET when on.
Resistance increases with load, if I remember well. That's why for higher loads you'll see more than one MOSFET on the circuit, in parallel I suppose.
I found a nice paper with the formulas to calculate dissipated heat VS Rds (on), a good reason to choose MOSFETs with a low resistance; IRFZ44N has a lower resistance of IRF540 (I have these two models on my desk)
Now after some more learning about MOSFETS I am thinking it is time to build my own circuit to run of an ESP32 so I can add remote control. Of course that becomes a big project and I need to decide what I want to use all the extra pins on the ESP32 for :)
What about ESP8266 (for e.g. Node MCU V3 or Wemos D1), costs less and has less pins. I also bought some ESP12, even smaller and less available pins, but needs some wiring (and resistors) to be programmed.
 

italianuser

Member
Joined
Feb 25, 2020
Messages
335
August Update.

Got to final testing phase: cell buckets are ready, copper busbar, wire fuses and spot welding.

Goal is the always the same, happy I got it set right during first phase of the project:
4x 14s20p batteries, 280 cells each, 1120 total cells. Cells are second hand ICR chemistry, possibly 90% SOC, range 2000-2400mAh, IR range 30-70mOhm (yes, I know @Wolf it's a big range) and a small percentage of INR cells (with lower IR, too) to evenly distribute in packs.

Cell buckets
I have a load of boxes full of cells and labels... 1016 cells ready, 47 in rest period (10 days), 57 missing cells for 1120 target. I would like to have neat single-kind of nice plastic containers but, instead, I have dirty old cardboard recycled boxes. So I'm in a love-hate-love-hate-love condition ehm LOL. Just over 1000 cells selected and tested (some reshrinked, too) and I can do a good round-robin picking of the cells to make the first (of four) battery.
  • Buckets are divided in mAamps ranges and IR ranges: 2000-2099, 2100-2199, 2200-2299, 2300-2399. IR ranges 20-29, 30-39, 40-49, 50-59, 60-69.
  • Currently tested total capacity is 7706KWh.
Cell distribution
Bucket 2000-2099: 278 cells. 27.36% of total cells. 570Ah. 2109KWh.
Bucket 2100-2199: 371 cells. 36.52% of total cells. 761Ah. 2814KWh.
Bucket 2200-2299: 329 cells. 32.38% of total cells. 674Ah. 2495KWh.
Bucket 2300-2399: 38 cells. 3.74% of total cells. 78Ah. 288KWh.

Copper busbar
Oh my... I have stranded 16mm2 copper wire. After seeing those good videos from @hbpowerwall pulling solid copper wire nice and tight I realized my copper was stranded... hundreds of super thin copper wires to make the busbars. Ok, tried last night until early morning and got a decent result. Obviously I checked in the forum if anybody had used stranded wire for busbars (y)

rig.jpeg rig copper.jpeg

Spot welding
Yee it worked! I was really concerned about the results of my previous tests: it worked well on nickel strips but not on 35AWG fuse wires (small explosions with hot-micro-metal-balls flying everywhere -and welder's copper points ruined/burnt at every test).
I don't know what the problem really was so I just proceeded with making a setup board and et voilĆ  it worked. This will be really time-consuming until I find a fast way of moving around and solder fuse wires. I suppose it's a matter of experience.

spot weld.jpeg spot welder.jpeg iron.jpeg
I will be careful in pulling the stranded copper as tight as possible and without leaving any loose wires. In the photos there's my 74Ah car battery from my wife's old car and my 13ā‚¬ AliExpress spot welder (put in a black box). Ah, and my lovely new 150W welder, love it!

Next steps (make first battery):
  1. Make 15 busbars (2 small ones for terminals blocks and 13 double ones to connect each block with the next one);
  2. Round-robin pick cells from buckets to make 14 series and mount them in 4x5 cell holders; 7 modules will have + on top view, 7 will have - on top view;
  3. Mount busbars on holders and lock with plastic strips;
  4. Spot weld fuse wire on cells using painter tape to keep busbar in place (stranded wire tends to move more than solid copper).
There's a step I'm not sure about: checking each series capacity; that is a 20P block for which I already know the single cell capacity. Lets say I do check each series capacity: 44Ah, 43.1Ah, 44.2Ah, and so on. Does it really make sense measuring each series? Just for and extra control or what?
 

OffGridInTheCity

Active member
Joined
Dec 15, 2018
Messages
1,706
>Does it really make sense measuring each series? Just for and extra control or what?
Testing before building the battery is a perfectly rational thing to do.

On the other hand, I don't but that's because (in my process) I make adjustments after the BMS (Batrium) shows me the sagging packs after they are online.

Great progress, thank you for sharing :)
 

italianuser

Member
Joined
Feb 25, 2020
Messages
335
>Does it really make sense measuring each series? Just for and extra control or what?
Testing before building the battery is a perfectly rational thing to do.
Ok, it makes sense. Will cost some extra work because I must test the single packs before mounting the busbars.

I could make two ad-hoc connectors (a sort of panel to cover the 20 cells in the 5x4 holder) for positive and negative sides and do a C-D-C cycle using a 12V 20A PSU and a powerful enough step-down board. Uhm, I think I'll cut open dead cells and keep top part of the cell, put them in a 5x4 holder, solder them all together and make a specialized connector for testing purpose.

I'll make mAh measurement using an Arduino and the circuit+code I posted in this thread with a 0.1A accuracy. I didn't buy a high-amp lithium charger/tester because entry level ones looked very bad in reviews.

Thanks a lot!
 

Oberfail

New member
Joined
Jun 22, 2021
Messages
22
My suggestion for spotwelding fuses to 18650 cells is, to use proper fuses and not a fuse wire.
The fuse-wires are made out of lead (or at least i think so) which is pretty annoying to spotweld to certain kind of other metals.

I bought me 1000 3A fuses of that kind on aliexpress and i am yet searching for a material to spotweld them against as a busbar. You can however solder them easily to your copper wire busbar.
1629606155706.png
 

floydR

Member
Joined
Aug 23, 2017
Messages
1,273
The fuse wire used most here is tinned copper wire 30-35awg. glass fuses are great but awkward to work with, keeping the legs the same length can be a problem. Cell level fused nickel plated steel strips are handy.
Later floyd
 

floydR

Member
Joined
Aug 23, 2017
Messages
1,273
To blow at the same amp value the fuses need to be the same, the length might increase the amp when longer, decease when shorter. Perhaps I am mistaken wouldn't be the first time. Tinned copper wire is the same length or close to the same length when used as fuse wire. Cell level fused nickel plated steel the fused part pf the nickel is the same length for all the cells.

Later floyd
 

cak

Member
Joined
Mar 14, 2021
Messages
52
You can however solder them easily to your copper wire busbar.
Are there any tricks you have for spot welding to bus bars? I am using 4AWG solid coper wire as a bus bar for some packs and am trying to attach .15 nickel strips to the bus bar but the KWeld wasn't melting on well even all the way up at 200jouls. I would rather not have to solder it on but that is my current plan unless you have tips.
 

floydR

Member
Joined
Aug 23, 2017
Messages
1,273
Sharpen Kweld tips to a finer point? How wide of a nickel strip, perhaps try a narrower strip? or use tinned copper wire. Very small contract area of the wire while still providing ample current capability.
Later floyd
 

Oberfail

New member
Joined
Jun 22, 2021
Messages
22
Are there any tricks you have for spot welding to bus bars? I am using 4AWG solid coper wire as a bus bar for some packs and am trying to attach .15 nickel strips to the bus bar but the KWeld wasn't melting on well even all the way up at 200jouls. I would rather not have to solder it on but that is my current plan unless you have tips.
Not yet, I still have to find ways myself as spotwelding directly to copper is a bit hard.
 
Top