jonescg's battery builds (historical reference - image heavy)

jonescg

100 MW
Joined
Aug 7, 2009
Messages
4,213
Location
Perth, Western Australia
I've decided to put this thread together as my original thread was in the For Sale section, and it was quite reasonably deleted along with hundreds of other threads which were taking up space. So it makes sense to put the details here.

Table of contents:
My first electric motorbike battery - built by cell_man
Early attempts at Hi-power LiPo packs from individual cells
Soldering technique (if you must)
Multiple race bike battery packs (700 V battery packs)
BMS solutions (or problems)
The screw termination revelation
Large packs using screw terminations
The production line of e-moto projects
High voltage (400 V) bench battery packs
Electric Prelude battery with liquid cooling (and the risks)
Cylindrical cell packs with nickel coated aluminium
Cylindrical cell packs with copper busbars and nickel tabs.

(This table will be updated with links and the like as I go along, hopefully I don't forget to keep at it)
 
For some background, I've been building electric vehicles since about 2010, when I caught the bug riding an e-bike while living in Vancouver. I got home to Perth and decided that while I loved the e-bike, I really wanted to build an electric motorcycle, so I started with Voltron the dual Agni-powered smoke machine. The batteries for this bike were made by cell_man (aka Paul Lynch) and were a set of A123 20 Ah pouch cells built as 4s3p blocks. I had eight of these in series for the ~100 volt system.

https://endless-sphere.com/forums/viewtopic.php?f=10&t=12480&start=175#p328112

Paul did a great job, but a few issues I noticed with these packs was the buildup of capacitive charge on the outside of the enclosure (resulting in ~100 V tickles) and the fact they weren't particularly waterproof. Not that this mattered much as the bike was very much a fair weather bike. Another issue was the size format - 20 Ah cells mean battery packs came in either 20, 40 or 60 Ah sizes. A lower capacity cell would mean I could tailor the battery to suit the space available.
 
When it was clear I needed a faster, more reliable bike I went looking for more energy dense cells. At the time (2011 or so) the answer was hobby LiPo, but I hated the pissweak XT60 connectors and silicone 6 mm2 cables used to link them up. Putting multiple blocks in series and parallel was a spaghetti nightmare, so there had to be a more compact, safer way to do it. I came up with this.

4p 20C cells.jpg

74 V 17 Ah pack 002.jpg

The cell tabs would slide into the PCB with slots, fold over and be soldered together on the PCB above. Then I would place 2 mm copper busbars on top and hit them with 400 W of soldering iron heat to bed them in. I needed a sophisticated rig to make this work...

Trilobe battery 001.jpg

https://www.youtube.com/watch?v=PbKCw6r5wZQ&ab_channel=ChrisJones

As you can see from the video it's not an elegant solution, but I got it to work.
 
Terminating them was always a problem. I tried conventional bullet connectors:
6 mm bullets close up.jpg
6 mm bullets.jpg

But in the end concluded that bolt-on terminals were more functional. I soldered brass nuts to the undersides of the copper busbars before soldering these to the top of the battery pack:
YouFly packs 002.jpg

The BMS would typically sit on top of the soldered busbars with a layer of G10FR4 (glass fibre reinforced resin) for electrical insulation. This was one of three modules, making up a 6 kWh, 120 volt battery I built for the crazy Entecho flying machine:
Hoverpod Lipo Batts 002.jpg
 
This was a soldered battery pack designed to replace an e-bike battery.
10s LiPo Replace2.jpg

Just a standard 15 A BMS to each cell tap. As you can see I started to work on polycarbonate enclosures for the cells because the number 1 cause of grief with built packs like this was physical trauma. Cells would be punctured, bruised or would rub and fret against a protrusion and eventually breach the pouch.

Initially I used 3 mm acrylic and attempted to glue them together:
20C packs Phasor 003.jpg

But later found more reliable results with 4.5 mm polycarb and screws.
 
At this stage I was ready to build the 168s,3p battery for the race bike. I worked out I could fit 9 kWh in the space provided, so I continued to hone my soldering and stacking skills. You can read the racebike thread here, with many of the pictures you might want to view in that thread.
https://endless-sphere.com/forums/viewtopic.php?f=10&t=29916&start=200#p536954

But for now I'll include some of the most informative images on this thread specific to pouch cell battery pack building.
Same technique, but the rig I used to build the battery pack was a little more refined.
BMS installation 1 done.jpg
Final assembly sub pack in.jpg
Voltron Evo pack assembled.jpg

The polycarbonate enclosure was made from 4.5 mm and 3 mm sheets, and screwed together using M3 screws initially. I later discovered self-tapping stainless steel screws did a better job, and were far less work. Crucially, there was a physical barrier between each module so there was never a risk of one cell potential touching another, much higher or lower potential.
 
Battery management on these packs was always a pest. Because you needed to run a wire from each cell terminal, but the BMS needed to be reasonably secure, compact and waterproof. I had none of these things :)

For the race bike I was using the EV-Power Batt-Mon system, which at the time did no balancing, but simply alerted me if a cell was out of spec (above 4.2 V or below 2.9 V). By positioning the BMS on top of the cell blocks, it added about 12 mm to the overall height of the battery, which in the case of the race bike, was acceptable. Future designs might have the BMS remotely located on a different shelf, perhaps as a single PCB.

In all these cases, the BMS was simple - sound the alarm when something was wrong. It didn't cut you off while riding, but it would cut off the charger while charging. On the dash of the race bike it was a green LED and a red LED. Nice and simple :)

In February 2021 I pulled the last soldered race bike pack apart. It wasn't taking a charge, so there was a broken series path connector somewhere.
Feb 2020 pack.jpg
There it is.

This was the main problem with solder - it was hard work, it was hot and dangerous work, and it didn't always work. So along came the screw terminal idea.
 
I started messing with the idea when I was in the process of moving down to Albany, 2015. I soldered brass nuts to the underside of the PCBs, and hole-punched the cells at exactly the right spot so I could fold the tabs over and screw a copper busbar down to take the current. The PCB and copper were made to the dimensions which matched the series-parallel arrangement for the pack. I tested at very high currents and found the cells and the connections were not getting hotter than they should be, even at very high currents. So the system was worth following up on.

hole punch tabs.jpg
Stack the boards.jpg
align the holes.jpg
Screw down the busbars.jpg
10s3p load test.jpg

The process was born.
20170122_134435.jpg
Again, the main terminals would protrude from outside the polycarb enclosure. This was a battery pack for Bruno (Poweeer guy).
 
So from here I started to learn a few tricks. Number one - glue the cells in position. Initially I thought I could just leave them there without any glue, so if a cell fails I can just replace the cell or cell group. But the very act of leaving them in position without glue would cause them to rub, fret, chafe and eventually fail. So glue it was.

A full build of a battery for a guy who later turned out to be a shitcuntdouchebag who made off with lots of free stuff.
[youtube]LcLOT0m67Bo[/youtube]

Sometimes the number of cells in series and parallel had to fit in a very specific shape and size. So I'd get creative with busbars and battery management.
Bretts pack iso.jpg
Bretts pack top.jpg

In this case I had to run a busbar over the top of another one because of the three-row nature of the battery.
 
This was the beginning of Sketch Coleman's super moto battery:
22s6p boards.jpg
Central insulation.jpg
The cardboard divider down the middle was essential for adding rigidity and helping ensure the cells couldn't short on each other.
 
Kirby asked me to build a battery where the enclosure was made of aluminium instead of polycarbonate. This was a challenge, as the risk of shorting a terminal on the chassis was pretty high, and the enclosure was quite expensive to machine. Not to mention it weighted 2.5 kg versus polycarb which was closer to 1.5 kg.
Enclosure mass.jpg

It meant thermal conductivity was important, but so was electrical insulation. The holy grail of heat management in batteries :roll:
Kapton in.jpg

20180318_123410.jpg
The terminals on this were not easy. Bending copper more than it liked, and mounting them on a Delrin block so they couldn't short on the aluminium was a challenge.
 
Aaron also had a supermoto project in need of a 120 V, 30 Ah battery. It was a pretty compact battery, and the BMS on top was a reasonable location.

PCBs soldered.jpg

Aarons pack built.jpg

Aarons BMS lit up.jpg

The EV Power Bat-mon battery seemed to be pretty robust once set up, but the processor on the board is rather s l o w.
 
Stuart Jameson asked me to build a special battery for a unique project - salt lake racer at Lake Gardiner.
IMG_20190306_144129.jpg
IMG_20190306_134618.jpg
20190202_141146.jpg
One problem with these LiPo cell packs, is that you don't know the exact dimensions of the battery until its built. The cell supplier might assure me they are 7.8 mm thick, but they are more like 8.0 mm thick, and after stacking 40 of them together you have an enclosure which is a wee bit too small...
Bugger.jpg

He specifically asked me not to glue the cells in place as he wanted the option of taking them apart. I advised against it, but the idea was it would so a handful of runs on the salt and then not get used for ages, so it was worth the risk.
 
Paul Sarris built a pretty cool cruiser e-moto which used a LiPo battery as long as it was wide. Just as well it was a cruiser.
Pauls cruiser.jpg
Pack without BMS.jpg
Splash screen 1.jpg
Splash screen 2.jpg
 
James got me to build a battery using these large NMC cells. A bit tricky to get the hole spacing right when punching tabs, but I got there in the end.
20161112_143620.jpg
20161112_143655.jpg
Coulomb Motorsport pack.jpg

This powered a super efficient reverse trike which sadly missed the chance to get the efficiency record because... someone had double-booked the venue they were testing at! :eek:
 
Last edited:
The UWA Electric Motorsport team needed a high voltage battery to do dyno testing with, so I built them a 96s2p pack. It used an EV-Power Bat-Mon BMS, but unfortunately these only manage 48 cells, so I had to make two of them work together. Some clever use of solid state relays made it work in the end, but so far I'm pleased to report the battery is still doing good things 3 years on.

Nuts soldered on.jpg
PC enclosure built.jpg
EV Power BMS.jpg
Pack finished top.jpg
Pack finished front.jpg
 
I built this LiPo pack for Heath who was putting it into a small motorbike chassis (if I recall, it's been a while).
28s3p battery enclosure part guide.jpg

The large acrylic battery building jig made things a lot easier for keeping cell in alignment and ensuring a nice rigid pack.

Heaths pack under construction.jpg
Heaths pack under construction2.jpg
The EV-Power Bat-Mon BMS was pretty simple to wire up, but it wasn't exactly convenient for a space-constrained application like a motorbike. I assume he mounted the BMS equipment somewhere waterproof.
Wiring guide 2.jpg

15 Ah and 28s (120 V max).
 
This battery was the biggest I've ever built. It was the traction battery for the Honda Prelude I converted. The goal was to fit as much high energy density battery in the car without compromising seats or boot (trunk) space, or going too far over weight. We didn't know exactly how much battery that was, but the dimensions were constrained such that the 96s (360 V nominal) pack would need to be about 10p. I'd got some pretty impressive 7000 mAh cells from First Energy Power in China and they checked out for power and capacity. The task here was to build eight 12s10p modules which would fit where the fuel tank was. Unfortunately, only 6 would fit there, meaning two more had to go where the share wheel went.
9050135 (1).jpg
As these cells were expensive, and being LiCoO2, they were more prone to thermal degradation than other chemistries, so I engineered a cooling plate beneath each module. The plates would pass chilled water thanks to a heat exchanger on the air conditioning loop, and the cells were bonded to the chill plate using a thermally conductive epoxy resin.
Cooling plate final.jpg
2s10p block assmbly.jpg
Battery sans BMS.jpg
Prelude modules.jpg
potting screenshot.JPG

The complete battery pack was this, plus two more in the boot space.
20190929_205201.jpg
20191124_170203.jpg

I used the ZEVA BMS which was quite user friendly and very functional.
Full pack balancing.jpg
 
The cooling system worked a treat.
[youtube]LBiUo_5MDAc[/youtube]

Unfortunately, one of the modules leaked coolant up into the cells causing it to corrode, swell and fail. The BMS caught it like it was supposed to, and I carefully drove it back to my workshop where I found this:

[youtube]nh86_lf7v14[/youtube]

I was able to salvage 6 of the modules, plus a spare I had, which all went to a friend who has since employed them in a home storage project.
20200606_162607.jpg
 
So the next step was to build a replacement battery for the Prelude with the same capacity and voltage, but if it were to be liquid cooled, make that a lot more robust so it can't leak like last time.
The cheapest way to buy cells (at the time anyway) was cylindrical cells, and that meant 18650s. I found that the 18650BD cell was satisfactory, so I set about working out how to make robust, waterproof, self-contained battery modules using these cells. The maximum space available lead to a 96s23p arrangement.

cylindrical cells arrived.jpg
Made in Japan carefully scratched out with a permanent pen :lol:

cylindrical pack 3d.jpg
Machined polycarbonate capture plates were the winning formula.

cut bars2.jpg
I decided to use aluminium as it was cheaper, and the laser cutting was quick. Plus, it seemed you could get the aluminium nickel-plated, which meant I could spot-weld nickel tabs to them, and link these tabs to the cells below.

robot drool.jpg
The unique shapes are testament to the challenge of fitting an odd number of cells into a rectangular shape, 24 times...
 
I used aluminium and had them nickel plated using an electrolytic process. Not cheap, but it appeared to work. It leaves a dull grey metallic look, and if you put it under a microscope it probably looks like a stack of spheres. In any case, the layer of nickel meant there was enough resistance to generate heat during spot-welding to take a 0.07 mm nickel tab.
[youtube]jumFyhtjylM[/youtube]

Multiply that about 10,000 times...
[youtube]N6nKPmhSE5g[/youtube]

In the end, the poor spotwelder had done so much work those long electrodes were worn down to a nub, my hands were sore and burned, but I made 4 complete double-sided modules.

more welds.jpg
Nickel plated aly sucks.jpg
20210118_201035.jpg
20210201_174648.jpg
top side welded.jpg

The next step was to coat the welded faces with a thermally conductive epoxy resin and glue a sheet of G10FR4 to it on either side.
thermal epoxy going on.jpg
Module glued up.jpg

End result - a pretty much sealed, waterproof, rigid, robust battery module. The BMS wiring went to a pair of ZEVA BMS modules as 12s units plus four thermistors per module.
 
I ended up building the pack as per usual - fitting the ~24 kWh battery in the same space as the original 6 modules where the fuel tank went. It meant the spare tyre could go back in the boot.
New Pack built sml.jpg

All's well with this battery except one of the 24s23p modules had a bank of cells bombing out in winter. I steeled myself to build another module, so that come summer I could do the replacement. Last summer (21-22) was a doozy in Perth - hottest on record. So I took the car for a long drive to get a baseline on range. Bloody thing clocked up 130 km of mostly highway driving! No cells bombing out - what gives?

Then I realised. Thermal expansion coefficient of set epoxy resin is pretty significant. I had built the battery in summer, so come winter, the epoxy shrunk somewhat. Thanks to the often lousy contact the nickel tabs made with the nickel-plated aluminium, during the cooler months, they made poor contact and the effective capacity dropped. When it warmed up again, contact was made, and the capacity returned!

So the replacement module was made using copper busbars with soldered nickel tabs. These stuck substantially better than the spotwelded ones, and were actually faster to prepare...

Ni tab solder Cu busbar underside.jpg

24s23p module.jpg
 
Between finishing the Prelude's first battery and starting the second one, I was tasked with building a 16 kWh LiPo battery with absolutely mental discharge power. Damien and Phil had been working on a Tesla motor powered hill-climb car, and needed a battery with lots of beans for a short time. LiPo it was! This video introduces the car at the AEVA Expo in Sydney 2019:
[youtube]_pKmIWxJTXc[/youtube]

Preparing the PCBs was slow, but I had a bit of help from Skevos:
Nut soldering.jpg
Rack em up, stack em up:
Blockmaking.jpg
Stacks o LiPo.jpg
Test-fit of the busbars:
Busbar test fit.jpg
The polycarb enclosures were made ahead of time, as the width of the cells was more reliable than the thickness...
Phils modules.jpg
Enclosure termination.jpg

And it was completed:
Battery built.jpg
Battery built 2.jpg
Batrium BMS placement.jpg
The brass nuts were purely for fitting the BMS. They soon found out that the electrical interference of such high currents made this BMS pretty useless for an electric car pushing 2500 A.
 
My latest battery is this e-bike pack built from Molicell 4.2 Ah cells. I took the soldered nickel tab method from the second Prelude battery module and extended it to these 21700 format cells.
20220424_142633.jpg
20220403_152553.jpg
20220425_153539.jpg
Spotwelding1.jpg
balance wires.jpg
Spotwelding2.jpg
20220508_120218.jpg
Once the sides were glued on with the epoxy resin, I could pot the whole pack with polyurethane resin. This particular resin goes off much slower than the previous stuff I tried, but that's also because I made this when the ambient temperatures were around 16'C and not 36'C...
PU potted pack.jpg
It's a nice, compact battery which should have an easy life on the Gazelle.
battery on bike.jpg

Perhaps a quirk of the BMS I bought, it's only delivering about 12 of it's 16 Ah. A bit annoying but the charger seems to stop at about 56 volts, when it should be approaching 58.5 V. I guess it will have a long life.

I do really like this battery building technique, and I think I'll use it to build the battery for my Honda CRX conversion. Also, if I need more range and more power, I can always build a LiPo battery to sit over the front, assuming it can fit.

The Formula SAE teams were very interested in this technique too, but they have a specific rule against solder in the high current path of the accumulator. I can't think of a better way to do this though. Perhaps 2 mm diameter copper rivets with positive engagement on either side?
 

Attachments

  • capacity test.jpg
    capacity test.jpg
    264.3 KB · Views: 1,359
Back
Top