Most active commenters
  • oulipo2(6)
  • kragen(5)
  • jacquesm(4)

←back to thread

CT scans of 1k lithium-ion batteries show quality risks in inexpensive cells

(www.lumafield.com)
287 points jonbruner | 20 comments | 24 Sep 25 19:29 UTC | HN request time: 0.001s | source | bottom
1. oulipo2 ◴[26 Sep 25 20:49 UTC] No.45390884[source]▶
After we monitored many battery fires, we decided to build a casing which can sustain lithium fires for typical e-bike batteries: https://www.youtube.com/watch?v=v0NXXfCA2CY
replies(5): >>45390933 #>>45391201 #>>45391275 #>>45391338 #>>45391418 #
2. djmips ◴[26 Sep 25 20:54 UTC] No.45390933[source]▶
That's a great idea. Are you making a product or sharing the build?
replies(2): >>45392822 #>>45395425 #
3. kragen ◴[26 Sep 25 21:28 UTC] No.45391201[source]▶
I think you mean "contain". Very nice.
replies(1): >>45395399 #
4. jeffbee ◴[26 Sep 25 21:38 UTC] No.45391275[source]▶
Neat. How do bikes that have the battery entirely inside the metal downtube compare?

Edited: Now that I looked at your product I have low expectations for an unbiased answer :-|

replies(1): >>45397367 #
5. avar ◴[26 Sep 25 21:45 UTC] No.45391338[source]▶
Your site (if this is your YouTube video) indicates Bosch compatibility with gen 2 motors[A]. I'm curious as to:

1. It says "Protocol implemented = no", but "Powers bike = yes", you can power a Bosch gen 2 motor without implementing the protocol, presumably this is referring to the CAN wire on the battery.

2. Your battery kit is 316 EUR without cells, and your BMS spare part is 99 EUR. If you can talk to Bosch motors then presumably this is where that special sauce is.

I for one might be interested in just something that serves to talk with the Bosch motor, and allows one to provide any arbitrary 36v source, do you or anyone else make/sell such a thing?

E.g. for range extending a cargo bike I'd find the cost/weight ratio of a beefy ~1220 Wh AGM 12v car battery coupled with a boost converter much more appealing than something that emulates the form factor of a ~500 Wh Bosch battery with Li-Ion 18650 cells, the range more than makes up for the extra weight.

A. https://infinite-battery.com/pages/compatibility-guide-of-th...

replies(1): >>45395411 #
6. amelius ◴[26 Sep 25 21:57 UTC] No.45391418[source]▶
Maybe complement that with a CT-scan of your batteries and AI to detect defects.
replies(2): >>45391642 #>>45395418 #
7. jacquesm ◴[26 Sep 25 22:28 UTC] No.45391642[source]▶
FLIR monitoring + Ri measurement is a much better predictor than a CT scan, especially because you're not going to do that continuously.
replies(1): >>45393534 #
8. toomuchtodo ◴[27 Sep 25 02:22 UTC] No.45392822[source]▶
https://infinite-battery.com/

(no affiliation)

9. kragen ◴[27 Sep 25 06:12 UTC] No.45393534{3}[source]▶
Ri is internal resistance?
replies(1): >>45394220 #
10. jacquesm ◴[27 Sep 25 08:57 UTC] No.45394220{4}[source]▶
Yes. This is where the cell's current warms up the cells, you can picture Ri as in series with the cell current so the power it develops there is Ri * I * I. As Ri goes up the power goes up, the ratio between the load resistance and the pack internal resistance determines how large a fraction of the total gets dumped into the battery as heat. That's why you want to be careful with ensuring the batteries can shed that heat faster than they develop it, otherwise you might end up with a thermal runaway even if the pack wasn't strictly speaking broken.

Most half decent cell testers report Ri, and any deviation from the norm is a good reason to discard a cell. I test the cells of all the packs I build and I've had to throw out two out of a few hundred cells over time so this isn't common but it does happen (those two were Sanyo's, I normally buy Samsung but those were out of stock so in my book Samsung > Sanyo but my sample size is still small enough that it could have been a random issue, and they're all reclaimed cells so that may have been due to some cause that was beyond the manufacturers control).

replies(1): >>45396412 #
11. oulipo2 ◴[27 Sep 25 13:06 UTC] No.45395399[source]▶
Indeed! Thanks :)
12. oulipo2 ◴[27 Sep 25 13:07 UTC] No.45395411[source]▶
Yes! We're compatible with Gen 1/2/3 (but not "smart system" yet!) the battery is indeed 316 eur + about 40/50 eur for the cells (while a comparable Bosch battery would be 700/800 eur)

Perhaps you can send us an email at contact@gouach.com with more details on your setup, so I can transfer it to the technical team?

13. oulipo2 ◴[27 Sep 25 13:08 UTC] No.45395418[source]▶
Would be interesting indeed! For now we don't build our own cells (the most important part for defects), in order to let our users choose their own depending on their needs (18650, 21700, brand, second-life, etc)
14. oulipo2 ◴[27 Sep 25 13:09 UTC] No.45395425[source]▶
We're building a product, but that's easy to adapt to many setup (we have many end-users mounting our battery on their e-bikes, for now we are compatible with most systems, Bafang, Brose, Bosch, etc, working on Shimano now)
15. kragen ◴[27 Sep 25 15:14 UTC] No.45396412{5}[source]▶
Thanks! For whatever reason the battery datasheets I've been looking at this week called it IR. I'll definitely be including it in the next cell tester I build.
replies(1): >>45396715 #
16. jacquesm ◴[27 Sep 25 15:42 UTC] No.45396715{6}[source]▶
IR = Infrared?

R = symbol of resistance

i = internal

https://en.wikipedia.org/wiki/Internal_resistance

So Rint is also used. The proper way would be with a subscript but I don't know how to do that.

replies(1): >>45398752 #
17. oulipo2 ◴[27 Sep 25 16:43 UTC] No.45397367[source]▶
We're working on an internal downtube battery too! It's basically the same tech, the batteries work the same, it's just the form-factor which changes
18. kragen ◴[27 Sep 25 19:41 UTC] No.45398752{7}[source]▶
There's a Compose key binding for it in https://github.com/kragen/xcompose/blob/master/dotXCompose. Usually a lowercase i is the small-signal current, but some ambiguity is probably unavoidable with short abbreviations.
replies(1): >>45399270 #
19. jacquesm ◴[27 Sep 25 21:00 UTC] No.45399270{8}[source]▶
That's an interesting thing you've got there. If this were standardized I might actually take the time to learn some of them. My own language has a whole pile of diacritical marks that I never use, I just transliterate everything to ASCII and call it a day. This obviously is very broken but what with every editor and every OS having their own way of doing super/subscript and various composition method it never seemed worth the trouble. But some people can get very offended when you don't write their name properly. German has picked some sensible transliteration defaults now that allow you to write German without resorting to special letters. But in other languages they would be horrified by such a thing. And for things like superscripts, subscripts and so on it is even worse.
replies(1): >>45399357 #
20. kragen ◴[27 Sep 25 21:14 UTC] No.45399357{9}[source]▶
I didn't realize Dutch had a lot of diacritics! I often use Emacs's builtin input methods in Emacs, for example when writing in Spanish, but Compose works in Emacs too. Because I don't yet have any Wayland devices, Android is the only operating environment where I can't use that .Xcompose file.