Comment by aetherspawn | Hacker News Reader

Comment by aetherspawn | Hacker News Reader

This is such a terrible idea because the packs have to be factory balanced before assembly, and everyday Joe doesn’t have the equipment (or probably the understanding) to do this part properly.

If the packs are not perfectly balanced, the batteries just short into each other and explode, and BMS can’t do anything because there isn’t any per-cell switch (cost).

It’s not just a matter of balancing voltage either, the cell profiles (voltage vs SOC) have to be the same otherwise you end up with 1 cell doing all the work. Simply put, when you mix and match cells of different brands, models, or even ages, they don’t integrate evenly. This results in a few or even just 1 cell doing a majority of the work during both charge and discharge, maybe 10x higher than its safety rating, guaranteed fire…

Also end-user is expected to do the math and input the battery’s total current rating into the motor controller? Yeah, nah, a hundred kids will think it’s cool to set this too high and set themselves and people around them on fire.

The cells which are in parallel will auto-balance when inserted. It's true that putting an empty cell and a full cell in, can indeed lead to a unextinguishble fire due to thermal runaway. But from the factory, cells will be very close already. They will auto balance once inserted (the parallell cells). Every single charger in existence will then balance the series cells actively on every charge cycle. So "terrible idea"? No, but maybe something to leave to the bike shop. Still better than throwing away the pack. And maybe fuses between every cell would help for protection against

(Gouach co-founder here) Exactly! The first kits are DIY for our early adopters, but we are working on an assembly pipeline in France, China and the US for our models in the coming months. Clients (B2C and B2B) will receive pre-assembled batteries

We are also looking for bike-shop owners who want to train on our batteries (it can be learned in one hour or so) to get a new revenue stream by offering to repair and assemble our batteries!

(Gouach co-founder here) Most of our users buy brand new cells, and we are working with partners who do have the equipment to retrieve second-life cells and match them in their factories, before selling them as "matched second-life cells packs" back to users for a huge discount :)

The Toyota Prius community already has a better solution for this - You can buy remanufactured batteries, then send your used battery back to get a credit. ChrisFix on YouTube has a swap guide for DIYers - it's mostly taking all the interior panels out of the back of the car.

(Gouach co-founder here) This is nice, but when you have an e-bike with a malfunctioning battery that you use for work everyday, you don't want to have the 2/3 months it generally takes for exchange programs.

Plus, once you sent your battery, they either discard it which is not ideal for the environment, or try to remanufacture it, which is very dangerous when it's done on a battery which wasn't designed for it, like ours

Seems an overly pessimistic take. TFA specifically mentions per-cell temperature monitoring, and I would assume there is also per-cell voltage monitors.

As long as the controller is made sufficiently conservative, there is no fundamental problem: you limit the current according to the cell that heats the fastest and shut down once one cell is near depletion.

Maybe they have even gone a more aggressive route and build a balancing circuit that can route significant current around a low-capacity cell. Or maybe just charging logic to keep as many cells as possible in the 20-80 regime if they will be limited by low-capacity members anyway. There are so many options here.

Sounds like a challenge: start monitoring individual temperatures, long term recording of performance down to the cell, recommend a mAh value of cell in a particular location, dynamic ohm chips in series with batteries to control differential discharging.

I am not enough of an expert to know if all this could work, but its got to be better than replacing an entire pack at a time to be worth trying

Cells in parallel will stay balanced and act as one big cell even if they're different capacities although they do need to be the same voltage first, but in series they need to be matched in capacity.

Slight mismatches in capacity can be OK in series too, as long as you have a BMS watching the cell voltages to make sure none get too low when discharging and none get too high when charging. The risk is that the lower capacity cells get too low or too high while the overall pack voltage seems fine, causing dendrites to form inside those cells, leading to a short and a scary fire. It really comes down to how well the BMS is designed / programmed and how much you want to trust it.

Most people aren't aware of how dangerous it is to try to "revive" a cell that got too low so this technology definitely comes with significant risks for user error.

It doesn’t magically morph into one big battery, suppose you have 5 different batteries in parallel, 1 new and 4 used different model, and you discharge this assembly at 50A: a large proportion of the current will flow through the lowest IR cell, say 45A of it. But the problem is that cell is only rated for 10A.

So yes, the net current is supposed to be okay, but it’s not, and rather than responding individually to all the other comments I may as well mention that no amount of temperature monitoring will fix this. Temperature sensors don’t detect internal cell stresses that cause spontaneous failures.

Unless they're wildly different chemistries, they'll have an IR that's inversely proportional to SoC and capacity, so you won't get that situation of 1 cell supplying 90% of the current unless it also has 90% of the capacity of the pack. Any cell that discharges a little more will have its IR increase, reducing its share of the current.

Do you honestly think they haven’t thought of that/tried it out?

Reply on news.ycombinator.com