Comment by rektide | Hacker News Reader

Comment by rektide | Hacker News Reader

Parent Full thread

Making heterogenous individual packs sounds like an absolute disaster. That's why this is various car companies teaming up with specific folk: because they'll have big quantities of cells from each individual vehicle they get, and there will only be so many different types of cell. There will be big supplies of a couple different cells, ready to go in to applications of all different sorts.

As it so happens, there's already a lot of agreement. Many many cells are either 18 mm, 21mm or 28 mm diameter cylinders & 650-700 mm long. "18650" cell is very common. 28650 used to be a "big" cell but it's not around as much I feel like. "2170" is a "21700" that is increasingly popular. Most of these cells have similar charge behavior, of applying a fixed amps up to a point, perhaps with temperature sensors to detect problems/slow down, and then a slower finishing charge, that ends around 4.2V per cell.

Using solar grid-tie micro-inverters would probably work fine to join together a bunch of different packs.

> Using solar grid-tie micro-inverters would probably work fine to join together a bunch of different packs.

No, they won't. The voltage range is all wrong for a full pack (microinverters work on one or two panels, so a normal working range of 30-80V, usually 50-60V per panel open circuit), and they tend to operate at their maximum power point.

You'd have far better luck with string inverters that can operate in that range (a typical residential string inverter will work from 200V to 600VDC which is just perfect for EV packs), and with a few modifications, you could get one that allows control over the output power (I expect some have that already, I'm just not sure which ones might).

> No, they won't. The voltage range is all wrong for a full pack (microinverters work on one or two panels, so a normal working range of 30-80V, usually 50-60V per panel open circuit), and they tend to operate at their maximum power point.

Actually, that sounds just about right, if you're working with modules rather than full packs. Tesla has various sizes of battery module, but 57 volts is one common size. (I think a typical model S has somewhere around 14 of them.) So, one microinverter per module would be fine. LG Chem also makes a module that's about 60 volts. (I think it's made up of around 16 pouch cells in series.)

Of course, if you're doing this on a large scale, you could just have any kind of inverter made that handles whatever voltage you want.

On the other hand, if you're using full packs you have the possibility of recycling motor controllers (also sometimes called inverters) from cars, which are designed to send 3-phase power to an AC motor, as a regular electrical system inverter running at fixed frequency. (Maybe fake the motor position sensor data so it always thinks the motor is spinning at a constant 3600 RPM, or 60HZ.)

I don't know if modern motor controllers output a real sine wave or if they just output a square wave. I'm guessing it's probably the former, otherwise I'd expect you'd get a lot of extra noise and vibration.

You could also re-use the car's original charger (the onboard part that converts 110V or 220V AC from a level1/level2 charger to DC at pack voltage). On the other hand, if the input power for charging is DC at the appropriate voltage to begin with, that part isn't needed.

my assessment is that 200-600VDC inverters would be far far far more economical & considerably more efficient since they're simple buck converters.

but there's no general market atm for such inverters. so building lots & lots of lower voltage packs & using low voltage solar targeting modules, which are everywhere, works out.

but it's for sure not as good & makes less sense.

repurposing motor controllers is an interesting notion. but I'd probably rather diy than try to circuit bend that. big awesome mosfets &c are cheap. motor controllers have to deal with inductive loads (big coils) so they tend to have slow slopes. which matches up ok for low 60Hz. but it's still be an effort to repurpose & reuse. just redoing the inverter seems like the win. don't even try to find out what design parameters don't line up between motor drive & inverter drive, don't spend the time, for something that won't work as well as having the right thing.

There's actually quite a market already for controller boards to allow people to use Tesla drive units (inverter/motor/differential setup) in non-Tesla vehicles. Not sure if such an approach would work if there are significant electrical differences between driving a motor and driving a utility backup power generator, but it's at least possible that it could work. Though on the other hand, for now Tesla drive units are in demand for vehicle conversions, so it would be kind of a waste to take the inverter out and to discard the motor and gearbox (which aren't useful without the inverter).

You're probably right that just engineering a specific solution or ordering them from a supplier that makes such things is probably less hassle and possibly cheaper in the long run.

I'm currently doing an EV conversion. I'm amazed that you can buy a box that's about ten or fifteen pounds or so for a couple thousand dollars that can handle about 90 kilowatts of power (in short bursts), converting around 100-200 volts to AC power to drive a motor. That's probably enough to power a whole residential block by itself, and yet it's not even a very powerful controller compared to what you find in a high-end OEM EV. I'd expect anyone building a large scale grid backup battery storage system to have access to inverter parts at least that good and at least that cheap, and if they don't they could afford to design it themselves.

I'm still trying to get over how awesome Google's 2014 Little Box Challenge was. Only a 2kW inverter, but one team (the Red Electronic Devils) managed a 14 cubic inch design. All the technical approaches are documented, as a part of the competition. A good number of extremely interesting ultra-high efficiency inverter designs!!

[1] https://ai.googleblog.com/2016/02/and-winner-of-1-million-li...

An 18650 is 65mm long. I think "eighteen six fifty" is more euphonic than "eighteen sixty five". A 2170 is 70mm long

And now, Tesla are developing an even bigger "4680" cell:

https://evannex.com/blogs/news/tesla-s-4680-cell-is-a-stroke...

A little too much fanboyism with that link.

Only Tesla uses cylindrical cells such as 18650s or 2170s. The other EV makers use prismatic cells or pouch cells.

I'd meant to mention pouch/prismatic cells, thanks. I had been under the impression that some of the big manufacturers had started using cylindrical cells, but I can't seem to find any evidence. Oops!

This is incorrect. Rivian also uses 2170s.

Reply on news.ycombinator.com