Roadster battery pack specs?

[iisjsmith]

Does anyone know the capacity of the Roadster's ESS?

We know it is made up of 6831 18650-form-factor batteries, but there are different types and nowhere on the website does it give the specs for a single cell. I mean, an HR18650 cell can be 1300-1400mAh whereas a LR18650 can range from 1800-2200mAh.

Also, it doesn't give any info as to how many of the 6831 cells are linked in series versus in parallel. I mean all 6831 3.7V cells can't be linked in series, otherwise you would have a 25275V pack!!

I'd like to know the specs on a single cell...and the total capacity of the whole ESS.

[WarpedOne]

If you study all TM material you learn that battery pack:
- stores about 50kWh of electric energy
- operates at nominal 375V
- consist of 6831 batteris that are organized into eleven separate modules with its own control electronics

So much from top of my head. For some more, dig into the site It has tons of info, but it is quite widespread.

[iisjsmith]

Yea, I got all that info a while back. I also know it weighs 450kg and will deliver up to 190kW of power.

What I'd like to find out is the capacity of the entire pack, the model number and/or specs of the individual cells, and how those cells are wired together (how many serial and how many parallel). Because I keep running the numbers through and I cannot figure out how they use 6831 cells at 3.7V each and only end up with 375V. That's only 102 cells linked in series. So they either linked a whole mess of batteries in parallel or the cells they are using aren't 3.7V each.

I'm just putting this out there, in case anyone comes across this info.

[Embassy]

The battery pack is almost certainly 99 cells in series by 69 parallel strings - it's really the only way to make the numbers work out. If you factor 6831 you get 3 x 3 x 3 x 11 x 23, which prettty much tells you how they designed it alongside the 375 nominal volts figure. The difference between 102 cells and 99 in series is probably the difference between 3.7 volts per cell and 3.8, which is reasonable for "just off the charger" numbers.

Here's the part I don't get. To justify a 50 kWh battery pack energy, and a 200 kW peak power, you need to have batteries rated for 4C discharge. You can calculate this for anyone's EV design by dividing the maximum power by the total energy. If the voltage is a nominal 375 for the pack the capacity of each cell has to be about 1950 mAh per cell. That's a good "real world" figure for a nominal 2000 mAh cell.

I don't see a battery like that on the market anywhere outside of Lithium Polymer cells. The 18650 cells I can find in the 2000 mAh class are rated for no more than 1.5 C discharge.

So what cell specifically is Tesla really using?

[dsacks6]

Welcome Embassy and thanks for the info.

[Embassy]

Thanks, Greenspeed.. it's really just reverse engineering, not any actual "info" per se.

[iisjsmith]

So if each cell is 2000mAh and 3.8V we get a battery pack with a 138Ah capacity and 376.2V. That works out to 51.9156kWh. Pretty close to the numbers Tesla Motors gives us.

But what about the fact that they say the ESS is made up of 11 sectors of 621 cells each? I guess you could just arrange them that way physically and then wire them 99S69P. But that seems awfully complicated. I'd think you'd want to keep the 11 sectors distinct, so you minimize the connections going from sector to sector.

I have seen 18650 cells with 10C discharge rates, but they have lower capacity and voltage levels. For instance, on the BiPowerUSA site (www.bipowerusa.com/products/li-ion.asp) they have a cylindrical cell called the HR18650. It's got a 1300-1400mAh capacity and 3.6V, and it can discharge up to 10C, while the normal 3.7V 2000mAh cells can only discharge up to 2C. So you are getting half the capacity at the same weight but with 5x the C rate.

I have also seen a pack called the Flion 1200, used for RC applications. It also has a 10C discharge rate, and has a 1200mAh capacity and 3.7V.

So if Tesla is using this kind of high discharge cell, our numbers suddenly look a lot different. Now if we have 69 parallel strings we are only looking at 82.8Ah capacity. We would need 115 parallel strings to get back to the 138Ah capacity, and that doesn't leave us enough cells to make the 375V.

I'm going to try another plea to Tesla Motors to let us know what cell they are using. I mean, it can't be THAT secret. The value of their product is the engineering behind the whole ESS...not just what cell they use.

[Embassy]

Well, I have a hypothesis that seems to fit all the facts we know about .. try this on:

The application is a 4C discharge application, we know the pack is about 375 volts and we know it's 6831 cells. Clearly that's 99 series by 69 parallel. But the 10C cells you can get in the 18650 format have too little energy, and the 2C cells have too little discharge rate.

I think they're using both kinds in parallel. Specifically, I think there are 47 strings of 99 2300 mah cells that have 2C discharge (such as the Sanyo 18650F), and 22 strings of 99 1500 mah cells that have 10C discharge characteristics such as the Sanyo 18650W. That stack gives you 536 Amps at maximum, about 52 kWh of total energy, 141.9 amp hours of capacity. I'm guessing each cell is probably two or three bucks, giving a total of 20K worth of cells in the battery pack.

Building a mixed-cell battery pack like this would be a significant challenge but the benefit is the ability to tailor capacity and discharge rate to the requirements accurately, and to use commodity cells. I guess if it were me I'd approach the problem with a current limiter on each string - a 4.6 amp limiter on the low rate strings and a 15 amp limiter on the high rate strings. That way, you can always draw the pack down at 1C or 2C.. but when you step on the amps you get an extra surge from the high rate strings.

If this is what Tesla is doing, then they have a right to be very proud of a tricky bit of engineering. Keeping a pack like this balanced and charging it safely is a very impressive achievement.

If that's actually what they're doing. I'm not sure I see another way to go about it, frankly.

[malcolm]

I think we have a problem in assuming that the cells are charged to their nominal voltage values.* Take a look at the bar chart/graph just over half way down this page:-

http://www.mpoweruk.com/bms.htm

Typical Hybrids only charge to 80% in order to leave enough headroom for Regen braking.

Under-running the batteries on pure EVs may have further benefits in terms of heat extraction and charge balancing.

The trouble is, we have no idea how much headroom (if any) Tesla have built into their battery system.*

[Embassy]

Malcolm, I don't think a pure EV needs regen headroom the same way a hybrid does. The regen energy comes out of the battery pack in the first place rather than from the gas tank as is the case with a hybrid. And once it comes out of a charged battery pack then it has plenty of headroom to go back in, given the inevitable losses. I suppose there's a "charge up at the top of a big hill" scenario that might change matters but if that were the case I would just make sure NOT to regen and switch to wheel brakes for speed control on the downhill stretch - you can always set the inverter to freewheel the motor, after all.

Also, I think that mid range charge cycling is more of a NiMH thing than an Li-ion thing. I have a Prius and I'm fairly sure that it cycles through the middle 200-300 Wh in a 1200 Wh battery for lifetime management issues.