-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
So DaveD's graph looks like it's leveling out at about 202 CAC after 6 months. That would support what the MS folks are seeing, with an initial drop then stability for many years. Bolosky's graph is not quite so clear in this regard, but it's only over 4 months.
I was personally just about to pull the trigger and order the 3.0 battery, but this has me a bit skittish... What is Tesla's warranty on the new 3.0 battery? If it's 8 years / 100k miles, like the original, it's probably worth the risk. If only 1 year / small miles, perhaps not?
Dave's chart certainly makes me feel better. We'll see in a few months if mine is also levelling out. I'd be perfectly fine with sitting at a CAC of ~200 for a long time.
I also found the datasheet for the cells that I think are the ones in the 3.0 battery. It claims that the spec is that the cells should have >= 70% capacity after 300 10A cycles. Figure a full cycle is 250-300 miles, so 300 of those is 75K-90K miles. The curve for my cells is much steeper than that; I've lost 6% in 6K miles, which also is consistent with an initial dropoff followed by a flat/slowly degrading period. Plus, you'd expect that the Roadster use (not full cycles, nowhere near 10A most of the time) would be much more gentle on them.
https://www.imrbatteries.com/content/lg_HG2.pdf
I also found the datasheet for the cells that I think are the ones in the 3.0 battery. It claims that the spec is that the cells should have >= 70% capacity after 300 10A cycles. Figure a full cycle is 250-300 miles, so 300 of those is 75K-90K miles. The curve for my cells is much steeper than that; I've lost 6% in 6K miles, which also is consistent with an initial dropoff followed by a flat/slowly degrading period. Plus, you'd expect that the Roadster use (not full cycles, nowhere near 10A most of the time) would be much more gentle on them.
https://www.imrbatteries.com/content/lg_HG2.pdf
This type of little kink at the start of the battery life is quite common for all lithium chemistry cells. I've seen it on datasheets for a long time including IIRC the original Roadster cells.
Anyway, I came across this. Scroll down to half way. What is the difference between the LG HE2 and LG HE4? Which is newer, better?
Anyway, I came across this. Scroll down to half way. What is the difference between the LG HE2 and LG HE4? Which is newer, better?
I forget; do we know what the 3.0 cell chemistry / pedigree is? I thought there was some confusion on that, other than they are not what the original cells are, not what the MS/X uses, and not what the M3 will use. So, what are they, and what is their lifetime capacity curve supposed to look like?
Ideally we should be able to take the original battery chemistry's curve and compare it for accuracy with the expected (published) curves. Given a match there (process sanity check), then take the curve for the 3.0 cells, and extrapolate what we should see. Then (pant, pant), we can determine if the CAC drops that folks are experiencing is expected, and where things should level out (assuming they should). I'm not qualified to do this... Any takers?
I looked at Dave's chart a little more closely, and I am no longer comforted.
I think that all of the flattening of the CAC curve is due to his driving less. It looks like the CAC dropped from 215 to 207 when he drove about 4000 miles, from 110K-114K, for a slope of -2/1K miles. After that, he drove another 2K miles and the CAC dropped from 207 to a little more than 202, or something around -2.5/1K miles. That is, the drop off is pretty much linear in miles driven.
Looking at the curve in the paper that dpeilow posted, it looks like the knee in the curve is around 25 cycles. If we think that a full cycle is 250-300 miles, that's 6K-7.5K miles. Dave's at about 5K miles on his battery, and I'm around 6K, so we should just be hitting it, more-or-less. And the amount of loss at the knee is about 10%, while I've "only" seen 6% so far on mine. So, maybe we still haven't gotten there yet. (Plus, that curve is for 20A discharge, which is almost certainly harder on the battery than what we're doing, which I figure is no worse than about 8A during a full-on acceleration, and <1A in cruise).
I'll keep monitoring and update you. Also, if anyone else would be willing to share their graph, that would be really helpful.
I think that all of the flattening of the CAC curve is due to his driving less. It looks like the CAC dropped from 215 to 207 when he drove about 4000 miles, from 110K-114K, for a slope of -2/1K miles. After that, he drove another 2K miles and the CAC dropped from 207 to a little more than 202, or something around -2.5/1K miles. That is, the drop off is pretty much linear in miles driven.
Looking at the curve in the paper that dpeilow posted, it looks like the knee in the curve is around 25 cycles. If we think that a full cycle is 250-300 miles, that's 6K-7.5K miles. Dave's at about 5K miles on his battery, and I'm around 6K, so we should just be hitting it, more-or-less. And the amount of loss at the knee is about 10%, while I've "only" seen 6% so far on mine. So, maybe we still haven't gotten there yet. (Plus, that curve is for 20A discharge, which is almost certainly harder on the battery than what we're doing, which I figure is no worse than about 8A during a full-on acceleration, and <1A in cruise).
I'll keep monitoring and update you. Also, if anyone else would be willing to share their graph, that would be really helpful.
Do any other EVs (cars) use this cell? Just curious if there is another experience base we can draw from. They seem to be popular (according to a very quick Google search) in electric bikes, but I don't think that's a good compare.
Someone stated these are 3Ah HG218650s in the 3.0. If true, this is an interesting call by tesla.
This battery (based on literature from LG on the net) is an NMC positive (cells have gone from LCO (roadster) --> NCA (S) --> NMC (3.0)) NMC has been quite well tested in the world, with great cycle life, much depends on the ratio of the Ni/Mn/Co used. What is special about these cells is that besides
graphite as the negative electrode there is also SiOx mixed in for added capacity. This is an "advanced" chemistry cell with high energy density. I'm not sure if Tesla switched the S to these yet,,
Many are starting to adopt this chemistry for the negative. I wouldn't worry about the front end loss, your capacity decay slope should level off. These cells seem to have a natural loss upfront then become quite stable. Keep an eye on it for a year.
My feeling is that the data from these cells will be watched closely by Tesla. If they aren't then they should. It is important
to get a feel for how these behave and how much better they are relative to NCA cells in S, and the benchmark LCO in 1.5-2.5.
Please keep us posted.
This battery (based on literature from LG on the net) is an NMC positive (cells have gone from LCO (roadster) --> NCA (S) --> NMC (3.0)) NMC has been quite well tested in the world, with great cycle life, much depends on the ratio of the Ni/Mn/Co used. What is special about these cells is that besides
graphite as the negative electrode there is also SiOx mixed in for added capacity. This is an "advanced" chemistry cell with high energy density. I'm not sure if Tesla switched the S to these yet,,
Many are starting to adopt this chemistry for the negative. I wouldn't worry about the front end loss, your capacity decay slope should level off. These cells seem to have a natural loss upfront then become quite stable. Keep an eye on it for a year.
My feeling is that the data from these cells will be watched closely by Tesla. If they aren't then they should. It is important
to get a feel for how these behave and how much better they are relative to NCA cells in S, and the benchmark LCO in 1.5-2.5.
Please keep us posted.
These are one off cells for the Roadster only. S/X/3 and future variants will use in-house designed batteries made at the gigafactory. Tesla outsourced the cells to LG because the Roadster is a small volume unique car. Each battery pack for example is hand built and the donor enclosure is re-used for subsequent 3.0 packs. Not much in common between the 3.0 pack and future Tesla's. NMC is the defacto choice for most production EV's (Nissan, GM, Ford etc) outside of Tesla.
The only relationship between the LG HG2 and the Panasonic derived tesla 3 chemistries, besides the now ubiquitous NMC you mention , may be the use of small % silicon in the negative. In some ways, I wonder if this is to get some early test data. If it is, it was a good plan. There were some reports in 2016 that the use of % silicon in the negative was a possible pathway for the tesla 3, I have not seen anything official, may be you have some info. Anyway, in the Li-ion battery world even these use of "NMC" is meaningless unless the ratios of Ni/Mn/Co are defined and then you have differences in the additives , electrolytes, electrode formulations, etc .all which make comparative generalizations near impossible relating to calendar life. Maybe for cycle life one can make some assumptions based on active chemistries, even then it is difficult.
Each new Li-ion battery is analogous to your manufacturer changing the design of your favorite IC engine for something "better" based on their exhaustive development testing. It may very well be be great, but time, use, and abuse patterns will finally dictate the final story. Regardless, the energy increase in the roadster 3.0 is very impressive as has the entire battery management system since the first roadster.
The energy and mileage is so large, I wish there was an easier way to further narrow the operating window of the cell through automatic narrowing of charge and discharge extremes (without manually limiting charge and keeping an eye on the discharge). Maybe a "long lifetime" charge setting (charge limited to 175 miles), standard (250), range 300+). With the long lifetime having typical taper charge protocol and incorporate cell balancing. This would be ideal. I would rather more numerous shallow charge-discharge cycles in the stable mid voltage ranges than less cycles of deep charge-discharges at even the moderate extremes any day.
dhrivnak
Active Member
The problem however with charging below 83% at the top end is the system is unable to balance. By charging up to 83% the system will remain in balance which is critical for long life.
Balancing is only dictated by the battery management software/firmware used in the 1.5-2.5roadster. IMHO, there really is no fundamental reason why one could not balance at other cutoff voltages as long as the battery voltage as a function of capacity curve has enough slope to it such that the rebalancing can be done efficiently. NMC, (depending on the ration of Ni/Co/Mn) has a higher and more continuous slope than the LiCoO2 used in 1.5-2.5 roadsters, so it could be plausible. It just becomes very difficult when the voltage profile becomes very flat as a function of capacity. LiCoO2 has a very flat voltage profile at the lower state of charge which may make balancing very difficult. NMC may be very advantageous for both flexibility in cell balancing and more accurate determination of state of charge and thus CAC and also calculated ideal miles. NCA would make this slightly easier also. There is much complexity in the CAC calculation which I don't know, but in general everyone wants a nice smooth slope in their positive electrode materials, and thus the output voltage of their battery. Just my opinion.
dhrivnak
Active Member
When I did testing of lithium batteries the voltage was very flat from 15% to 85% so it was impossible for me to check charge level in the broad middle level. I had to wait for either the low or high knee before I could tell differences.
Do you know what was the positive electrode? NMC and NCA cells I have tested have a nice slope (depending on composition), LCO cells have a flat phase transformation at the lower levels of charge in the range you mention.
As you state, really tough to check charge levels on the latter cells unless you have prior discharge capacity data.
It's in the log file that you can get with a USB drive. That's where the whole car histories that Dave and I used to make our graphs came from.
dhrivnak
Active Member
My testing was on Li-FePO4 so a different chemistry. But what I saw was one could use a Volt meter on lead acid batteries as there was a clear voltage slope. But the lithium ones were extremely flat except at low and high states of charge. All cells drift with time and one of Tesla's secrets I believe is the balancing. Between 30 and 70% it was very difficult to determine charge level. But at 10% or 90% the differences became apparent. Since a car rarely spends a lot of time at a low state of charge it means sence to balance at the high end. NCO chemistry may be different but if they are like Li-FePO4 I would not try to balance in the middle states of charge.
I am not convinced the HG2 cells will be as good long-term as the panasonic small form factor LCO cells. No other auto manufacturer has deployed small form factor NMC cells in an automotive production setting. The Panasonic LCO cells were tested extensively for several years prior to the Roadster release. My hope is Tesla utilizes their production GF cells into a Roadster enclosure and that is the pack I will wait for. I think the HG2 cells are merely a stop gap that will only last till Tesla is able to get to scale with their GF cells. I am not a believer in the HG2 cells for automotive.
dhrivnak
Active Member
Unfortunately that is very unlikely as the GF has moved to a different cell format that is a little larger. So they will not fit unless the battery is totally redesigned which I do not have much hope they will do.
Similar threads
