3.0 Battery Longevity

[dpeilow]

Thanks for plotting those (however I think you mean after 1k miles, not 3k miles for mine to return to expected CAC).

I pulled the log and ran it through @tomsax 's parser and it confirmed that:

- The upgrade happened on 16th June but the car briefly thinks it is 25th Jan 2010. It still had the old firmware at that point. Probably powering up after the battery swap.
- Firmware 5.1.2 was installed on that date 30 minutes later. I can't see any evidence of it changing since then.
- In the first log entry 1 day after the install the car is giving CAC of 215, but when they full charge from 42% to 94% it drops to 205 in the entry a day later.
- When I dropped the car off on 4th of September the CAC was 206, after I did a few range charges earlier in August.
- The SC then did a full standard mode charge, then range charged it, drove it for a short drive (interestingly with the TC off for 28 minutes but not going far), then did a series of 'top ups' in quick succession on the 6th.
- The CAC now jumps back up to 215. It has since fallen down to 213.6 after approximately 225 miles of driving.


06/17/2017 14:09:55 | 1497704995 | DAY | odo = 29265.0 range soc = 42%, brick ave 214.574Ah, brick min 214.574Ah, min Ah brick 0, CAC 215.04 Ah
06/18/2017 14:09:55 | 1497791395 | DAY | odo = 29265.0 range soc = 94%, brick ave 205.681Ah, brick min 204.940Ah, min Ah brick 80, CAC 205.38 Ah

It is quite interesting that the CAC value when the pack was installed on June 16th - 215.04 - is exactly the same as it was on Sept 7th when the fix was applied.

09/07/2017 09:58:22 | 1504774702 | DAY | odo = 30304.3 range soc = 96%, brick ave 214.574Ah, brick min 214.574Ah, min Ah brick 0, CAC 215.04 Ah

I wonder if this is a set of default theoretical values on "Day 1" of the 3.0 battery install which the system then iterates, probably with some hysteresis, as it samples the real Ah capacity?

Obviously as my pack has aged by over 3 months at this point, the CAC is pretty unlikely to be the same with the kind of drops that everyone else is seeing. It has to be an artificial starting number.


I'll put my tin foil hat on for a minute and also suggest this: What if the real CAC is nowhere near 215 and the algorithm is just trending towards the capacity the pack really has? It would explain perhaps why "cannot calculate range" now kicks in with ~45 miles left and someone said they got flatbedded shortly afterwards.



Full logs from the period when they did the fix below. Unfortunately the transient log starts on 21st of June, so I just missed the week when the original install was done.

09/04/2017 19:26:23 | 1504549583 | DAY | odo = 30301.0 range soc = 65%, brick ave 206.593Ah, brick min 205.738Ah, min Ah brick 15, CAC 206.18 Ah
09/05/2017 09:01:33 | 1504598493 | ERR | error code 1092 6 bytes
09/05/2017 09:33:32 | 1504600412 | ERR | error code 1146 6 bytes
09/05/2017 09:33:51 | 1504600431 | ERR | error code 1146 6 bytes
09/05/2017 13:20:44 | 1504614044 | ERR | error code 1146 6 bytes
09/05/2017 13:21:22 | 1504614082 | ERR | error code 1146 6 bytes
09/05/2017 15:39:54 | 1504622394 | ERR | error code 283 7 bytes
09/05/2017 15:39:57 | 1504622397 | ERR | error code 65535 5 bytes
09/05/2017 15:39:59 | 1504622399 | ERR | error code 65535 5 bytes
09/05/2017 16:02:51 | 1504623771 | VINF | L xxxxxxxxxxxxxxxx330 '5.1.2 42'
09/05/2017 16:02:53 | 1504623773 | XX05 |
09/05/2017 16:02:53 | 1504623773 | XX05 |
09/05/2017 16:02:53 | 1504623773 | XX05 |
09/05/2017 16:02:53 | 1504623773 | XX05 |
09/05/2017 16:02:54 | 1504623774 | XX05 |
09/05/2017 16:02:54 | 1504623774 | XX05 |
09/05/2017 16:02:54 | 1504623774 | XX05 |
09/05/2017 16:02:54 | 1504623774 | XX05 |
09/05/2017 16:02:55 | 1504623775 | XX05 |
09/05/2017 16:02:55 | 1504623775 | XX05 |
09/05/2017 16:02:55 | 1504623775 | XX05 |
09/05/2017 16:02:55 | 1504623775 | XX05 |
09/05/2017 16:02:55 | 1504623775 | XX05 |
09/05/2017 16:02:58 | 1504623778 | XX05 |
09/05/2017 16:02:59 | 1504623779 | XX05 |
09/05/2017 16:09:50 | 1504624190 | ERR | error code 1146 6 bytes
09/06/2017 08:40:32 | 1504683632 | VINF | L xxxxxxxxxxxxxx330 '5.1.2 42'
09/06/2017 08:40:34 | 1504683634 | XX05 |
09/06/2017 08:40:34 | 1504683634 | XX05 |
09/06/2017 08:40:35 | 1504683635 | XX05 |
09/06/2017 08:40:35 | 1504683635 | XX05 |
09/06/2017 08:40:35 | 1504683635 | XX05 |
09/06/2017 08:40:35 | 1504683635 | XX05 |
09/06/2017 08:40:35 | 1504683635 | XX05 |
09/06/2017 08:40:35 | 1504683635 | XX05 |
09/06/2017 08:40:35 | 1504683635 | XX05 |
09/06/2017 08:40:35 | 1504683635 | XX05 |
09/06/2017 08:40:35 | 1504683635 | XX05 |
09/06/2017 08:40:39 | 1504683639 | XX05 |
09/06/2017 08:40:40 | 1504683640 | XX05 |
09/06/2017 08:40:40 | 1504683640 | XX05 |
09/06/2017 08:40:48 | 1504683648 | XX05 |
09/06/2017 08:59:34 | 1504684774 | ERR | error code 65535 7 bytes
09/06/2017 08:59:35 | 1504684775 | ERR | error code 65535 5 bytes
09/06/2017 09:07:25 | 1504685245 | ERR | error code 1146 6 bytes
09/06/2017 09:08:19 | 1504685299 | ERR | error code 1146 6 bytes
09/06/2017 09:09:00 | 1504685340 | ERR | error code 1146 6 bytes
09/06/2017 09:43:24 | 1504687404 | ERR | error code 1146 6 bytes
09/06/2017 09:46:21 | 1504687581 | ERR | error code 1146 6 bytes
09/06/2017 13:17:58 | 1504700278 | ERR | error code 1146 6 bytes
09/06/2017 13:30:18 | 1504701018 | ERR | error code 1146 6 bytes
09/07/2017 08:27:12 | 1504769232 | ERR | error code 1146 6 bytes
09/07/2017 09:58:22 | 1504774702 | DAY | odo = 30304.3 range soc = 96%, brick ave 214.574Ah, brick min 214.574Ah, min Ah brick 0, CAC 215.04 Ah




09/04/2017 12:08:44 - 09/04/2017 13:43:21 (01:34:37) Drive 73% -> 44% 75.9 mi 57.8 mph 70.0 mph 30300.9 mi 22.9 kWh 7.3 Ah 69.8 Ah 302 Wh/mi
09/04/2017 14:30:33 - 09/04/2017 14:32:25 (00:01:52) Drive 44% -> 44% 0.0 mi 0.0 mph 10.2 mph 30300.9 mi 0.0 kWh 0.0 Ah 0.1 Ah
09/04/2017 17:18:57 - 09/04/2017 17:19:38 (00:00:41) Drive 44% -> 44% 0.1 mi 9.5 mph 5.3 mph 30301.0 mi 0.0 kWh 0.0 Ah 0.0 Ah
09/05/2017 08:43:56 - 09/05/2017 08:44:51 (00:00:55) Drive 44% -> 44% 0.0 mi 0.0 mph 7.0 mph 30301.0 mi 0.0 kWh 0.0 Ah 0.1 Ah
09/05/2017 09:01:50 - 09/05/2017 14:31:44 (05:29:54) Charge 44% -> 78% 228V 32A of 32A 33.4 kWh 0.0 Ah 0.0 Ah
09/05/2017 16:05:31 - 09/05/2017 20:03:46 (03:58:15) Charge 78% -> 97% 231V 33A of 32A 23.4 kWh 0.0 Ah 0.0 Ah
09/06/2017 07:18:30 - 09/06/2017 08:07:12 (00:48:42) Charge 96% -> 98% 237V 32A of 32A 2.8 kWh 0.0 Ah 0.0 Ah
09/06/2017 08:25:40 - 09/06/2017 08:26:12 (00:00:32) Drive 98% -> 98% 0.0 mi 0.0 mph 6.5 mph 30301.0 mi 0.0 kWh 0.0 Ah 0.0 Ah
09/06/2017 08:49:58 - 09/06/2017 08:51:54 (00:01:56) Drive 98% -> 98% 0.0 mi 0.0 mph 8.2 mph 30301.0 mi 0.0 kWh 0.0 Ah 0.1 Ah
09/06/2017 09:06:19 - 09/06/2017 09:12:18 (00:05:59) Drive 98% -> 98% 1.4 mi 17.0 mph 35.0 mph 30302.4 mi 0.6 kWh 0.1 Ah 1.4 Ah 399 Wh/mi
09/06/2017 09:42:45 - 09/06/2017 10:14:54 (00:32:09) Drive 98% -> 97% 0.3 mi 4.1 mph 28.0 mph 30302.7 mi 0.5 kWh 0.0 Ah 1.2 Ah TC off

for 28 minutes
09/06/2017 13:15:16 - 09/06/2017 13:21:25 (00:06:09) Drive 96% -> 96% 1.4 mi 18.2 mph 33.6 mph 30304.1 mi 0.4 kWh 0.2 Ah 1.1 Ah 283 Wh/mi
09/06/2017 13:27:00 - 09/06/2017 13:28:11 (00:01:11) Drive 96% -> 96% 0.0 mi 0.0 mph 14.9 mph 30304.1 mi 0.0 kWh 0.0 Ah 0.1 Ah
09/06/2017 13:28:12 - 09/06/2017 14:38:58 (01:10:46) Charge 96% -> 98% 236V 32A of 32A 3.4 kWh 0.0 Ah 0.0 Ah 1096 Wh/mi
09/06/2017 14:43:39 - 09/06/2017 14:45:15 (00:01:36) Charge 98% -> 98% 231V 13A 0.1 kWh 0.0 Ah 0.0 Ah
09/06/2017 14:45:51 - 09/06/2017 14:47:26 (00:01:35) Charge 98% -> 98% 234V 12A of 32A 0.1 kWh 0.0 Ah 0.0 Ah
09/06/2017 14:47:37 - 09/06/2017 14:48:12 (00:00:35) Charge 98% -> 98% 239V 10A of 32A 0.0 kWh 0.0 Ah 0.0 Ah
09/06/2017 14:49:21 - 09/06/2017 14:50:57 (00:01:36) Charge 98% -> 98% 232V 9A 0.0 kWh 0.0 Ah 0.0 Ah
09/06/2017 14:51:07 - 09/06/2017 14:52:42 (00:01:35) Charge 98% -> 98% 236V 9A of 32A 0.0 kWh 0.0 Ah 0.0 Ah
09/06/2017 14:52:52 - 09/06/2017 14:54:27 (00:01:35) Charge 98% -> 98% 240V 10A of 32A 0.0 kWh 0.0 Ah 0.0 Ah
09/06/2017 14:55:28 - 09/06/2017 14:59:03 (00:03:35) Charge 98% -> 98% 240V 10A of 32A 0.1 kWh 0.0 Ah 0.0 Ah
09/06/2017 15:00:00 - 09/06/2017 15:01:35 (00:01:35) Charge 98% -> 98% 240V 8A of 32A 0.0 kWh 0.0 Ah 0.0 Ah
09/07/2017 08:18:23 - 09/07/2017 08:19:54 (00:01:31) Drive 96% -> 96% 0.1 mi 4.6 mph 19.2 mph 30304.2 mi 0.1 kWh 0.0 Ah 0.2 Ah
09/07/2017 08:25:10 - 09/07/2017 08:28:47 (00:03:37) Drive 96% -> 96% 0.1 mi 10.3 mph 9.4 mph 30304.3 mi 0.0 kWh 0.0 Ah 0.1 Ah

 

[dpeilow]

@Kerios kindly sent me his log file. From that I can see that:

- The CAC also starts at 215.04 like mine.
- The CAC drops literally every time the car moves. With the 2.0 before you had to do a significant drive before it recalculated.
- The firmware is also 5.1.2 42

 

[dpeilow]

Hmm looks like on my log file the pack size calculation is weird.

09/04/2017 12:08:44 - 09/04/2017 13:43:21 (01:34:37) Drive 73% -> 44% 75.9 mi 57.8 mph 70.0 mph 30300.9 mi 22.9 kWh 7.3 Ah 69.8 Ah 302 Wh/mi

=> 22.9 kWh / (73-44)% = 78.97 kWh so probably correct.

09/05/2017 09:01:50 - 09/05/2017 14:31:44 (05:29:54) Charge 44% -> 78% 228V 32A of 32A 33.4 kWh 0.0 Ah 0.0 Ah

=> 33.4 kWh / (78-44)% = 98.24 kWh so wrong

09/05/2017 16:05:31 - 09/05/2017 20:03:46 (03:58:15) Charge 78% -> 97% 231V 33A of 32A 23.4 kWh 0.0 Ah 0.0 Ah

=> 23.4 kWh / (97-78)% = 123.16 kWh so wrong

Is this really in the log file or an artifact in @tomsax 's software?


UPDATE: Choosing a few random charging and driving sessions from @Kerios 's log file I see the same behaviour, namely that driving sessions have a broadly correct percentage (if I work out for 100% DoD it's close to 80 kWh), whereas charging sessions are way off.

Am I missing something here or is this a possible firmware bug that is screwing the CAC algorithm?

 

[bolosky]

Thanks for plotting those (however I think you mean after 1k miles, not 3k miles for mine to return to expected CAC).

Sorry, I misread my own graph. It is indeed about 1K miles.

 

[hcsharp]

I'll put my tin foil hat on for a minute and also suggest this: What if the real CAC is nowhere near 215 and the algorithm is just trending towards the capacity the pack really has? It would explain perhaps why "cannot calculate range" now kicks in with ~45 miles left and someone said they got flatbedded shortly afterwards.

I've wondered that a few times myself. The good news would be that it's not really degrading. The bad news would be that it was never anywhere near 80 kWh to begin with.

 

[dpeilow]

I've wondered that a few times myself. The good news would be that it's not really degrading. The bad news would be that it was never anywhere near 80 kWh to begin with.

I'm less inclined to believe my own conspiracy theory now that I've seen the drive side of the log file consistently give ~80 kWh when you normalize for 100%. Looks like an issue with the charging side.

I'm going to test it on an old 2.0 file when I get home.

 

[dpeilow]

OK so I see the same effect in an old log file - then it struck me that the energy shown in the cases I was looking at was correct for the energy in. But it is not correct in the new file. There is a big discrepancy between the energy shown and the VxA going in. In the 23.4 kWh case above the VxA is 30.5 kWh. But it is the lower number that gives the normalised total of 123 kWh, so it cannot be the case that it is the pre-charging losses value.

 

[bolosky]

Adding dhrivnak's #255 and a few weeks' more data for my #670, as well as changing the color of some lines, and fixing up the gridlines to make the graph a little easier to read (because I clearly need that, since I misread it earlier).

#255 seemed to have unusually low degradation for the first few hundred miles, then had a large, sudden downward correction that put it right in line with the others.

The real news seems to be that my car continues to reduce its rate of decline. It's barely moved for the past thousand miles, which didn't include any range charging (though that will change tomorrow night). But regardless, the slope of the line after the big bump up around 7K miles to now is about half what it was from 0 to just under 6K where I started range charging it (and it would be even less than half if I measured it from 6K rather than the top of the bump). So, that's somewhat encouraging. I'm still very curious as to whether #181 will plateau around CAC 191 like #670 did.

 

[dpeilow]

So I imported an old 2.0 log file and the new 3.0 log file into Excel and started analysing a bit more.

There are definitely discrepancies with the logging of energy use when charging, much more so than the driving. I don't know the car's power factor but that certainly doesn't explain everything I am seeing. However as the same effect is seen on both files it doesn't necessarily explain these drops.


2.0

3.0

 

[dpeilow]

The real news seems to be that my car continues to reduce its rate of decline.

Is this true though? It seems to do something similar between 14-15k and 16-17k miles. Could just be temporary.

 

[bolosky]

Is this true though? It seems to do something similar between 14-15k and 16-17k miles. Could just be temporary.

Hard to say. And since I'm taking it on a long drive tomorrow and will be range charging it, that will likely throw the data off some.

One thing that's a little interesting is that if you draw a line from 0 to the top of the peak at 7K miles, it more-or-less tracks the history of the car from 7K miles on, as if it decided that it learned a new, lower rate of decline then and just stuck with it. I should make a picture of that and post it.

 

[ion_1]

This is a solid set of data. Thank you all that have been participating in Bolosky's study. Again, we don't have confirmation on what type of cells these are from LG (some felt they are HG2) but if they are Si-graphite negatives, these should level and cycle nicely by 20% capacity loss (hopefully).

There is enough data here to start to extract important guidelines on how to treat these cells. Some takeaways for discussion (I ignored the extensive no change in capacity when the batt was fresh, mostly focused on fade rate once they were used):

1. The strongest comparison to watch is between #670, 181, and 33. These 3 have virtually identical fade rates as a function of time, and have the same age. However, there is a very wide difference in the number of miles (i.e. #cycle and/or depth of discharge). There is a slight trend for worse behavior with greater miles, but 670 exceed 33 by a significant 16k miles. In this case calendar life is dominant factor towards fade.

2. Regarding 670,181 and 33, it would be interesting to see if there any other factors involved that could change this conclusion. #1 priority would be adding a chart with # of range charges...

3. most #670 changed fade rate (ignoring jump in capacity) significantly at mile 6k, what's your magic change in behavior?

4. 425 and 537 will be a nice pair to watch also. Miles are similar, but loss is contrasting, 537 is older. Again calendar aging.

5. 660 is very interesting, once it started to fade, the fade rate is significantly less than the others

--> As these cells are significantly different than generation 1, I would assume Tesla would be using this as an opportunity to optimize the charging protocols. Is anyone seeing changes in firmware with SC visits? Do all of you have the same firmware?

--> with all this extra capacity, it is simply a waste of significant degradation to keep the cells at full charge if you do not need it. It would be great to have a way to limit charge simply and still have rebalancing. Personally, I would want a way to keep the the charge to about 150 miles and extend longevity for most of my needs... maybe I should know this by now , but is there an easy way to do this without manually stopping the charge?

Wow, this is great data. I expected more effect as a function of cycles with these cells, but it does not seem to be the case. It is early
and all of the conclusions may be different in the end. There are 4 main factors to battery aging: the chemistry, the quality of the near-hermetic cell enclosure that the manufacturer uses, quality of battery management/thermal control by Tesla, and the behavior of the consumer. At least we can control one of them.

Thanks to all, especially Bolosky.

 

[dpeilow]

these should level and cycle nicely by 20% capacity loss (hopefully).

I hope not as that is close to the 160 CAC of the original packs.

Do all of you have the same firmware?

Here is my firmware:

And here is @Kerios 's:

So our EU cars are the same.

 

[ecarfan]

As these cells are significantly different than generation 1, I would assume Tesla would be using this as an opportunity to optimize the charging protocols. I

I would not assume that. Earlier this year I had the opportunity to speak to a Tesla firmware engineer who has been with the company for over 10 years and wrote some of the firmware for the Roadster. He told me that the Roadster 3.0 battery was done as a "side project" with minimal resources and I got no indication from him that Roadster firmware was under continuing development.

That is not a criticism of Tesla, simply a statement of the reality that currently Tesla does not devote any resources to ongoing Roadster development. It is a model that ended production 6 years ago.

with all this extra capacity, it is simply a waste of significant degradation to keep the cells at full charge if you do not need it. It would be great to have a way to limit charge simply and still have rebalancing. Personally, I would want a way to keep the the charge to about 150 miles and extend longevity for most of my needs... maybe I should know this by now , but is there an easy way to do this without manually stopping the charge?

See above. In my opinion, and based on what I stated, no new features will be added, even minor things like being able to set the charge level beyond the 3 current choices.

So if you want to charge your Roadster to about 150 miles you have to manually stop the charging at that point.

 

[hcsharp]

--> with all this extra capacity, it is simply a waste of significant degradation to keep the cells at full charge if you do not need it. It would be great to have a way to limit charge simply and still have rebalancing. Personally, I would want a way to keep the the charge to about 150 miles and extend longevity for most of my needs... maybe I should know this by now , but is there an easy way to do this without manually stopping the charge?

I'm not convinced that stopping at a lower charge level would be of any help. In fact I believe there's greater potential to do more harm than good. We don't know everything about this chemistry but the std charge level is already below the point where it would have a significant impact on longevity, at least in the relatively short time since the 3.0 packs were released. A pack that's charged to a higher SOC generates less heat in the cells while driving. Less heat = less impact on cell life. Not to mention the challenges with balancing.

+1 to your praise for @bolosky for helping the whole community with this. Thank you!!!

 

[bolosky]

1. The strongest comparison to watch is between #670, 181, and 33. These 3 have virtually identical fade rates as a function of time, and have the same age. However, there is a very wide difference in the number of miles (i.e. #cycle and/or depth of discharge). There is a slight trend for worse behavior with greater miles, but 670 exceed 33 by a significant 16k miles. In this case calendar life is dominant factor towards fade.

Perhaps we'll get an update on #33. 670 is mine, so I update it all the time, and Dave sends me logs for #181 every few months, but the latest date in the log for #33 is 7/1/17.

2. Regarding 670,181 and 33, it would be interesting to see if there any other factors involved that could change this conclusion. #1 priority would be adding a chart with # of range charges...

I could do this, but I'm pretty sure that I'll get the opposite result to what you're expecting. See below.

3. most #670 changed fade rate (ignoring jump in capacity) significantly at mile 6k, what's your magic change in behavior?

I thought that possibly the problem was that the battery wasn't balancing due to the lower % charge in standard mode compared with the old battery, so I range charged it every day for a week, then went back to normal behavior. I no longer believe that balancing was (much of) the problem, because it didn't seem to balance all that much. However, I suspect that the CAC computation algorithm is able to gather more data at full charge and the actual capacity loss was less than what it projected.

I made a graph of the CAC vs. mileage for just my car, then drew a straight line from the starting point through the top of the bump at 6K miles, on the assumption that the car learned a new fade rate then. It's actually not too far off from observation.

5. 660 is very interesting, once it started to fade, the fade rate is significantly less than the others

660 is hardly ever driven. It seems like cars that don't get much mileage take steep jumps down to the trend line when they are driven a decent amount. I suspect that if it had a day with a couple of hundred miles that you'd see it snap to the trend line for its calendar age.

 

[bolosky]

Oh, and the other bumps in #670's line (like at 9500, 11500, 15K and 17K) are also when I range charged it, typically once or twice, not for a whole week.

 

[ion_1]

I'm not convinced that stopping at a lower charge level would be of any help. In fact I believe there's greater potential to do more harm than good. We don't know everything about this chemistry but the std charge level is already below the point where it would have a significant impact on longevity, at least in the relatively short time since the 3.0 packs were released. A pack that's charged to a higher SOC generates less heat in the cells while driving. Less heat = less impact on cell life. Not to mention the challenges with balancing.

+1 to your praise for @bolosky for helping the whole community with this. Thank you!!!

I am quite convinced that higher state of charge is detrimental to long term longevity. This is based on solid fundamental science regarding acceleration of deleterious reactions at the interfaces between electrode and electrolyte, subsurface phase transitions etc . This combined with significant quantity of research of impact of state of charge on longevity of Li-ion 18650 cells clearly support this position. This holds whether it is LCO-graphite, NCA-graphite, NMC-graphite etc. Anything you can do to bring your charge state lower without causing you to push the lower end when driving hard is good. 75% charge would be a nice safe area if you have a shorter commute which could support it.

Your point regarding heat is a good one. This depends much on the slope and output voltage of various cell chemistries. LiCoO2 is good with respect to this effect vs NCA/NMC. I'm not saying 150 for everyone, just for my shorter commmute. I wouldn't want to push the lower end too much in the same spirit as you comment.

Finally, if the cells don't balance at lower states of charge, then it could be detrimental...

 

[ion_1]

Perhaps we'll get an update on #33. 670 is mine, so I update it all the time, and Dave sends me logs for #181 every few months, but the latest date in the log for #33 is 7/1/17.

I thought that possibly the problem was that the battery wasn't balancing due to the lower % charge in standard mode compared with the old battery, so I range charged it every day for a week, then went back to normal behavior. I no longer believe that balancing was (much of) the problem, because it didn't seem to balance all that much. However, I suspect that the CAC computation algorithm is able to gather more data at full charge and the actual capacity loss was less than what it projected.

I made a graph of the CAC vs. mileage for just my car, then drew a straight line from the starting point through the top of the bump at 6K miles, on the assumption that the car learned a new fade rate then. It's actually not too far off from observation.


660 is hardly ever driven. It seems like cars that don't get much mileage take steep jumps down to the trend line when they are driven a decent amount. I suspect that if it had a day with a couple of hundred miles that you'd see it snap to the trend line for its calendar age.

View attachment 252484

This is truly an amazing change in fade rate. It even seems to affect your calendar fade rate...

 

[ion_1]

I would not assume that. Earlier this year I had the opportunity to speak to a Tesla firmware engineer who has been with the company for over 10 years and wrote some of the firmware for the Roadster. He told me that the Roadster 3.0 battery was done as a "side project" with minimal resources and I got no indication from him that Roadster firmware was under continuing development.

That is not a criticism of Tesla, simply a statement of the reality that currently Tesla does not devote any resources to ongoing Roadster development. It is a model that ended production 6 years ago.

See above. In my opinion, and based on what I stated, no new features will be added, even minor things like being able to set the charge level beyond the 3 current choices.

So if you want to charge your Roadster to about 150 miles you have to manually stop the charging at that point.

Interesting info.

However, there may be another twist on this story that could be happening which would justify some attention by Tesla. So regarding the 3.0 upgrade you could assume two positions:

a) Tesla was very kind and arranged the 3.0 for us

b) Tesla was very king and arranged 3.0 for us. Also, it was a way to test the behavior of some new cell chemistries using the roadsters, and demonstrate phenomenal range. That being said, Tesla would/should watch these batts carefully and possibly offer firmware upgrades to evaluate how changes to charging protocol affect cell. Offering users an option to have a charge state slightly lower than normal to further extend life would be interesting to explore. This would not be for the benefit of the roadsters but help generate a gameplan for new options for next gen cells for S,X, and 3. I would be all over this data, the roadster community is a good test community.

But you may be probably right, Tesla may have many more issues to care about and this was a one time side project...