3.0 Battery Longevity

[ICON]

Very smart analogy mark thanks.
Did you observe any similar degradation with the Hong Kong R80 battery you pulled data from?

 

[gregd]

Right, I understand all that. But if we extrapolate the diminishing-returns curve for the range charges, do we get to a place where the battery's real capacity (presumably then known) is reasonable? My recollection is that it still is not. And, we should be seeing much larger jumps in the positive direction for more recent range charges, since they are farther out of sync from reality, for example. Looking at the graphs, I don't think I am seeing that. (Bolosky, can you correct me?) Remember, we're something like 2-3x the rate of decline... This isn't a margin-of-error sort of situation.

Further, the results of a sudden "Oops!" at the low end of the charge, while not surprisingly confirming that the algorithm is inaccurate, also shows us that it's inaccurate in the wrong direction (really has less capacity), no?

Perhaps I'm expecting too much, but what I expect for my $30k (a significant portion of a whole new car) is a new battery that has a higher capacity than my original (since it is a decade's worth newer technology), and a life span that is at least as good as what the original battery turned out to be. I'm at 8 1/2 years and down something like 13% - 14% in capacity. If Tesla would warrant the new battery for anything close to this, I'd buy it. Instead we get a 1-2 year "repair parts" warranty. This is not a repair of an original part, it is a new part. A fresh upgrade. It should carry a warranty commensurate with that.

Bottom line, we should not let the excuse of not knowing precisely what capacity the battery has to misdirect our attention away from the apparent fact that it is declining far faster than the original battery did, and in fact, faster than any of the other batteries in any of the other Tesla cars. None of the algorithmic adjustments suggested so far can account for what we are observing, as compared for what we observed for the original battery.

 

[slcasner]

I too suspect an algorithm error. We know they don't have people working on the roadster firmware any more, and that algorithm must have been tuned for the behaviour of the old cells.

An ex-insider told me that minimal changes were made to the firmware when the R80 battery was developed. That suggests to me that Mark's assumption is correct.

The actual range you can drive concerns the 'bottom' of the pack, and is much simpler. When the lowest brick hits the low voltage limit, you're done.

So then the question is, how does the range prediction progress over the course of the drive? I think we know that the number of ideal miles displayed at full is ratio of calculated Ah capacity to the original design Ah capacity times the original design ideal miles (340). Is the ideal range displayed during a drive simply based on starting with the number of ideal miles displayed after a range mode charge and then reducing that number according to the measured Ah discharge during the drive? If so, the prediction could be substantially wrong (in either direction) near the bottom. What triggers the display to switch to "cannot calculate range"? Could OVMS give us a better idea of how much range we have left by reporting the lowest brick voltage?

But, yes, the way to know for sure is to do a drive from 100% down to 0%, measuring the actual kWh used (or charge the car back up measuring at the wall, while taking into account charging losses).

I have a meter on the wall and find that the kWh number displayed on the VDS when first turning on the car after a charge is within 1 of the measured value. Calculating the charging losses accurately is tricky, though.

In addition to the experiment of running a battery all the way from full to empty and seeing how much charge (or Ah) it actually produces compared to the CAC, another thing to try would be to take a car that hasn't had all that many range charges and do a bunch in a row. If it's really doing measurements plus averaging (and if it's really consistently been underestimating the top), then the overall increase should be much bigger than what you see charging my car, since I've been doing it regularly. Probably the best car in the dataset to try this with is #33, @slcasner , since it's had few range charges and a pretty linear dropoff. If it wound up climbing back up to 200-205 Ah (roughly where my car was at the same mileage after range charging), that would indicate that the estimates just keep getting worse with no data, and also that mileage is more important than age (which is the opposite of what the CAC algorithm is saying).

Doing a bunch of range charges won't work for me because I don't drive enough. However, I do have an upcoming long trip where I can do four range charges over five days.

Been there, or very close to it last year after getting my 3.0 pack fitted. Did a range charge, topped out about 329miles, drove to Manchester (UK) and back to Oxford, about 300 miles roundtrip, on the way back in range mode the car suddenly jumped from 45 miles remaining to can not calculate range, leaving me fundementally terrified that I was about to brick the pack. Got home with a heart in mouth moment, everything switched off to minimise power drain less headlights (it was night!) doing another ten miles or so, and got car on charge, did the will charge faster with a 120V message - and after 24 hrs all was normal. That sufficiently scared me that I now never use range mode except for a range charge which I then switch back to normal mode at the 100mile mark. My SC say they're looking at it, but what that means I have no idea. Summary, it's not fun. :-(

I agree that is scary, but again I ask what triggers the display to switch to "cannot calculate range"? I don't think it is just the point where standard mode would indicated zero miles because with the original battery there were one or two occasions where I was driving in standard mode and nearing zero so I switched to range mode. I don't recall every hitting the point where the car said it could not calculate the range. Could that message be displayed when the apparent remaining capacity (and therefore range) is greater than what the algorithm had been calculating, as well as when it might be less?

On long trips I charge in range mode but always switch to standard mode when driving so that performance is not limited if I need it. That means the VDS battery gauge is pegged at 100% for a while, but that's OK. I plan my charging stops so that I don't reach zero in standard mode.

Yes had the same experience and when I THOUGHT I had about 20 miles of range left the car died, no warning just died. It did charge just fine but it was a rather expensive experiment. But I also set a personal record of 375 miles on a charge and I was NOT hypermiling

Do you mean that the VDS displayed 20 miles but then the car suddenly died? Or did the VDS say it couldn't calculate but you thought you had 20 miles left because you had only driven, say, 50 miles from the point at which the VDS said 70? Also, if you went 375 miles without hypermiling, that's significantly more than 340.

I think it's supposed to display "cannot calculate range" at 10% SOC, or ~35 ideal miles. So getting it at 45 means it was only off by 10 miles, or about 3% of the battery capacity, which isn't all that bad. Still scary.

I thought that 10% SOC is zero for standard mode, but again, I don't think the display is expected to change to "cannot calculate range" if you are in range mode at that point.

 

[gregd]

Doing a bunch of range charges won't work for me because I don't drive enough. However, I do have an upcoming long trip where I can do four range charges over five days.

Would leaving the car on in the garage (door open) with the A/C and Heat on full-blast be equivalent?

I thought that 10% SOC is zero for standard mode

From some analysis I had done earlier (see https://teslamotorsclub.com/tmc/threads/roadster-soc-mileage-mode-decoder.92322/), I believe this is correct. Something in the 10%-ish range, based on miles when you flip between Standard and Range mode. I think my analysis came in at 11%, but it's in that general area.

I have a meter on the wall and find that the kWh number displayed on the VDS when first turning on the car after a charge is within 1 of the measured value. Calculating the charging losses accurately is tricky, though.

Yes, those two numbers match well. The trouble comes in when you look at the logs and try to square it against what goes into the battery... See: https://teslamotorsclub.com/tmc/posts/2169156/ Still trying to clean up the stain on the floor from all the loose electrons.

 

[slcasner]

Yes, those two numbers match well. The trouble comes in when you look at the logs and try to square it against what goes into the battery... See: https://teslamotorsclub.com/tmc/posts/2169156/ Still trying to clean up the stain on the floor from all the loose electrons.

Initially the car's report of kWh was not correct. At the Model S unveiling party I complained about this to JB so he asked for some data, which I provided in a followup email. After that they fixed the firmware.

I wish we still had that degree of attention.

 

[dpeilow]

I too suspect an algorithm error. We know they don't have people working on the roadster firmware any more, and that algorithm must have been tuned for the behaviour of the old cells.

The 3.0 firmware is an integer upgrade, so I hope it is more than just a tweak. Furthermore, given the price of this upgrade, I am entitled to a properly engineered product - including the all important SOC calculation algorithm.

Besides, as others have said, the "cannot calculate range" warning kicks in at a higher ideal miles level now, I suspect that is based on the actual voltage measured. It also trickle charged back up to a higher level, so I suspect it is really that low and not just an off calculation.

As I have said before, Tesla could answer this quickly based on the information we put into that letter.

 

[gregd]

Hi folks,

A bit of good news, for a change... TL;DR version: Tesla knows we have an issue, and are actively working on it. I guess the message is to hang in there, and let them do their work. We've got their attention.

Long version: I listen to a number of podcasts (too many, probably), several related to Tesla, and on one, "Ride the Lightning", the host Ryan mentioned he knew someone who happened to end up working at Tesla. I took a chance, and sent him an email with a HUGE request: could his contact determine what the status of this is, or if it even had made it into the organization. We've only heard crickets. I also gave him a link to the thread, and a summary of the problem such as we have discussed here already.

Well, I heard back from Ryan, and with his permission, here is what he said:

Hi Greg,

With the help of a contact at Tesla I was able to get this forwarded along, and I have some good news for you. Without getting into details about who specifically this got to, they are aware of it. My contact told me a direct quote from a person in the right place to do something about this, and that direct quote is: "It appears this is a known problem and lots of people are working on it."

Feel free to relay this to the thread publicly or privately; I'm not after any credit or anything so I'll leave it to you to post if you like -- I'm just glad that I was able to help you guys out!

All the best,

Ryan​

So, a big thank you to Ryan and the folks who helped let us know what was going on. To the thread contributors and lurkers, please don't try to pick apart the nuances of the wording of this - I was only asking for an acknowledgment that our letters and data have made it to the right people, and that is what I got. Now, I just need wait for their work to conclude, and a definitive answer to be reported. My $30k is still firmly immobilized until then.

 

[dpeilow]

^ I have to say, that's not really news. Several people have said the same thing - that it is a known issue - but the minimal feedback we have had is either 1) it's Roadster owners not charging their cars properly or 2) it's being handed off to Service, which means they want to deal with it on a car-by-car basis.

Call me a cynic, but I will believe that real action is being taken when someone from Tesla corporate responds either directly to the letter we put in or they make a statement to say they are looking into the issue and/or have a solution. So far the only official response I have seen by email is that it has been passed to one place and then another place in the company. The fact that we have to use these friend of a friend type of channels isn't encouraging.

 

[Marius]

David, how is your car battery performing. We seem to have the last update still when it was CAC 212. Your pack is now 1 year old. If it is still around this number you will be the first one with a good performance and you are the only one who got a refresh of software installment.
Could this be the answer?

 

[dpeilow]

David, how is your car battery performing.

At the present time the CAC is 207.57.

I did not drive much over the winter, so the CAC stayed around 212. Now that the sun is out and I have been using the car more for the past 5 to 6 weeks, I am seeing the same drop (I predicted this in the text of my letter). Basically, it is dropping by 0.5 to 1 Ah every time I use it.

 

[backmost]

Hey bolosky,

How often do you want updates on the CAC? Since I'm taking over #209 I want to continue to supply data to monitor these batteries.

 

[bolosky]

Hey bolosky,

How often do you want updates on the CAC? Since I'm taking over #209 I want to continue to supply data to monitor these batteries.

Every few months, maybe quarterly. Msjulie sent me an update right before so sold the car, so you've got a while.

And thanks!

 

[thebabydoc]

Every few months, maybe quarterly. Msjulie sent me an update right before so sold the car, so you've got a while.

And thanks!

I e-mailed you with my 0707 logs and a question. Check your spam folder.

 

[dpeilow]

I've just sent @bolosky my latest log file.

I am down to 205.78 Ah now. That's lower than when I first had the battery installed with that strange hiccup and means I have dropped 10 points in 9 months. This weekend was the worst yet though, a 200 mile round trip with a partial range charge and use right away cost almost 2 CAC points.

I am starting to think that hard, fast driving (and hence power use) is the big factor.

At the event I went to on Saturday I also met Markus Doessegger from the 80 eDays team. He got the 3.0 upgrade last year and is already down to a CAC of 188. I told him to send in his log files.

 

[bolosky]

Here's the latest update, with new data for 707, 330(EU) and 670. Thanks thebabydoc and dpeilow.

707 stands out as the quickest CAC drop by mile of any car, but it's in the middle of the pack in the by-day graph, so I think all we've learned is that calendar time is more important and it's a very low mileage car (1500 miles in a year, roughly).

330(EU) is doing better than most cars both by-day and by-mile, but if you were to just eliminate the section where it was reset at about 2.5 months/1K miles it would be pretty typical. This may be among the most interesting pieces of data in the whole data set. If what we were seeing was based on what was happening in the cells, then you'd expect the effect of the reset to go away, since it was all software and no hardware. The fact that it's just a constant offset makes it look like most of the signal isn't all that related to what's happening in the cells.

And, yes, it's taking some pretty big drops recently, but it also mostly wasn't driven over the winter, so it's just catching up to the baseline, I think. Though I do like the idea that hard driving has a disproportionate effect. I'm not sure how I could measure that from the data that I have. While I do have some of the short-term logs were I could figure out speed and acceleration, they're pretty limited. Maybe I could work with what I've got, I'll have to think about it some, it's an interesting idea.

I also added about 6 weeks' data for 670. It's continuing to be remarkably level. It first hit its current CAC back in November, 8300 miles ago. Maybe it really has hit a knee in the curve, but I'm still somewhat skeptical.

 

[bolosky]

For some reason, the non-zoomed version of the by-mileage graph didn't upload correctly. Here it is.

 

[dpeilow]

I'm not sure how I could measure that from the data that I have.

@tomsax 's log file parser includes the -v switch that gives you the Wh/mile for each drive, which might be a good indicator.

 

[JRP3]

Does 209 appear to be leveling out?

I realized I mixed up the colors of 209 and 670. Looking at the latest data 670 does in fact show leveling out. Since the car with the most use is showing flattening capacity loss either it's the first one to hit that part of the curve or the use profile allows the algorithm to more accurately calculate actual capacity. Or a bit of both I suppose.

 

[bolosky]

I realized I mixed up the colors of 209 and 670. Looking at the latest data 670 does in fact show leveling out. Since the car with the most use is showing flattening capacity loss either it's the first one to hit that part of the curve or the use profile allows the algorithm to more accurately calculate actual capacity. Or a bit of both I suppose.

Maybe. If you look at the pattern, you'll see it drop down to about 185 Ah and then climb back up to about 190Ah over and over. The climbs correspond to range charges, often several in a row (this is my car, so I know). It seems like several times it's climbed to the same place around 190Ah and stopped. It's possible that this has convinced the algorithm that the battery isn't degrading as fast anymore, hence the flat spot at the very end.

I had thought that maybe the algorithm just had a natural bend in the curve around 185 Ah, but #33 blew on through that level without flinching at all, so maybe not. Perhaps something else is happening, and I have an idea of what.

I'll put up another post tonight describing the idea. I'd do it now, but I'm at work and don't have access to the data (plus, I should actually do some work...)

 

[Marius]

Does someone understand why 189 is such a good performer?