Model 3 Performance Battery Degradation One Month (Story)

[AlanSubie4Life]

No, that isn't true, you aren't ever allowed to drive into the bottom buffer. That is to prevent bricking the pack when you run it to "zero".

There is this video showing someone drive for about 25km past zero (which is the buffer being discussed in this context). Obviously there is an additional buffer below that to prevent bricking but that’s not what is being discussed. Is this video some sort of fluke or is it normal? I apologize for the German language video but you can turn on automatic subtitles to get the general idea.

 

[Paddy3101]

This is a BS discussion. The estimator gets more accurate as you near zero, because it measures voltage. even if it didn’t, It’s simple math. If it’s off 10% at 90%, that’s a lot of miles. 10% at 10 miles is 1 mile.

Just because it's off by 10% at 90% DOES NOT imply that its off by 10% at 10miles. There is NO evidence that is the case.

The inaccuracy could be linear, as in there is 30miles different, all the way down the scale.
The inaccuracy could be proportional to the discharge (i.e. inaccuracy degrades to zero)
The inaccuracy could be inversely proportional to the discharge, in that the inaccuracy increases as the measurement becomes smaller.
Then there is the 'Real' degradation of the battery to factor into the equation.

There are very few natural/physical properties that actually follow a linear path. Most are some form of S or U or other curved path.


In all probability, there are almost certainly multiple curved profiles of different factors, which are unique to each car, and it's current environment.

Also, don't mistake a small sample-set of measurements with a large body of data indicating a trend. i.e. difference between Weather and Climate.

 

[TimothyHW3]

No, that isn't true, you aren't ever allowed to drive into the bottom buffer. That is to prevent bricking the pack when you run it to "zero".

Sure, whatever you say... Have you tested it? Because people have...

Not sure what you "define" as bottom buffer, but the 3.5kWh, if that is what you call the bottom buffer, are exactly below 0 miles.
And this is actually common knowledge for ALL Teslas.

Here in this video he drives for about 25km more under 0km which is exactly the 3.5kWh buffer (he drives under typical consumption)

Here is another video from an S

There might be another hidden buffer somewhere (top of bottom), but Tesla actually allows you to go way below 2.8V to almost dead 0 of the battery so this right here is proof that there isn't much more buffer below the 3.5kWh...If you know how Li-ion batteries work. Maybe another kWh or so more hidden around at the bottom that we don't have access to, but not more.

Is this video some sort of fluke or is it normal?

No fluke. There are tons of videos with BMS readouts and other videos where people drive below 0km for 3.5-4kWh buffer size. Of course only when the BMS is perfectly calibrated.

It is sad to see so many disinformation in this forum...

 

[MP3Mike]

No fluke. There are tons of videos with BMS readouts and other videos where people drive below 0km for 3.5-4kWh buffer size. Of course only when the BMS is perfectly calibrated.

It is sad to see so many disinformation in this forum...

And there are plenty of reports, and videos, showing the car shutting down at/before 0. Both shutting down before and after 0 is a symptom of the BMS being out of calibration.

 

[TimothyHW3]

Just because it's off by 10% at 90% DOES NOT imply that its off by 10% at 10miles. There is NO evidence that is the case.

The inaccuracy could be linear, as in there is 30miles different, all the way down the scale.
The inaccuracy could be proportional to the discharge (i.e. inaccuracy degrades to zero)
The inaccuracy could be inversely proportional to the discharge, in that the inaccuracy increases as the measurement becomes smaller.
Then there is the 'Real' degradation of the battery to factor into the equation.

The BMS readout for the rated miles is pretty simple:

Full capacity (including the buffer) / typical constant (in the USA it is called rated apperantly, the straight line)

And it calculates from there - so it is actual pretty linear.

Once the battery gets into temperature and the BMS is settled, the dropout will be linear all the way to 0.
The BMS doesn't re-calibrate itself while you are driving, if it is off it is off. The only way to "calibrate it" is if Tesla resets it, or if it recalibrates itself after a charging (apperantly)

There are actually a bunch of people with BMS uncalibration, not sure why they intend to stick on pressing dislike instead of actually proving us wrong if they believe their BMS uncalibration is just a "display" thing?!

Just get your car out, drive for the BMS available 250 miles at 240Wh/m without stopping, shoot a video and see how far you go?

 

[AlanSubie4Life]

And there are plenty of reports, and videos, showing the car shutting down at/before 0. Both shutting down before and after 0 is a symptom of the BMS being out of calibration.

Do we have CAN bus data to support this? I'm very curious about how the readbacks look as one takes a properly calibrated BMS to 0 rated miles and beyond...

I'd be kind of surprised if someone hadn't taken a video of doing just this with CAN bus readbacks, at some point...

 

[TimothyHW3]

And there are plenty of reports, and videos, showing the car shutting down at/before 0.

Yes, and how is that contradicting what I said before that?!
Let me quote you and me again:

or many miles after 0% depending on how off the calibration is.

And my answer:
Many miles after 0% actually is a proof that the BMS is perfectly calibrated, not the opposite, since there is a 3.5kWh buffer below 0%.

But the other way around is true.

(the other way around is what you just said in this post - so you basically confirmed what I said in my previous reply)...

So to summarize:
If you can drive the 15 or so miles below 0 at rated consumption - BMS perfectly calibrated.
If your car shutsdown before 0% (5% or more for example) - BMS uncalibrated.\

If the car doesn't drive the rated miles at rated consumption from 0-96% - BMS uncalibration OR Battery degradation OR both

 

[TimothyHW3]

Do we have CAN bus data to support this? I'm very curious about how the readbacks look as one takes a properly calibrated BMS to 0 rated miles and beyond...

You don't need CAN data for the below 0. If the car goes to below 0 and drives as in these videos posted, meaning roughly 3-4kWh * by the consumption equals driven km/miles, then the BMS is perfectly calibrated because this is by Tesla design. Obviously, if the car performs as it was instructed to do, there can't be any BMS uncalibration - logic!

Bjorn Nyland has an old video with CAN data when his car stopped at about 15km left or something. It was BMS uncalibration...

 

[AlanSubie4Life]

You don't need CAN data for the below 0

Not sure why you say that...having the actual data showing it is depleting the buffer would be evidence that what you say is actually true. I'm just saying it would be nice to see it...and the data from the following recharge, showing the buffer getting replenished, followed by the available energy getting replenished...etc.

It may be that the CAN does not report the buffer being used, of course...as that parameter may be a statement of the TARGET buffer size, not the amount of energy left in the buffer (since the buffer energy is included in the energy available readback, and that presumably would be decremented, nearly all the way to zero, until the car shuts down). And the available energy readback would presumably NOT show zero energy remaining when rated miles reach 0. For a properly calibrated BMS.

If it is the target size only, you'd expect it to always say 3.3kWh or whatever, during the discharge, and during the recharge, you'd expect the rated miles to inch above zero during the recharge once the available energy readback goes above 3.3kWh. Etc. It would just be a demonstration of how the mechanics of the whole system of readbacks work and what they represent and how they get translated to the UI.

As you can see from above, I don't know exactly how it works which is why I'm curious to see the data, and draw my own conclusions.

 

[TimothyHW3]

Here another video where he drove about 10km under 0, which is just like Tesla intended it, but just stopped because he was scared.

 

[TimothyHW3]

Not sure why you say that...having the actual data showing it is depleting the buffer would be evidence that what you say is actually true. I'm just saying it would be nice to see it...and the data from the following recharge, showing the buffer getting replenished, followed by the available energy getting replenished...etc.

Well I actually send you a couple of screenshots already, just check them out...You can see that you go way below 0% and the buffer is being depleted.

Your tests of the "constant" you conducted also proof the existance of the buffer below 0, because otherwise you can drive from 0-100% at rated consumption and not your "calculated constant", which is below the rated straight line...

And why I say that you don't need CAN? Simple - logic and math.

In the video you and I posted the german guy actually does the math and calculates the buffer exactly without even having access to CAN by driving below 0 until the car shutsdown(Wh/km multiplied by the km driven - is the math for the buffer in the video)
He also tested how much kWh go into the battery, but unfortunately his measurements were wrong - he calculated the heat loss for charging too short and was off by 2kWh. But his general idea was pretty close without even having access to CAN - just basic math.

Also note that this was a brand new car - barely driven 500km with it, so no degradation and no BMS uncalibration.

 

[AlanSubie4Life]

Well I actually send you a couple of screenshots already, just check them out...

I can't find these in that conversation...sorry. Maybe it was another thread.

In any case I do think from what I've seen from CAN bus readbacks that I agree with you, and there is circumstantial evidence that it works as you say, including the discussions about constants that we have had (still waiting on more data from you on that... ), and the video you reference with the manual buffer calculation.

However, I've found in order to convince others, it's good to be able to say something other than "I insist I am right about this, it is just logic and math." Hard data from a dispassionate CAN bus readback can go a long way to convince people...

It's often good to reprove hypotheses in multiple ways. Makes them more likely to be solid theories.

As a specific example, do you know whether the CAN bus readback keeps the "buffer kWh" readback at 3.3kWh even as available energy goes to 0kWh? Or does it start decrementing and match "available kWh" as soon as you reach 3.3kWh available energy, and you continue to drive below zero rated miles (displayed)? Do you know how it behaves?

 

[TimothyHW3]

Here is Here is the screenshot.
You can see the -%. the energy buffer being "eaten" and its nominal value (3.50kWh is the buffer size and 3.2 indicates it is being eaten, also negative % SOC)

This also proofs that the buffer is below 0% and that 0-100% uses the nominal full pack including the buffer, but the actual rated miles/km don't account for it - hence the difference between the "typical rated constant" and the "real constant"

 

[tomas]

Just because it's off by 10% at 90% DOES NOT imply that its off by 10% at 10miles. There is NO evidence that is the case.

The inaccuracy could be linear, as in there is 30miles different, all the way down the scale.
The inaccuracy could be proportional to the discharge (i.e. inaccuracy degrades to zero)
The inaccuracy could be inversely proportional to the discharge, in that the inaccuracy increases as the measurement becomes smaller.
Then there is the 'Real' degradation of the battery to factor into the equation.

There are very few natural/physical properties that actually follow a linear path. Most are some form of S or U or other curved path.


In all probability, there are almost certainly multiple curved profiles of different factors, which are unique to each car, and it's current environment.

Also, don't mistake a small sample-set of measurements with a large body of data indicating a trend. i.e. difference between Weather and Climate.

Hmm. Lemme see. I’ve had one Tesla since 2012, and two others since 2018. Been through almost 10 total years of charge cycles. I have no technical proof, since I don’t have all the code and metrics. But I do have 10 years of experience that actual miles remaining converges with rated range as you get towards 0. And that’s exactly why Tesla says cycle between 10 and 100!% to recalibrate. BMS understands empty and full. Everything between is an estimate.

 

[AlanSubie4Life]

Here is Here is the screenshot.
You can see the -%. the energy buffer being "eaten" and its nominal value (3.50kWh is the buffer size and 3.2 indicates it is being eaten, also negative % SOC)

This also proofs that the buffer is below 0% and that 0-100% uses the nominal full pack including the buffer, but the actual rated miles/km don't account for it - hence the difference between the "typical rated constant" and the "real constant"

Makes sense, thanks. That's what I expected for a properly functioning BMS. I still don't understand all the different SOC numbers, but never mind. 3.2kWh is 5.4% (SOC avg) of 59kWh for example (makes no sense). Is this the SoC of individual bricks maybe? But the average doesn't make sense either...

SOC seems to be defined as -0.3/(76.9-3.5) = -0.41%

The buffer would go to zero at in above case at SOC of -4.78%. Maybe the other SOCs represent the actual true SOC (0% meaning bricked)? Oh well.

 

[AlanSubie4Life]

actual miles remaining

Are you talking about "actual miles remaining" on the Energy "Consumption" screen? (On that screen it is called Projected Range.)

If so, that makes sense since that value is based on the formula:

Charging Constant * Rated Miles Remaining / Current Selected Efficiency

So as rated miles goes to zero, projected range will also go to zero.

But I'm not sure what you meant by actual miles remaining.

 

[TimothyHW3]

Makes sense, thanks. That's what I expected for a properly functioning BMS. I still don't understand all the different SOC numbers, but never mind. 3.2kWh is 5.4% (SOC avg) of 59kWh for example (makes no sense). Is this the SoC of individual bricks maybe? But the average doesn't make sense either...

SOC seems to be defined as -0.3/(76.9-3.5) = -0.41%

The buffer would go to zero at in above case at SOC of -4.78%. Maybe the other SOCs represent the actual true SOC (0% meaning bricked)? Oh well.

Nevermind the SOCs. The SOC % is the calculated value from nominal remaining. Hence the negatives.

The other SOCs are from the full capacity, whereas the BMS gathers information and makes prediction - max min and avg.
And as far as I can see Tesla uses SOC min from full pack incl. buffer for its prediction just to be on the safe side.
Tesla's rational:
" We think we have this max and this min. We might have that max, but if we believe the min is our break point, let's use min just to be sure. We still might not have min at all(see post about the car shutdown before 5%), but if that is all the info we have we might as well use the minimum value"

The only SOC% you need from that screenshot are SOC and SOC min.

 

[VT_EE]

Nevermind the SOCs. The SOC % is the calculated value from nominal remaining. Hence the negatives.

The other SOCs are from the full capacity, whereas the BMS gathers information and makes prediction - max min and avg.
And as far as I can see Tesla uses SOC min from full pack incl. buffer for its prediction just to be on the safe side.
Tesla's rational:
" We think we have this max and this min. We might have that max, but if we believe the min is our break point, let's use min just to be sure. We still might not have min at all(see post about the car shutdown before 5%), but if that is all the info we have we might as well use the minimum value"

The only SOC% you need from that screenshot are SOC and SOC min.

Have they always had this usable buffer at the bottom or is this something that was introduced recently (within the last 6 months)?

 

[tomas]

Forever

 

[TimothyHW3]

Have they always had this usable buffer at the bottom or is this something that was introduced recently (within the last 6 months)?

Always, and it works like that on all Teslas.
AFAIk since the beginning, but at least since 2014/15.
And the buffer a) is different depending on the car battery capacity b) is being reduced with degradation(mine is now at 3.4 down from 3.5)

Like I said, there possibly is another hidden buffer below this one, but seeing the voltage around the end of this buffer, it can't be that much, maybe 1-2kWh.

This is why Tesla is actually using this buffer below 0% and is playing with it to make sure you never use it. The usable capacity, from 0-100% is without the buffer.

And this is why the rated range displayed while driving AT RATED/TYPICAL consumption can't be achieved from 0-100%, but to 104%(or if you go to below -4%), even though you would expect to be able to achieve it from 0-100% at the typical consumption.

What this means is that while driving at rated/typical consumption you can only drive ~96% of the 310 or whatever miles the car shows you depending on the model, unless you go below 0.

I explained this in this video.