10500 miles, 10% “degradation” already?

[hgmichna]

It takes 24 hours to fix 1mv of cell imbalance. If you have a friendly local Tesla ranger who will connect your car to toolbox and find out, or someone with a ScanMyTesla setup... you could know how imbalanced you are. With a 10% drop, you could easily be out of balance by 50mv. Your car needs to spend as much time as possible with the Battery SOC >85% and the car asleep. If you're in balance range for 10 hours a day, it could take you over 108 days to get to 5mv charge imbalance level.

Am I the only one whom this strikes as bad design? If you have to let the battery sit at 90% for 1080 hours, aren't you losing more to degradation than you gain through rebalancing?

Is there any good reason to design it that slow? I cannot think of one.

Is there any verifiable source for the rebalancing rate?

 

[smatthew]

Am I the only one whom this strikes as bad design? If you have to let the battery sit at 90% for 1080 hours, aren't you losing more to degradation than you gain through rebalancing?

Is there any good reason to design it that slow? I cannot think of one.

Is there any verifiable source for the rebalancing rate?

Normally, as soon a imbalance gets over 5mv the car starts trying to fix it. You always fix imbalances slowly. Tesla knows what they're doing.

However, if you never charge above 85% the car never gets a chance to fix the block imbalances.

You can use "ScanMyTesla" to view the amount of battery imbalance your car has.

The source for this data is the Tesla Model 3 Theory of Operations, part of the service manual.

Interestingly, "Battery University" gets this one wrong. They say Teslas use active cell balancing. Which would be superior. That's where instead of burning excess power in a tiny resistor, you move power from brick to brick. So the highest charged brick is used to bring the lowest charged brick up. Unfortunately, that's much more complex and Tesla has not decided to do that.

 

[camalaio]

Am I the only one whom this strikes as bad design? If you have to let the battery sit at 90% for 1080 hours, aren't you losing more to degradation than you gain through rebalancing?

Is there any good reason to design it that slow? I cannot think of one.

Is there any verifiable source for the rebalancing rate?

I've mentioned this elsewhere on the forum, see this quote:

I've written, rewritten, deleted, written again, and again deleted this post a lot. But one last try, in an uncharacteristically short summary for me:

Regarding balancing, we'd probably be best to forget about the doc we've seen. By my measuring, it is either incorrect or misleading. We don't know which parts are still (or ever were) true. Even if the doc is partly correct, imbalance is so incredibly unlikely that the imbalance scenarios I've tried outlining are plainly asinine.

If you are reading this, your battery is almost certainly properly balanced. If it is not, you're likely in talks with Tesla, not this forum.

I've spent a lot of time on this, and I think we honestly just don't know. The only thing that makes any sense is that it balances at higher SoC - where that threshold actually is, only Tesla knows right now (and the oft-quoted document does not seem correct). I can tell you now that I've been watching it, it's pretty conclusive that some sort of balancing happened when my car was sitting at 80% (and it also went beyond correcting to 5mV).

As for making it that slow, it's probably for accuracy. Putting any sort of load on a battery generally drops its voltage. When you're accounting for a few mV, you need to be very accurate. Using as small a load as possible to increase the accuracy of your voltage reading is best.

It also doesn't need to be fast. Catastrophic imbalance is indicative of an actual defect, and is not expected (especially if sudden). Only minor differences need to be dealt with on a day-to-day basis.

 

[smatthew]

I've mentioned this elsewhere on the forum, see this quote:

​

I've spent a lot of time on this, and I think we honestly just don't know. The only thing that makes any sense is that it balances at higher SoC - where that threshold actually is, only Tesla knows right now (and the oft-quoted document does not seem correct). I can tell you now that I've been watching it, it's pretty conclusive that some sort of balancing happened when my car was sitting at 80% (and it also went beyond correcting to 5mV).

As for making it that slow, it's probably for accuracy. Putting any sort of load on a battery generally drops its voltage. When you're accounting for a few mV, you need to be very accurate. Using as small a load as possible to increase the accuracy of your voltage reading is best.

It also doesn't need to be fast. Catastrophic imbalance is indicative of an actual defect, and is not expected (especially if sudden). Only minor differences need to be dealt with on a day-to-day basis.

I think part of the confusion is that only the lowest brick needs to be above 4.0 volts. Depending on the brick imbalance, the SOC could be significantly different. 85% is Tesla's approximation.
Also, documents indicate that it takes more than a 5mv imbalance to start balancing, but don't specify when it stops. I bet it goes until the car switches to drive mode.
One other thing - almost everybody is quoting a copy of a a document from April 18th of 2018. Everything in this car is controlled by computers. Tesla could have changed the parameters for rebalancing since then.

 

[hgmichna]

Thanks for the information!

At least we have good reasons to think that Tesla gets the balancing right.

Any info on balancing within a brick? Or are all cells connected in parallel within one brick?

 

[smatthew]

Thanks for the information!

At least we have good reasons to think that Tesla gets the balancing right.

Any info on balancing within a brick? Or are all cells connected in parallel within one brick?

All cells are in parallel within a brick.

 

[camalaio]

... almost everybody is quoting a copy of a a document from April 18th of 2018. Everything in this car is controlled by computers. Tesla could have changed the parameters for rebalancing since then.

This, exactly. It could have been true at some point or in some context we don't understand. But to me, it's clearly not accurate to today's behaviour with the context I have with access to the CAN data and as an owner. It was clearly never meant as guidance for owners, nor something for us to hold them to (or even need to be concerned about in the slightest, as there's not much we can do if we have a bad brick). Thus my rather strong assertion that we should just forget about it - it's not adding value to anyone right now, except to say that the Model 3 does in fact have some form of balancing and the details of how/when are a mystery.

 

[smatthew]

This, exactly. It could have been true at some point or in some context we don't understand. But to me, it's clearly not accurate to today's behaviour with the context I have with access to the CAN data and as an owner. It was clearly never meant as guidance for owners, nor something for us to hold them to (or even need to be concerned about in the slightest, as there's not much we can do if we have a bad brick). Thus my rather strong assertion that we should just forget about it - it's not adding value to anyone right now, except to say that the Model 3 does in fact have some form of balancing and the details of how/when are a mystery.

I don't think it's a good idea to disregard what we know about brick balancing just because Tesla might have changed something. If you want to assert that something is different I suggest you pay for 24hrs of service manual access and check the latest version.

 

[camalaio]

I don't think it's a good idea to disregard what we know about brick balancing just because Tesla might have changed something. If you want to assert that something is different I suggest you pay for 24hrs of service manual access and check the latest version.

I'm not USD$100 invested in this haha. I might pay for it one day, but I need a long list of things I want to find out before I'd ever do that. I don't even know what I'd find in there! And I'm against the whole concept of temporary access to such a thing (no idea if this is normal for the industry).

My main point was that the data I've got clearly contradicts it. Their 85% is my 72-73% in terms of voltage (apparently Bjorn Nyland's numbers match mine too, thankfully). This, to me, is actually comforting, since it means most of us wouldn't need to worry about charging any higher for balancing benefits. And it seems like I've witnessed balancing at just 80% (I don't see any sort of flag that concretely indicates balancing unfortunately). It's not even necessarily that Tesla changed something - what they refer to as "85%" might not be the same percentage we see on the screen and app (FWIW, it's still not even close if I remove the bottom buffer from the percentage calculation).

I don't see value in worrying anyone about balancing based on old, seemingly inaccurate data, especially when we have much more concrete things we can concern ourselves regarding range estimates and degradation. We've also yet to see a single example of degradation specifically due to imbalance (though I'd love to see this case if it existed!)

 

[smatthew]

I'm not USD$100 invested in this haha. I might pay for it one day, but I need a long list of things I want to find out before I'd ever do that. I don't even know what I'd find in there! And I'm against the whole concept of temporary access to such a thing (no idea if this is normal for the industry).

My main point was that the data I've got clearly contradicts it. Their 85% is my 72-73% in terms of voltage (apparently Bjorn Nyland's numbers match mine too, thankfully). This, to me, is actually comforting, since it means most of us wouldn't need to worry about charging any higher for balancing benefits. And it seems like I've witnessed balancing at just 80% (I don't see any sort of flag that concretely indicates balancing unfortunately). It's not even necessarily that Tesla changed something - what they refer to as "85%" might not be the same percentage we see on the screen and app (FWIW, it's still not even close if I remove the bottom buffer from the percentage calculation).

I don't see value in worrying anyone about balancing based on old, seemingly inaccurate data, especially when we have much more concrete things we can concern ourselves regarding range estimates and degradation. We've also yet to see a single example of degradation specifically due to imbalance (though I'd love to see this case if it existed!)

The problem is that the only way to know if there is an imbalance is to use ScanMyTesla and view CAN messages. But we know that lithium ion battery packs suffer from imbalances. So it seems prudent to me to make sure the car has the opportunity to remedy the issue. Especially since prevention consists of a) charge to 90% b) Let the car sit.

 

[djrodtc]

Well, I’m back with some news.

This week I didn’t charge the car at all, I’ve been draining the battery around 20% per day until today I got it down to 12% and started charging it.

Previous week I kept the car at 90%, the 100% estimate from the app was 458km / 284.58 miles.

Every day the battery was at a lower SOC, the 100% estimate was even lower, I saw it as low as 441km / 274 miles, that was when I was around 50% SOC

Today getting the battery down below 20% changed the 100% estimate up to 461km / 286.4 miles, biggest improvement I’ve seen since I started this experiment.

Right now the car is charging at 16A, in 18 hours I’ll get to 100% SOC and I will see what new estimate I get then.

 

[KenC]

Well, I’m back with some news.

This week I didn’t charge the car at all, I’ve been draining the battery around 20% per day until today I got it down to 12% and started charging it.

Previous week I kept the car at 90%, the 100% estimate from the app was 458km / 284.58 miles.

Every day the battery was at a lower SOC, the 100% estimate was even lower, I saw it as low as 441km / 274 miles, that was when I was around 50% SOC

Today getting the battery down below 20% changed the 100% estimate up to 461km / 286.4 miles, biggest improvement I’ve seen since I started this experiment.

Right now the car is charging at 16A, in 18 hours I’ll get to 100% SOC and I will see what new estimate I get then.

When your SOC is lower, the margin of error in the Rated Range increases. Good luck.

 

[djrodtc]

The car just charged to 100%.

The range is now...

460km / 285.8 miles

I’ve recovered about 10km, not breathtaking or anything...

TeslaFi reports that from 12 to 100% SOC I’ve added 62KWh

I started charging at 12%. If 100% is 75kWh, 12% is 6kWh.

6 initial +62 added = 68kWh battery

My battery has lost about 7kWh, so a little under 10% loss, more or less what I’ve been seeing from the beginning.

Also TeslaFi says only one car in my mileage is worse than mine, 99% of other cars in my mileage has about 485km / 301.3 miles.

I’m going to contact Tesla again...

 

[jmaddr]

You calculations above are based on Tesla allowing us to use the entire 100% for range...they don't. They leave room at the top and bottom to keep from destroying the battery. I'm not sure if anyone would know this tolerance, but if they save us 4% on the top and 4% on the bottom, you would have essentially zero. And that's assuming we even have a 75kWh (I've heard 72 and 73 often).

Here is my output from this year. Note the high is 309 and the low is 296. It's been in the 280s before that. The end take away is the BMS range is just an estimation. Look at the average and see how that declines over time.

 

[jmaddr]

Sorry, for some reason the pic was not uploaded. That little bump in early June was when I read it was good for cell rebalancing to charge to 90 and let it sit for a few hours. Normally I only charge to 70% (60% now during the pandemic)

 

[rhatguy]

Adding my stats to this thread. I have a 2018 AWD that I purchased used a bit less than a year ago. For months I charged to 80% every night and did very little supercharging. My projected range in teslamate increased marginally as the weather warmed up in Georgia as expected but overall had stayed relatively consistent around 290.

When Covid hit, I dropped my charge limit to 50% as I was not driving as much. Projected range stayed basically the same. As I started driving more, I returned my charge limit to 80%. Around that time, you can see on my graph a significant and fairly quick drop started occurring from around 289 down to around 275 now. I've recently bumped my charge limit back to 90% to see if any balancing might occur but haven't seen anything yet. I'm very curious what could cause such a significant drop (~15 miles over about 15 days) when the weather was basically unchanged. I haven't called Tesla yet, but thinking about it just to make sure something more serious hasn't happened.

 

[camalaio]

The car just charged to 100%.

The range is now...

460km / 285.8 miles

I’ve recovered about 10km, not breathtaking or anything...

TeslaFi reports that from 12 to 100% SOC I’ve added 62KWh

I started charging at 12%. If 100% is 75kWh, 12% is 6kWh.

6 initial +62 added = 68kWh battery

My battery has lost about 7kWh, so a little under 10% loss, more or less what I’ve been seeing from the beginning.

Also TeslaFi says only one car in my mileage is worse than mine, 99% of other cars in my mileage has about 485km / 301.3 miles.

I’m going to contact Tesla again...

Quick correction on the calculation: You mixed ideal capacity and your capacity ("if 100% is 75kWh"). This only works if 100% is only reported at 75kWh, but the percentage is relative to your battery's capacity.

If adding 88% added 62kWh, you have very roughly ~70.5kWh usable 100% capacity. That number from TeslaFi can be confounded in many ways unfortunately, but I'll take it as roughly accurate. At new, I'd expect about 74.3kWh is "usable" at 100% (this would rapidly decline for short time, but not very much).

With a 4.5% of total capacity bottom buffer (seems about right from other data), this means a total capacity of about 73.8kWh on your pack. Again, very roughly, and the number can be wrong in many ways.

Strangely, my total capacity is around 74.5kWh (just 0.7kWh more), but reports significantly more range at 100% than yours. This is very interesting to me, and I can't currently think of why this would happen. Gonna have to go pull some numbers and see if I am remembering anything wrong, or calculated something wrong. Additionally, the above numbers imply you've only lost 5%, which doesn't correspond to 460km@100% at all. Data from TeslaFi could just be a bit off, we're doing a lot of extrapolation here.

 

[djrodtc]

New update, I checked one list time before setting the UI to % and now the car is reporting 370km / 292 miles at 100%.

 

[jmaddr]

I’m hoping others read this and see that the mileage are just estimates. Even gasoline fluctuates 1% every 10 degrees, it’s just that you can’t tell 1% on a gas car gauge but you can tell 3 miles) on a Tesla. Set it to percentage and forget it. Heck, since Tesla shows 1/100 accuracy on percentage, or even worse, 1/300th on miles, even that shines a light that the BMS is just an estimate. The phenomenon is just that the data displayed by Tesla has SOOOO much more precision, it highlights the everyday inaccuracies of modern engineering compared to what we are used to. It will get better with time with the car.

 

[rhatguy]

Interestingly after changing my charge threshold to 90% I've seen a noticeable bump on the estimated range (as well as when dragging the slider to 100% in the app). Not sure what that means yet, but definitely interesting. I'll be leaving my charge at 90% for a while to see if it continues to return to the level it was before the sudden reduction.