HV Battery Health Test Confusion

[Nonlin]

This seems to have happened for my car after it installed 2023.44.1 on Dec 03, 2023. That update seems to have impacted the CAC reporting . CAC has been off ever since then. I asked Amund Børsand from SMT but at the time he was waiting for that update on his car. Haven't heard anything else since. Has anybody else observed that?

That seems really alarming if updated can fudge the results like that.

 

[AAKEE]

This seems to have happened for my car after it installed 2023.44.1 on Dec 03, 2023. That update seems to have impacted the CAC reporting . CAC has been off ever since then. I asked Amund Børsand from SMT but at the time he was waiting for that update on his car. Haven't heard anything else since. Has anybody else observed that?

My is of on my MSP 23 as well.

 

[AAKEE]

That seems really alarming if updated can fudge the results like that.

Not at all.

The car have everything under control, it is just Scan My Tesla that have to be updated.
The grabbing of data from the cambus most be updated to get the correct values.

 

[Sti-g]

Balanced somewhere in the middle SOC.View attachment 1010000

This makes sense. Cell balancing will only ever bring back capacity if a given cell somehow dissipated more charge then others. I'm not sure if that can happen. But if the capacity of a single cell has been reduced due to difference in aging between cells, then you are really stuck with that. Whether you balance at top or bottom, you will have a problem at the other end.

Based on this, is the battery capacity calculation that we are using (from the energy screen) really correct? Or would it be better to use lowest CAC and calculate battery capacity based on that?

 

[Steve446]

Based on this, is the battery capacity calculation that we are using (from the energy screen) really correct? Or would it be better to use lowest CAC and calculate battery capacity based on that?

There are several ways of estimating capacity and I guess they all work according to your preferred poison. We should just remember that the data all comes from the same place though, it's just expressed in different ways then rounded: however, the more ways you express the data the better the solidity of the data.

I'll prefer to let the experts, who may have other insights, to comment here.

For info, here's the data comparing CAC Average and NFP from SMT for my car (CAC until it stopped working on Dec 3 2023 after 2023.44.1). There's a good correlation between NFP and CAC average.

 

[AAKEE]

This makes sense. Cell balancing will only ever bring back capacity if a given cell somehow dissipated more charge then others.

We do not actually balance to give back capacity. The imbalance most probably is not even a part of the capacity (incl. Max range displayed).

The capacity is CAC x nominal voltage, so a cell being lower in soc do not change the battery capacity estimate. The capacity is there.

Theres a common misconception about
”balancing the BMS”.
That never happens.

We can ”calibrate” the BMS. The calibrate term is a forum invented term.
What we do is showing the battery capacity for the BMS, helping it to better calculate/estimate the battery capacity.

 

[ran349]

Based on this, is the battery capacity calculation that we are using (from the energy screen) really correct? Or would it be better to use lowest CAC and calculate battery capacity based on that?

The BMS determines the battery kWh capacity and it is given as the Nominal Full Pack value from the CAN bus reading.
The energy screen calculation gives you the same value, although because it contains uncertainty errors, it may be slightly different than the SMT reading.

So the energy screen is a very good and simple way to get a good capacity value, as long as it is done at a high battery SOC level to minimize the errors.

 

[Steve446]

The BMS determines the battery kWh capacity and it is given as the Nominal Full Pack value from the CAN bus reading.
The energy screen calculation gives you the same value, although because it contains uncertainty errors, it may be slightly different than the SMT reading.

So the energy screen is a very good and simple way to get a good capacity value, as long as it is done at a high battery SOC level to minimize the errors.

Nicely said. The energy screen is enough for an occasional update on battery capacity. However, if you want to look more deeply into the car's data, and over the whole battery operating range, then SMT becomes a more useful tool.

 

[AAKEE]

From when the time when SMT did’nt confuse the numbers.
Also a example of a battery with very similar cells.

At that time also daily charged to 55%, with a normal imbalance of 4mV.
Here on the picture in the end of a supercharhing session, just reached 100%.

 

[KenC]

The BMS determines the battery kWh capacity and it is given as the Nominal Full Pack value from the CAN bus reading.
The energy screen calculation gives you the same value, although because it contains uncertainty errors, it may be slightly different than the SMT reading.

So the energy screen is a very good and simple way to get a good capacity value, as long as it is done at a high battery SOC level to minimize the errors.

If you look at your service menu, you can get a more accurate BMS estimate, because there, they give you the SOC to the tenths, so less rounding error. As you can see, 156mi divide by 50.2% is 310.8miles, and the error is now +/-0.6miles. SMT shows my Full pack is 76.3kWh and my Range is also 310miles. So, they match.

I just took this SMT reading because I came back from skiing yesterday and ran my battery down to 5%! When I left the ski resort, the projection was for 1.2%, but I knew I'd do a little better than projected! Whew! I was wondering if running it that low would affect the BMS calculation! Nope, it's the same.

 

[Sti-g]

The BMS determines the battery kWh capacity and it is given as the Nominal Full Pack value from the CAN bus reading.
The energy screen calculation gives you the same value, although because it contains uncertainty errors, it may be slightly different than the SMT reading.

So the energy screen is a very good and simple way to get a good capacity value, as long as it is done at a high battery SOC level to minimize the errors.

My point was if energy(or SOC) per OCV was linear, capacity would be limited by how much charge can be stored by the weakest cell to avoid too high/too low charge. But I don’t think it works that way due to non-linear relationshi of SOC and OCV.

 

[Sti-g]

We do not actually balance to give back capacity. The imbalance most probably is not even a part of the capacity (incl. Max range displayed).

The capacity is CAC x nominal voltage, so a cell being lower in soc do not change the battery capacity estimate. The capacity is there.

Theres a common misconception about
”balancing the BMS”.
That never happens.

We can ”calibrate” the BMS. The calibrate term is a forum invented term.
What we do is showing the battery capacity for the BMS, helping it to better calculate/estimate the battery capacity.

I didn’t say what I was thinking correctly. I meant that due to imbalance you won’t be able to use the full capacity of the battery because you cannot charge to 100% due to imbalance.

Perhaps the fee mV imbalance is not very critical at low SOC since the curve is pretty sharp at that point (ie, less energy). balancing at the top would make more sense given that there’s 4.5% buffer at the bottom.

PS. I understand there is some misuse/overuse of BMS and balancing and calibration etc.

 

[Steve446]

If you look at your service menu, you can get a more accurate BMS estimate, because there, they give you the SOC to the tenths, so less rounding error. As you can see, 156mi divide by 50.2% is 310.8miles, and the error is now +/-0.6miles. SMT shows my Full pack is 76.3kWh and my Range is also 310miles. So, they match.View attachment 1010748

I just took this SMT reading because I came back from skiing yesterday and ran my battery down to 5%! When I left the ski resort, the projection was for 1.2%, but I knew I'd do a little better than projected! Whew! I was wondering if running it that low would affect the BMS calculation! Nope, it's the same.
View attachment 1010749

Nice, thank you. You can also find battery temperature in the HV battery and HVAC coolant cct sections of service mode. The temp seems to correlate broadly with SMT, good enough for many people probably.

I imagine that it all depends on where the data comes from, SMT data=BMS reporting I guess; but, given that the values I saw are not exactly the same, the difference could either come from rounding or separate, independent temperature measurements or cct in/out measurements - its also notoriously difficult to match separate readings even if the probes were calibrated initially. I get nervous playing around too much in the service mode though...

 

[Steve446]

I just took this SMT reading because I came back from skiing yesterday and ran my battery down to 5%! When I left the ski resort, the projection was for 1.2%, but I knew I'd do a little better than projected! Whew! I was wondering if running it that low would affect the BMS calculation! Nope, it's the same.

Yes, and for an individual trip I also find it's very reliable right down to 0% on screen, temperature permitting. It's only after a few weeks, when the BMS gets frustrated and wanders a bit (~6-8), if you follow the low SoC principle, that the real energy value might be a little bit different to that previously observed. Isn't that what the Mod calls the rabbit hole?

 

[Nonlin]

Thought they recent update was interesting.

This is exactly what I'm talking about.

I wonder if I do the health test now if it'll be more accurate now that they changed how they assess things.

 

[KenBlub]

If you look at your service menu, you can get a more accurate BMS estimate, because there, they give you the SOC to the tenths, so less rounding error. As you can see, 156mi divide by 50.2% is 310.8miles, and the error is now +/-0.6miles. SMT shows my Full pack is 76.3kWh and my Range is also 310miles. So, they match.View attachment 1010748

I just took this SMT reading because I came back from skiing yesterday and ran my battery down to 5%! When I left the ski resort, the projection was for 1.2%, but I knew I'd do a little better than projected! Whew! I was wondering if running it that low would affect the BMS calculation! Nope, it's the same.
View attachment 1010749

2023 Model 3 LR. used Ken’s method of calculating range by using the service mode SOC. Miles / SOC and got 330 miles. That is the same 100% range Tessie gives me. But…when I use Alan’s formula of mileage * Wh/mi / SOC I get 77.4 as my capacity. 77.4 / 78.8 does not equal 330 / 333 as far as % degradation. I’m just trying to learn so thanks. I’m not obsessing over the battery. Really.

 

[AlanSubie4Life]

2023 Model 3 LR. used Ken’s method of calculating range by using the service mode SOC. Miles / SOC and got 330 miles. That is the same 100% range Tessie gives me. But…when I use Alan’s formula of mileage * Wh/mi / SOC I get 77.4 as my capacity. 77.4 / 78.8 does not equal 330 / 333 as far as % degradation. I’m just trying to learn so thanks. I’m not obsessing over the battery. Really.

Depends on how high the SOC was, even with fractional SOC % from service mode or the new energy screen tabs.

Could be the degradation threshold. Hard to know without more info but it could be 78.1kWh or so. No real info about this pack.

 

[robertvg]

Is it anywhere formally stated by Tesla (on their site?) that the HV battery test from the service menu and resulting SOH number can be used for warranty claims, or is it just considered as an indication of the actual health of the battery?
Is there any other test they do (at a SC?) when this test gives a to low SOH number?

 

[AAKEE]

Saw a post a few days back from a model Y with the smaller LG pack (74.5 kWh FPWN).

The service mode battery test showed 97% health, but the nominal full pack is around 69.7 kWh (established via a picture of the energy screen).
Nominal full pack / FPWN —> 6.5% degradation.

Theres a huge delta and from what I have seen before Tesla didn’t use degradation thresholds on the smaller LG packs, they seem to vary the range when new like there where no threshold.

Anyway, the service mode battery test seem to lie to us, at least about thecstarting point. Its seem to be a huge gap to the delivered EPA test energy.

 

[Nonlin]

Tesla forces another class action from customers to go to arbitration

Tesla has successfully managed to kill another class action lawsuit from customers by forcing them to go to arbitration. It’s...

electrek.co