Battery Imbalance after 8 years

[wk057]

Lowest imbalance is always near the center of the SoC range, because the delta-voltage at those SoCs is much lower vs near the ends of the range. For example, 3.1V to 3.6V (500mV change) might represent 0-20%, but 3.6V to 3.9V (300mV change) might represent 20-80% (not exact figures, just illustrating). So usually what you'll see is an imbalance be exaggerated near either end of the SoC range.

Not sure where the folks get their info from, but the permitted delta is based on a lot of factors and varies substantially. There's no fixed allowed delta anywhere below 100mV, but there can be disallowed deltas as low as 20mV in extreme cases. Generally 60-100mV is where the BMS is likely to be picky.

A brick that has a lower voltage at high SoC and higher voltage at low SoC is likely to be a stronger brick. Higher voltage at high SoC and lower voltage at lower SoC, likely a weaker brick.

Again, voltage is only part of the picture. The BMS doesn't really expose a lot of the data needed to paint a better picture of pack health without additional coercion.

Overall, I usually suggest just letting the BMS do its job and not dwell on any particular metric.

 

[vanwho]

Loving this "new" refurbished battery Passes 100mV cell imbalance often. Lowest I see it is 40mV, often 70s and 80s. Has over 56000kWh charge total, more than 24000kwh DC charging. Jan 2015 pack. My failed pack was a Mar 2015 pack with 30k kWh total charging, 8k kWh DC, usually read about 10-15 mV imbalance, 40mV max I saw sometimes.

 

[kirkusss]

Have you tried to see how it does if you discharge it close to 0 and charge to 100%? From what you said before, even the capacity on the “new” one is lower than your failed pack. Have you told them this? At least in North America the warranty states that the new pack should have equal or better capacity.

 

[wk057]

Yikes. 100mV in the center of the SoC window (~3.5-3.7V) is pretty bad usually.

 

[David99]

Interesting discussion. I guess people dwell on balance so much because it is the only number that we can see easily. Other factors, for example how much the voltage sags under a certain load, would be very telling about battery health, but it's a number we can't easily judge because we don't have an easy way to compare an old to a new one. One would have to capture a lot of data and record exactly how much voltage drop happens under a specific load at a specific SoC and battery temperature. Heck I even saw it makes a difference how long ago I had charged my battery given the same SoC.

 

[kirkusss]

From what I saw on my 85 pack (Jan 2015 build) and a 90 pack (Ludicrous enabled) end of 2015, voltage sag is very similar and the entire pack voltage drops close to 250V below 70% SOC and cells in the 2.5-2.6 V range. Not sure anymore that this is a valid pack health metric. On 85D this drop triggers power limits if pushed a bit harder due to the lower amperage permitted, p90d does not suffer from this. Cells unbalance can go up to 400mV under WOT, but sits nice and chill at 6-12mV stationary. Tried the limits on my pack and the only thing that gave up are the tires.
I bet the 100 and the new 350V 90 don't suffer from this as they have a different parallel distribution.

 

[brainhouston]

Loving this "new" refurbished battery Passes 100mV cell imbalance often. Lowest I see it is 40mV, often 70s and 80s. Has over 56000kWh charge total, more than 24000kwh DC charging. Jan 2015 pack. My failed pack was a Mar 2015 pack with 30k kWh total charging, 8k kWh DC, usually read about 10-15 mV imbalance, 40mV max I saw sometimes.

Interesting though that this pack is operating without errors but your failed one had good numbers and still failed...
Sure hope it'll last. Let us know how it does as time goes.

 

[5JWX]

Does anyone know what a normal cell difference would be while driving underload? I read above up to 400mv is the typical or the maximum??

 

[Wim_V]

Does anyone know what a normal cell difference would be while driving underload? I read above up to 400mv is the typical or the maximum??

You cannot use SMT to read imbalance under load. There is way too much delay between the measurements of the different bricks.

 

[brainhouston]

the only way i found its somewhat useful is to look at BMS tab
it refreshes slow but u can tell by the color (orange mostly) if its roughly in balance (try to do steady load)

also, i've noticed that any other tab including my own will not display anything close to reality while driving (large imbalance) until u go to BMS tab n it refreshes
then for a second, all other tabs show correct imbalance...
i think it might be doing some long averaging including regen so i always see 150-250mV imbalance while driving
but on BMS tab its around 30mV while driving...

 

[howardc64]

Finally successfully triggered rebalancing visibly showing up on SMT BMS/Battery tab. Here is a summary

- tried charging to 95% on multiple occasions, even close to 100% (and bled it off afterwards). No change, always ~22mV imbalance at 70% SOC (74k miles on original RevB 85 pack)
- car has been cycling mostly 50-70% SOC for years. Nearly all short trips.
- finally read this article ( link ) which was quite illuminating along with the imbalance vs SOC plots from my battery ( post #60 ). Here is a guess
- the plots show imbalance is not linear vs SOC. Mine was lowest at low SOC and 70% but rises away from 70% in both slightly higher and lower SOC
- Based on this non linear pattern, I'm guessing BMS probably wants to see the entire SOC range before deciding to rebalance?

Finally now seeing 16-17mV at 70% SOC after the following sequence

- Let the car sit without being charged. Remaining parked at each 10% SOC decrements. I don't drive much, this took like 2 weeks.
- Finally nearing 20% SOC a couple of days ago and <=- 50miles remain (85 pack) acceleration limit bars shows up and car eventually said plug in to charge at ~40mi remaining. Decided to not push my luck further given recent spade of older pack failures if you breath on it haha.
- Charged it up to 50% SOC 2 days ago. Was planning on doing +10% increments all the way to above 90% SOC
- 60 mile trip came up earlier today. Charged it to 95% SOC 2 hrs before trip. Saw imbalance was ~20mV @ 95% SOC which was surprisingly low compared to prior measurements. Eventually settled at 17mV at 70% SOC. Might still be going down if rebalancing is active.
- Am thinking to charge up to 95% again soon, wait 2 hours, bleed off 20%

Anyhow, quite interesting. Doesn't seem to be just a matter of charging to a specific SOC %. BMS seems like it wants to see the entire SOC range don't have major problems before trigging rebalancing effort.

Another interesting coincidence is SMT now connects to the car instantly at all times. Previously had to launch Tesla app and flip to say Climate screen to get the car woken up enough for the BT dongle talk to SMT. Now it seems always on even though the car is quietly "asleep" haha. Maybe some relationship to observed rebalancing?

Will report back if the imbalance keeps on dropping and perhaps do another all 96 brick imbalance graph from 90-20 SOC. All of this does take some time to complete haha

I figure a better balanced pack can't be bad. Of course is orthogonal to

- water ingress (my fuse cover seal looks pretty good with a garaged car)
- dendrite bridge formation. Not sure any signs of that... maybe just reflected as part of the increased resistance but not sure how an end user can see that. Perhaps Tesla's supercharge rate reduction is a reflection. No idea.

 

[cousin_IT]

(We're actually considering exposing an estimated date range for this on the BMD dashboard for customers once enough data is collected to assess the trends.)

Hope the BMD will be released soon. I'd like to get a better understanding of my battery's health now that I have about 6 months of warranty left.

 

[Snerruc]

Just got my Raven back with a rebuilt pack- brick failed during charging. Tesla charged the new pack to 99% and it sat overnight until I could pick it up. Apparently this is standard procedure. Also, whenever we have a hurricane, Tesla blasts an email to charge to 100%. I’m not sure Tesla thinks 100% is as big a deal as we do. Since damage to the pack following their recommendation or practice would be on them warranty or not, perhaps 100% is more conservative than we think.

 

[jensk2]

Finally successfully triggered rebalancing visibly showing up on SMT BMS/Battery tab. Here is a summary

- tried charging to 95% on multiple occasions, even close to 100% (and bled it off afterwards). No change, always ~22mV imbalance at 70% SOC (74k miles on original RevB 85 pack)
- car has been cycling mostly 50-70% SOC for years. Nearly all short trips.
- finally read this article ( link ) which was quite illuminating along with the imbalance vs SOC plots from my battery ( post #60 ). Here is a guess
- the plots show imbalance is not linear vs SOC. Mine was lowest at low SOC and 70% but rises away from 70% in both slightly higher and lower SOC
- Based on this non linear pattern, I'm guessing BMS probably wants to see the entire SOC range before deciding to rebalance?

Finally now seeing 16-17mV at 70% SOC after the following sequence

- Let the car sit without being charged. Remaining parked at each 10% SOC decrements. I don't drive much, this took like 2 weeks.
- Finally nearing 20% SOC a couple of days ago and <=- 50miles remain (85 pack) acceleration limit bars shows up and car eventually said plug in to charge at ~40mi remaining. Decided to not push my luck further given recent spade of older pack failures if you breath on it haha.
- Charged it up to 50% SOC 2 days ago. Was planning on doing +10% increments all the way to above 90% SOC
- 60 mile trip came up earlier today. Charged it to 95% SOC 2 hrs before trip. Saw imbalance was ~20mV @ 95% SOC which was surprisingly low compared to prior measurements. Eventually settled at 17mV at 70% SOC. Might still be going down if rebalancing is active.
- Am thinking to charge up to 95% again soon, wait 2 hours, bleed off 20%

Anyhow, quite interesting. Doesn't seem to be just a matter of charging to a specific SOC %. BMS seems like it wants to see the entire SOC range don't have major problems before trigging rebalancing effort.

Another interesting coincidence is SMT now connects to the car instantly at all times. Previously had to launch Tesla app and flip to say Climate screen to get the car woken up enough for the BT dongle talk to SMT. Now it seems always on even though the car is quietly "asleep" haha. Maybe some relationship to observed rebalancing?

Will report back if the imbalance keeps on dropping and perhaps do another all 96 brick imbalance graph from 90-20 SOC. All of this does take some time to complete haha

I figure a better balanced pack can't be bad. Of course is orthogonal to

- water ingress (my fuse cover seal looks pretty good with a garaged car)
- dendrite bridge formation. Not sure any signs of that... maybe just reflected as part of the increased resistance but not sure how an end user can see that. Perhaps Tesla's supercharge rate reduction is a reflection. No idea.

Hi Howard

Claid you could trigger balancing on your car, which appeared to need it

But, 70mV imBalance is NOT a problem if the Voltage ImBalance is not on the very same string/brick that have an Ah imbalance, which I guess is why I have failed to trigger rebalancing on MY car?

Here is my Model S 70D Oct 2015 at 87.000 km.

- Voltage imbalance: 30 mV,
- Min Voltage: 3,51V
- Max Voltage: 3,54V
- Avg Voltage: 3,52V

- CAC imbalance: 4,30Ah
- CAC Min: 221Ah
- CAC Max: 225Ah
- CAC Avg: 223Ah

(CAC is Calculated Amp-Hour Capacity)

Currently ScanMyTesla Beta does not indentify the String/Brick consistently, so it Claims my Lowest Voltage String is #111 which is a little high given that my Model S 70D only has 84 strings

Untill I saw the CAC, I was surprised that my Model S did NOT balance when my average imbalance voltage was way higher than what people brag about on the internet. But Tesla BMS appears smarter than that! Why would it Voltage Balance, potentially in conflict with the Amp-Hour balance? And my AmpHours imbalanced is 225Ah / 223Ah = 1,008 ~ 0,8% and my Voltage imbalance is 0,030V / 3,52V ~ 0,85%?

'Fixing' the Voltage imbalance ONLY makes sense if the too low Voltage is on the very same too low amp-hours cell. The imbalance is - as well - so neglible, that balancing makes little sense.

Seeing the new Tesla/SMT CAC numbers is nice, I can mmediately calculate, that the minimum brutto capacity of my pack is 221Ah * 3,6V * 6 strings * 14 Modules == 66,84kWh. This can be compared to the 71,2 kWh from new so capacity remaining is around 66,84kWh/71,2kWh or 93,8%

 

[jensk2]

Seeing the new Tesla/SMT CAC numbers is nice, I can mmediately calculate, that the minimum brutto capacity of my pack is 221Ah * 3,6V * 6 strings * 14 Modules == 66,84kWh. This can be compared to the 71,2 kWh from new so capacity remaining is around 66,84kWh/71,2kWh or 93,8%


View attachment 956730

Having slept on above numbers

If Tesla had Active Balancers my total capacity would be the Avg Amp-Hours times 3,6V or 223Ah *3,6V * 6 Strings * 14 Modules = 67,44kWh
That is only 0,6kWh more or 1%.

I plan to make an Imbalance Graph to determine the SoC my battery was last balanced to. My guess is that it is the average used SoC, because of this Salomon logic:
- If Bottom Balanced, Tesla would throw away total capacity, because Amp-Hours at high voltage provide more energy than at low voltages
- If Top Balanced, strong cells will be unnecessarily stressed and will cycle in an unecessary high range (On the other hand, they are stronger so who cares)
- If Average SoC balanced, all cells are evenly stressed and people like me, will be able to optimize/minimize wear, say by staying below dV/dAh voltage peak at around 55%

I could fear however, that Maximum Range (Top Balance) would be used, because of Marketing.

Supporting my Average SoC Balance would - for my Car - only reduce total capacity with less than 0,53% based on this simple estimate:
A: Average string capacity can only be raised by 0,8%
B: An Amp-Hour at 4,2Volt is 4,2/3 better than an Amp-Hour at 3,0V.
So potential is around 4,2/3 * 0,8V * 0,5 = 0,53%
(0,5 factor mimics, that all the better cells are evenly distributed between Avg-Amp-Hours and Max-Amp-Hours. That is surely not the case, they are likely Gausian or Bell Curve distributed around Avg-Amp-Hours, s0 the factor is far less than 0,5)

If my 'Big 92% Cycle' triggers a ReBalance after next Charge, I sure hope that my October 2015 Model S does not suffer from Stuck CMOS issues syndroms! THAT would be a big bommer, given that I have nursed my battery for years, by AC Charging small Cycles around 40% SoC and Parking with less than 55%

 

[jensk2]

Having slept on above numbers

If Tesla had Active Balancers my total capacity would be the Avg Amp-Hours times 3,6V or 223Ah *3,6V * 6 Strings * 14 Modules = 67,44kWh
That is only 0,6kWh more or 1%.

I plan to make an Imbalance Graph to determine the SoC my battery was last balanced to. My guess is that it is the average used SoC, because of this

Here is an unplanned imbalance graph. Badly planned, because the two outlier high samples at 51,2% and 58,4% clearly was taken right after slowfox charging with 11kW and so reads too high. ((Results should have been taken at Stable Rest Voltages :-(

But I think the graph show sthat the balance is best at the center of my cycles which is around 35% - 40%? (Which get much more pronounced if I 'correct' the two outliers)

That ofcourse does not mean, that Tesla balances for my Average, because same pattern will be the result, if Tesla Balances whenever BMS detects significant imbalance, And I have deprived the BMS from measurements outside my normal SoC

But newer the less - Well Done Tesla!

 

[Zuikkis]

What do you think about this?

2016 MX P90DL, with the 1088792-00-A pack. 145000km on odo. Car is (slowly) charging when I took these screenshots, but it doesn't dramatically change the results.

Nominal is quite low 67.6kWh.

At first look battery looks quite unbalanced (90mV) but then on the BMS page you can see that it's only single cell that shows 70mV higher voltage. If you take that cell out all the rest are really well balanced actually. Is it possible this is some kind of measurement error which BMS can work around somehow?

It was already exactly like this on year ago, with 30000km less mileage. The very same cell was +70mV.

Car charges and supercharges fine. I have charged it a few times to 100% but I have never logged the voltages. I'll have to try that.

 

[meomyo]

I WISH TESLA showed this in the car screen

 

[Russell]

I WISH TESLA showed this in the car screen

Are you referring to this?



For about $50 you can have it on any phone or tablet
https://www.amazon.com/NEXAS-NexLink-Bluetooth-Diagnostic-Enhanced/dp/B08BPC179W/ + https://www.aliexpress.us/item/3256803492827610.html + Scan My Tesla app.

 

[jensk2]

What do you think about this?

2016 MX P90DL, with the 1088792-00-A pack. 145000km on odo. Car is (slowly) charging when I took these screenshots, but it doesn't dramatically change the results.

Nominal is quite low 67.6kWh.

At first look battery looks quite unbalanced (90mV) but then on the BMS page you can see that it's only single cell that shows 70mV higher voltage. If you take that cell out all the rest are really well balanced actually. Is it possible this is some kind of measurement error which BMS can work around somehow?

It was already exactly like this on year ago, with 30000km less mileage. The very same cell was +70mV.

Car charges and supercharges fine. I have charged it a few times to 100% but I have never logged the voltages. I'll have to try that.

If you opt in on the ScanMyTesla Beta (in Google Play AppStore) and wait for approval, then installs the downloaded update twice (My first install reported unchanged App version, but added Front Motor Signals, so was somewhat newer than the Releas. My second install totally changed APP UI and Version and after a drive and Phone connected 20 afterwards, it reported the CAC (Calculated Amp-Hour Capacity) Min/Max/Avg. SMT has problems identifying WHAT Brick is Min, but if your Min, Max, Avg is skewed say 200, 213, 215, then only one (or a few Bricks) have very low value. If that is the case then that brick will vary more in Voltage when SoC is changed and the imbalance will be high probably in th ehigher SoC range, given that - by nature or algorithm - balancing gets best at you average SoC, which is lower.

Dislaimer: I have NO knowledge - yet - about whether the BMS actually compensates for CAC Imbalance and so cause FURTHER INCREASED mV imbalance, but given it calculates CAC and knows mV imbalances, it is feasible

BTW: My Nominal does NOT match the CAC Capacity! Nominal kWh was reported as 64.7 - 65,0 kWh and my CAC calculation argues that Total is 66,7 kWh. I will however assume that my Car just increased my nominal, because TeslaMate just jumped Projected Typical Range from 342 km to 350 km.

Tesla appears to report Typical Range as Nominal kWh / 0,190 kWh/km (all other possible formulas performed on my OBD2 Signals give larger StdDev over multiple SoC values, than that formula, which is always spot on
And
- 242 km * 0,190 kWh/km== 64,98 kWh (== Old Nominal)
- 250 km * 0,190 kWh/km == 66,5 kWh (== Assumed new Nominal)

So potentially SMT now reports a higher Nominal now after my less than bi-yearly charge to > 90% and discharge to < 20%. I will post results in this thread!

(350 Typical Km is higher than ever recorded by TeslaMate - I hope it is not an outlier!