2013 MS 85 upgrade to MCU2 then HV battery problem

[NV Ray]

hello all, I was reading those 5 pages of battery change and was wondering : when you have a original S85 and you change the battery pack. You take it new they now install a 90 battery pack. Then, you have like a S90 or a S85 restrained (meaning that if you charge up to 100%, in fact you don't charge at the max of the battery).

That's what happened to my 2012 P85 - ordered 85, 90 came in and installed. 90 corked (restrained) to 85. I decided to keep status quo - 90 kWh pack software limited (corked) to 85 kWh.

Decision made after feedback from posts in this thread Uncork 90 kWh HV Battery ($750) ? and phone conversations with @howardc64 , @Recell , and @wk057 .

For my driving I don't need the extra 15-20 miles range. Further, I like the feeling that I have a built in top buffer. Having a top buffer is similar to my Prius Prime with an 8.8 kWh Lithium Ion HV battery - when charge meter says 100%, State of charge is actually around 84%.

 

[wk057]

I'll note that there is some newer code (that I've not seen active in practice yet since I don't have a ton of data from active software limited cars) that appears aimed at keeping the capacity calculations accurate on software limited cars. Short cycling the pack all the time tends to throw off the math on that over time, and could result in things like lost range (miles under 0) and misreported range (real 0 is somewhere over 0).

This code looks like it will occasionally shift the real behind the scenes charge window of software limited cars towards the high end vs the low end in order to get data from a higher SoC starting point, meaning if you set it to 100% all the time, sometimes 100% will really be 100%.

It looks like it still won't go above the user's set point, though. So for example, if set to 90% it won't go above real 90%.

Overall this would seem to effectively make the capacity calculation behavior more on par with non-locked packs.

 

[NV Ray]

I'll note that there is some newer code (that I've not seen active in practice yet since I don't have a ton of data from active software limited cars) that appears aimed at keeping the capacity calculations accurate on software limited cars. Short cycling the pack all the time tends to throw off the math on that over time, and could result in things like lost range (miles under 0) and misreported range (real 0 is somewhere over 0).

This code looks like it will occasionally shift the real behind the scenes charge window of software limited cars towards the high end vs the low end in order to get data from a higher SoC starting point, meaning if you set it to 100% all the time, sometimes 100% will really be 100%.

It looks like it still won't go above the user's set point, though. So for example, if set to 90% it won't go above real 90%.

Overall this would seem to effectively make the capacity calculation behavior more on par with non-locked packs.

Interesting. So would corresponding range increase reflect your comment? In my case, have 90 kWh corked to 85 kWh. Currently 80% charge shows 217 miles. Would the newer code you mention cause that to increase to say 229ish emulating a 90 kWh at 80% SOC?

 

[wk057]

If there were a capacity estimate mismatch, then it's possible. I don't have a ton of data on the effects of this code just yet, but it's definitely geared towards making the range estimate more accurate.

217 @ 80% would be 271 at 100%, which is pretty much what a really good new "85" would have gotten (almost 80 kWh usable), so that looks normal for a new s/w locked pack.

 

[NV Ray]

If there were a capacity estimate mismatch, then it's possible. I don't have a ton of data on the effects of this code just yet, but it's definitely geared towards making the range estimate more accurate.

217 @ 80% would be 271 at 100%, which is pretty much what a really good new "85" would have gotten (almost 80 kWh usable), so that looks normal for a new s/w locked pack.

Thanks.

FYI. Assuming this is Mr. Hughes, you were quoted in this recent article about the growing trend of older Model S BMS_U029 errors. BMS_u029 Is the Code Killing Several Tesla Model S Battery Packs

 

[geordi]

@wk057 One more datapoint for you on this - when my 90 battery was very much corked back to 60, you (and some others) had made that suggestion that charging to 100% would actually be ~66% and clearly not have any issue on range. As I was on that road trip then, I needed the range and HAD TO charge to 100%. I did observe both the charging speed drop off as it approached that 100%/66% finish... AND the computer complaining at me that repeated charging to 100% would affect the range and nagging me to drop the total.

Thankfully I didn't have to endure that for very long, but I also seem to recall that I observed the claimed range drop by a mile. When it very much should not have, if the programming is to maintain that "available" battery size at whatever the locked size is by slowly shrinking the upper buffer as the battery ages.

So since we know that a locked pack has plenty of available capacity that isn't used, the reduction in range is entirely being reported by the computer when it doesn't have to. That's not nice of them to program it like that. Just report the capacity accurately!

 

[wk057]

I observed the claimed range drop by a mile. When it very much should not have, if the programming is to maintain that "available" battery size at whatever the locked size is by slowly shrinking the upper buffer as the battery ages.

So since we know that a locked pack has plenty of available capacity that isn't used, the reduction in range is entirely being reported by the computer when it doesn't have to. That's not nice of them to program it like that. Just report the capacity accurately!

The usable portion of a capped pack is a percentage of the real pack, not a specific kWh. So as the pack ages and loses capacity, the usable capacity will proportionally decrease. It won't use up the locked capacity to maintain the capped capacity.

Silly, but this is how they do it.

 

[howardc64]

The usable portion of a capped pack is a percentage of the real pack, not a specific kWh. So as the pack ages and loses capacity, the usable capacity will proportionally decrease. It won't use up the locked capacity to maintain the capped capacity.

Silly, but this is how they do it.

Curious if capped pack will also limit BMS on triggering rebalance? Brick voltages will be lower in the capped pack's "cork scaled SOC" compared to uncapped pack at same SOC.

 

[geordi]

That's the stupidest programming logic. That means that any perceived "buffer" on the pack versus the rating is meaningless, where they COULD have chosen a more tech-like logic of spare capacity utilized to blunt any capacity loss from degradation. That's how SSD manufacturers do it, a 1 TB drive is actually more like 300GB larger of spare sectors, which remain reserved until and unless cellular losses are detected and wear-levelling calls the reserve into use.

The capacity remains XYZ purchased capacity for considerably longer based on this.

 

[MP3Mike]

That's the stupidest programming logic. That means that any perceived "buffer" on the pack versus the rating is meaningless, where they COULD have chosen a more tech-like logic of spare capacity utilized to blunt any capacity loss from degradation.

Why would they do that? That would give you an "unfair" advantage over non-locked packs, and would essentially preclude them from selling you the unlock. (If it has already given you almost all of the extra capacity as your pack degraded, then it would essentially already be unlocked.)

 

[cloneguru]

View attachment 842378

Some data to go with my note above. (Please excuse the crudity of this model. I didn't have time to build it to scale or paint it.)

This is data from one of 057's customer vehicles in the extended service plan program (shared with permission). A very early 2013 S. This is data from a couple of weeks ago. Our monitoring device, using the data above along with other data used internally on the device, determined that the car was going to eventually throw some errors related to an SoC imbalance. The customer was notified, and they were actually already in the middle of a small road trip.

"Imbalance? But wk057, that pack looks very well balanced! Only off by 5mV!" Ah yes, the myth of voltage being the end-all-be-all of balancing. NOPE.

Check out the capacity of some of the groups in the front hump. Notice anything? One of these things is not like the other....

Other data suggests this is the "standard" moisture issue with these older packs.

Sure enough, near the end of their trip, the car threw the "Maximum Charge Level Reduced" error, as the SoC imbalance grew beyond the BMS's ability to handle. (Fortunately, all is well and they managed to make it home. The car is en-route to us for a pack replacement.) Letting the car sit for weeks can temporarily resolve some of the presentations of this error by giving the BMS time to catch up with the capacity delta, but with use it will always end up back in a position where it can't keep up and will have to severely limit charge levels and discharge rates.... right up to the point where you can't drive anymore.

To Tesla's credit, in much older firmware you would have just been left stranded instead of the BMS just reducing usable capacity and power.

May I ask what software/device was used to gather this info and generate these reports?