Help! One more time...roadster 1.5 battery question

Hi Marc,

I know your life would be easier if I said "true", but I'm going to have to go with "false".

The number 1 thing to check on is whether the range number is "ideal" miles or not. The Ideal Miles (which is what I have displayed on my VDS) is indeed meant to be a measure of capacity without any modifications based on driving style. But the "range" displayed on my dash (although it's programmable, so it can be changed to Ideal miles too) is different; that DOES take past driving in to account. Comparing the two can sometimes be instructive, as they show very different things. Make sure you are being told the Ideal Miles. It actually says "Ideal Miles" on my VDS; plus it's usually not hard to figure out as it's generally going to be the larger number.

Even if you have Ideal Miles, though, there is more to the story - the past can still affect the number of miles you see. If you drove gently before the charge, you may get more Ideal miles. If you charged slowly, you may get more Ideal miles. If you checked as soon as charging finished, you may find that Ideal Miles will rise a short while later as the car sits and the batteries level. If you check a while later, you may find it's gone down as the car has used power to keep the battery temperature at a certain spot. If you have owned the car for a long time and not given the batteries a chance to level (sitting at full Range Mode charge), the reported Ideal Miles may be less than the batteries are really capable of holding.

I would add that it is best to compare apples-to-apple in one more sense-- I have unproven anecdotal observations that the VOLTAGE at which one charges affects the end-point charge. Higher charging voltages SEEM to result in lower end-charge capacity, in my experience. Our house's voltage runs high, so we are charging at 245-247V most of the time; on a recent roadtrip, I charged at a station delivering only 195V and it charged my battery higher than ever before. So if one Roadster was charged in one environment, and the other in another, the end-charge values MAY not be comparable. Chad said this also in another way. I wanted to just emphasize and give more colour on the point.

That said, if these are both true Ideal Miles range mode charges under similar conditions, the 220 value is pretty good for a 1.5, the 200 value is a bit less ideal. (Note, though, this is a 10% difference, not 20%.)

Vger said:

That said, if these are both true Ideal Miles range mode charges under similar conditions, the 220 value is pretty good for a 1.5, the 200 value is a bit less ideal. (Note, though, this is a 10% difference, not 20%.)

Click to expand...

Good math!

ChadS said:

If you have owned the car for a long time and not given the batteries a chance to level (sitting at full Range Mode charge), the reported Ideal Miles may be less than the batteries are really capable of holding.

Click to expand...

ChadS, can you please explain this leveling a little more? I recently bought a used Roadster 2.0 sport. I have been charging it using the standard wall outlet charger at 15 amps, till I can get a NEMA 14-50 installed. Thus far I have only used the standard charge, and the car charges to 165-167 miles. I got the car with ~7000 miles on it, and it was reportedly kept plugged in by the prior owner. What is the best process to level the batteries and see if the car can give me more range in standard mode?

BryanW, below is an explanation that Tesla sent me when I asked about it (my local service manager forwarded it; I think it had been written by somebody else in the company).

By way of example, over 3 years my Roadster had "lost" capacity so that a Range charge only got about 210 miles. But that was largely because I rarely charged in Range Mode, and when I did I timed it so I took off as soon as it was full - the batteries never got a chance to level. So I charged it in Range Mode, then unplugged it and let it sit for a week. Then I took it for a drive and charged again - and this time I got 224 miles.

These are some observations of what happens when the Tesla Roadster battery pack becomes unbalanced and subsequently, is balanced using a semi-manual process.

This is a single case of a particular problem initially alerted on the VDS:

"Battery Service Required"
ID: 7
DATA: 0XFF000000
Dash Fault Status: Red

The problem was a mechanical defect regarding a battery pack rivet(s) which affected the proper operation of the battery monitoring system (false brick bleed warning.) The cells were fine. The solution was to replace the battery sheet, one of eleven in the pack. The process of replacing a sheet can result in a battery pack that is unbalanced, and in this case that is what happened.

The nominal delta-voltage between Vmin and Vmax bricks is 0.02 Volts at full standard charge. This pack had a 0.20 Volts delta (3.87 Vmin/4.07 Vmax,) due to the higher charge in the new sheet caused by a delay in installation. The imbalance yielded a low full charge level, measured by ideal miles (133 miles at standard full, about 63% soc.) The solution was to charge to full in range mode and leave plugged-in. If this didn't work, an alternate manual process was available.

The process took about three weeks and was finished when the roadster diagnostics indicated a fully balanced table of bricks and the ideal standard/range miles reached 187/237 miles (0.05 Volts delta at full range.) In later months, the balancing process would start again, especially if not having been in full standard or range plugged-in mode for a while (winter storage mode.) Periodically, the ideal miles would increase (up to 192/242 miles,) the Amp-hr level would increase (incrementally up to 158.89 Amp-hrs) and/or the delta-voltage would decrease (down to 0.03 Volts,) indicating a slightly better balance each time. This was a 2008 Roadster with 5000 miles and 3.15.17 firmware. Data was collected from the vds service diagnostics.

Here's a synopsis of the balancing:

cause/trigger: 0.03-0.06 Volts delta-voltage over-threshold between Vmin and Vmax bricks.
note: threshold may vary due to delta-Ahr levels and sheet/brick charge states.
remedy: fully charged and plugged-in in standard (slower balancing) or range mode.
method: balancing by slow discharge of high charge bricks.
resolution: balancing table changes to nominal once balanced.
monitoring: 0.01 Volts delta-v improvement per day when balancing.
Amp-hrs/SOC/ideal miles levels update when balancing improves.


Further details:

The balancing table was an array of the 99 bricks, with a 0 indicating a balanced brick and a 1 indicating an unbalanced one. The over-voltage bricks in the new sheet were shown as 1's. Gradually, as the balancing progressed, the 1's would change to 0's. During the charged/plugged-in/sitting mode, the high voltage bricks would discharge slightly more than the others.


Vmin would alternate between multiple bricks, indicating no one brick was persistently the lowest charged. The cycling between the changes would include 4-10 different Vmin bricks within an hour. After balancing, cycling between Vmax bricks included 6-15 different bricks per hour. The bricks within each cycling group would change periodically as a further indication of charge balance and health of all cells.

The amount of time spent at high charge did not appear to affect the battery capacity in any noticeable way, partially due to the fact that most of the pack was not at full charge. Vmin was between 4.00 - 4.11 V (the 10 old sheets, 91% of the pack) and Vmax was between 4.08 - 4.18 V (new sheet, 9% of the pack.) Active cooling was also a factor.

The pack temperature varied between 15 - 21 degrees C.

The range mode charging was done at 240 Volts/16 Amps.

The water pump remained on for all of range mode and early standard mode periods.

The balancing worked better if Vmin at full charge-sitting was > 4.00 Volts. In the beginning, it was necessary to top off the charge before the next daily cycle charge. Otherwise, Vmin would fall below 4.00 V during the sitting period. Range mode produces a larger voltage delta and balances more quickly than standard mode.

Vmax at full standard charge was 4.07 Volts and Vmax at full range charge was 4.18 Volts (4.16 Volts after sitting one hour.)

When charging to full in range, charging would halt during the Vmax = 4.18 V state. When discharging to the 25 ideal range miles remaining state, the vds soc alert would activate at 10% SOC, rather than voltage. This meant that when the pack was discharged, Vmin was in a slightly lower than normal state with respect to the low-charge imbalance (e.g. Vmin/Vmax = 3.66/3.76 Volts at 10% soc with a 0.10 Volt delta.)

The voltage delta would decrease after discharge and increase again after charging.

Continued range charging would timeout and revert to standard mode after 3.5 days and 3 range charges. The charge mode needed to be manually changed back to range mode at that time to continue the fastest balancing process.

It took some time for service to figure out how the balancing worked exactly, and the first half of the balance period was spent in various states of standard and range mode charging, charged and plugged-in, driving discharge and sitting discharge. The information about the best way to handle such an imbalance was ultimately read from company operations notes.

Click to expand...

BTW, I have heard several people say that charging on 110V increases the number of ideal miles they get. I have never charged on 110V; I have always used 240V as that's the outlet next to my car. I tried dialing the amps way down on my 240V connection (which typically is over 240V, like Vger's), figuring the responsible factor was low power over a long period of time, but didn't notice any benefit so I forgot about it. Vger's note above indicates that it may really be low voltage, rather than low power, that matters. I have noticed that I get different ideal miles after charging at different chargers, but never thought to try to see if that correlates with voltage. So I'm going to move some stuff around in the garage and give 110V a try on my Roadster.

Thank you ChadS and Doug_G,
I haven't done a range charge because I haven't yet needed the range. It's pretty great to be able to drive my daily needs and then just plug in on a standard charge each night. The joy of waking up with a full tank every morning!

I'll try your suggestions, charge in range and let it sit a bit, then keep it plugged in nightly on standard charge and report back in awhile with the results.

To bamashrink's question, one of the best things about a Roadster is that you can pull the log files and check them out prior to purchasing. This will tell you things like if the battery was driven down to empty, how often it was range charged, Average/Max/Min Ahr (capacity) of the pack, etc.

If you can get the owners to pull the logs for you I'm sure one of us would be happy to compare and contrast to help you pick the "best" one.

ChadS said:

BryanW, below is an explanation that Tesla sent me when I asked about it (my local service manager forwarded it; I think it had been written by somebody else in the company).

By way of example, over 3 years my Roadster had "lost" capacity so that a Range charge only got about 210 miles. But that was largely because I rarely charged in Range Mode, and when I did I timed it so I took off as soon as it was full - the batteries never got a chance to level. So I charged it in Range Mode, then unplugged it and let it sit for a week. Then I took it for a drive and charged again - and this time I got 224 miles.



BTW, I have heard several people say that charging on 110V increases the number of ideal miles they get. I have never charged on 110V; I have always used 240V as that's the outlet next to my car. I tried dialing the amps way down on my 240V connection (which typically is over 240V, like Vger's), figuring the responsible factor was low power over a long period of time, but didn't notice any benefit so I forgot about it. Vger's note above indicates that it may really be low voltage, rather than low power, that matters. I have noticed that I get different ideal miles after charging at different chargers, but never thought to try to see if that correlates with voltage. So I'm going to move some stuff around in the garage and give 110V a try on my Roadster.

Click to expand...

Hi ChadS, I'm the one who wrote that document in your post, "These are some observations of what happens when the Tesla Roadster battery pack becomes unbalanced...."

It looks like you have it there in its entirety and I hope it has provided some benefit. I don't work for Tesla. I made the text in a simple, systems management format as I thought it would be useful for future sorting and quick, at a glance knowledge retrieval. Lots of improvements could be made in this area of IT, I believe. The intention was to also make this document public, including this forum.

Thanks to Tesla service two years ago, I was correctly directed to keep my 2008 plugged-in at full range charge for a few weeks due to a rare situation with a bad battery imbalance. This worked beautifully and the pack recovered quite nicely and in fact, has improved even further since then.

What follows is my best guess as to what happened in your recent situation. I can't say for sure as I don't know all the details of your case and my information is only based on measurements I made regarding one issue with my vehicle. My service involved replacing a single sheet that inadvertantly caused a 0.20 delta-voltage imbalance at full standard charge. I carefully monitored and documented everything, then gave to Tesla service the detailed description of what happened during the entire balancing process. If you happen by the Seattle EVA meeting today, perhaps we can discuss it.

The average soc change per day when in full range balancing mode is 0.8% soc improvement/day (about 2 miles of range.) This is an average only and varies considerably (0.3% - 1.2%,) resulting in some days of little improvement and some days more. When the pack is largely out of balance, there is a somewhat linear delta-voltage change of 0.01 Volts/day improvement. As the pack is close to balanced, the delta-v change per day can be much less.

The balancing table will become active (some bricks shown as 1's) somewhere between 0.03 - 0.06 delta-v, depending on various factors. Coupled with an out of calibration soc/ideal miles reading, a battery that is only slightly out of balance may, at a glance, appear to be largely out of balance. For example, the balancing table may be active, the delta-voltage may be 0.06 Volts and the capacity estimate off by 6 miles. Together, this may yield an ideal range estimate that is 12-15 miles lower than normal, say 207 miles compared to a normal 220.

It sounds like this may have happened in your recent case:

Initial charged/plugged-in mode didn't have much effect. This was because that is the expected improvement in only one day (about a couple miles to 210.)

Driving the car interrupts the balancing process, but does allow the capacity estimate to recalibrate if followed by a full charge. The combination of discharge/charge and a few days of range charged/plugged-in mode should have balanced the battery pack (charged to 224 miles.) It's difficult to identify what the exact contribution was of each effect without more data.

If the vds battery diagnostic was still showing a large number of unbalanced bricks after a 3-7 day period of persistent plugged-in mode, then there may have been something else going on. Given your next day delta-voltage measurement of 0.03 V and the two separate readings of 210 range miles at full charge, it's unlikely the original delta-voltage was much higher than 0.05 Volts, That value only changes very gradually so it seems your battery may have been unbalanced, but not extremely so.

On rebalancing, the range estimate can be slightly high for a short time (separate type of exaggerated range from the common post-charge high voltage effect.) I don't think you actually lost any miles. Once the pack is mostly balanced, I would highly suggest to simply stay with standard/storage charging and only charge in range when you need. Your delta-voltage should be low so with that, regular driving and every once in a while, driving to 50 ideal miles range mode remaining should keep the pack balanced sufficiently. You really have to look at the history of the balancing and keep an eye on the vds diagnostics in order to have the complete picture - again, my guesses about your case are just that.

On 120v vs. 240V charging, I suppose it depends on taste. I prefer 240V as the temperature management is so much better. Gaining a bit of range at 120V may be handy sometimes, but continued charging at this lower voltage would push your average temperature up too much resulting in a bit more capacity loss over time.

This battery pack is amazing. When you witness the process of the battery balancing itself, there's a great sense of the achievement in software and systems design. To account for so many variables and elegantly maintain efficient balancing, it must have involved a staggering amount of research. Lots of characteristics of the pack change when balancing such as voltage, temperature, impedance, etc and each cell and brick varies in its own unique way. Watching the diagnostic numbers over a long period is like watching a fine ballet. It's dramatic as Vmin/Vmax dance around and various values change in big and small leaps and you don't exactly know what will happen next due to measurement uncertainties, but in the end you see this wonderful resolution of everything as the curtain falls on the sweet sight of the balancing table going all zero's.

Carl at Tesla service was always just great. I know he's been promoted and I do hope he continues to rise in the ranks and spread his common sense, keep it tight/do it right attitude.

Thanks, johns! For the original write-up, and this follow-up.

I did check the balancing table several times, but never saw anything but 0's. Probably not too surprising given the small delta-v of my pack.

I did try 110V charging, and sure enough got about 4-5 more miles the next day. But as you note perhaps 240V is better as matter of course.

Carl has been very helpful, and in fact it was him that forwarded your missive.

johns said:

...
This battery pack is amazing. When you witness the process of the battery balancing itself, there's a great sense of the achievement in software and systems design. To account for so many variables and elegantly maintain efficient balancing, it must have involved a staggering amount of research. Lots of characteristics of the pack change when balancing such as voltage, temperature, impedance, etc and each cell and brick varies in its own unique way. Watching the diagnostic numbers over a long period is like watching a fine ballet. It's dramatic as Vmin/Vmax dance around and various values change in big and small leaps and you don't exactly know what will happen next due to measurement uncertainties, but in the end you see this wonderful resolution of everything as the curtain falls on the sweet sight of the balancing table going all zero's.
...

Click to expand...

@johns thank you for your fascinating contribution of information! I have also been overwhelmed with the degree of sophistication, research and development that went into the system design and implementation of the Roadster battery pack. You only made one post but we would love to hear from you more!

I've always thought that the Achilles heel of the battery is the end of life race to the bottom scenario. Under this theory if one cell dies in a brick, then the other 68 have to take all the current, accelerating their failure and creating a failure runaway situation. The speed at which you ultimately lose the whole brick increases with time while the other bricks in the pack are still degrading at a very slow rate. Ultimately the whole pack is dead when you lose one brick. Obviously the system has a way of bleeding from one brick to another for balancing and to help ensure no bricks are stressed any more than any others.

What I'm wondering is to what extent, if any, the system is capable of bypassing a brick while either charging or discharging. If the answer is "none", then I think my failure runaway scenario will be the case in the final years of the battery's life. Obviously they can help with this a little by managing the A-hrs of each brick. For example, I think the balancing is not just with the min/max voltage. It also focuses on balancing the min/max A-hr. I don't know the relative weight they are giving each one of these parameters.

The biggest fault that I have with Tesla's R&D is due to California's rather nice environment compared to many other parts of the country. They weren't able to compile enough data from places that have extreme temperatures, road salt, and dirt roads.

You're welcome chads and hcsharp. I'm glad to contribute.

hcsharp, I believe cell failure is handled differently than you described and the system does well at gracefully aging the pack. If a cell dies, all 99 cells in that series chain are deactivated. It's not too much of a loss (1.4%) and it protects the rest of the pack from uneven wear. The C rating will only get slightly nudged lower due to the loss of a cell in each brick so normal aging will continue. It seems strange to remove good cells, but it's a reasonable solution if cell degradation is limited and managed well. My understanding of this is years old now and I wouldn't be surprised if they found clever ways to improve upon it.

There may be various kinds of balancing processes in use here and different passive/active discharge/charge methods, depending on the arrangement of the imbalance. Yes, as I noted in the case note synopsis, delta-Ahr and charge states are some of these factors in balancing. I think part of the challenge is how to actually measure everything accurately given all the uncertainties, and then somehow statistically put it together in a meaningful way so the software can always make the right decision, or at least correct itself within proper limits. It certainly would be fascinating to continue unraveling the onion and find out how every bit of it works.

Regarding outdoor environments, yeah I get your point, although I think Tesla did alright overall. I agree the doorsill issue is a problem anywhere and of course, gets worse with slush and salt. The stock headlamps and cabin climate controls are not perfect. These are all areas they did address in one way or another and I was impressed with Tesla's radiator fan shroud that made cooling more efficient. I'm sure they did their homework, it's just that they had to make some hard choices building such a low-slung sportscar tuned for efficiency and fun. And as far as extremes go, I have to admit, those old snow track speed and handling tests were fairly incredible so they do deserve praise, including the pretty good job they did on traction control.

johns said:

I believe cell failure is handled differently than you described and the system does well at gracefully aging the pack. If a cell dies, all 99 cells in that series chain are deactivated.

Click to expand...

Johns are you sure about being able to deactivate all 99 cells in series? I thought the lowest level that Tesla manages is at the brick level (69 cells in series) If a cell fails they brick has slightly less capacity but they work to keep bricks in balance and thus that would reduce the effective capacity of all bricks. In the few detailed drawings of the battery I have seen it appeared that all cells in the brick were electrically connected to the same common terminal. One ends up in close to the same place in overall capacity loss. Only if there are 2 or more cells in the same brick would one notice it.

dhrivnak said:

Johns are you sure about being able to deactivate all 99 cells in series?

Click to expand...

No, I'm not sure, that was just my understanding when I asked about it years ago. I remember it seemed odd that it should work that way as I originally thought only one cell would be affected. According to Tesla, the pack does have the ability to isolate on a cell basis.

I imagine our firmware upgrades have battery management improvements based on 10 million miles of gathered data.

Henry, we may not want to use the word "brick' in describing the battery parts. Fortunately they use the word sheets.

johns said:

No, I'm not sure, that was just my understanding when I asked about it years ago. I remember it seemed odd that it should work that way as I originally thought only one cell would be affected. According to Tesla, the pack does have the ability to isolate on a cell basis.

Click to expand...

johns I suspect what Tesla told you years ago is right. It would solve a lot of the engineering problems to be able to discharge with 69 separate series strings of 99 cells. First, if a cell died it would disable that whole string, wasting 98 good cells, but would ultimately result in longer battery life. Second, it enables them to easily do the switching necessary to also put the cells in each brick in parallel with each other to help manage balancing.

I have seen the diagrams that dhrivnak refers to that look like the cells in each brick are hard wired in parallel. But that doesn't mean they are. It only means they were capable of doing that either by hard wiring or with electronics. The more I think about it, the more likely it is that they chose a design with 69 separate strings. It's a lot easier controlling 10 amps than 700!

I have noticed that my lowest brick for voltage is #1 and has stayed the same for a long time. Several months in fact. As long as I've been watching. This concerned me at first. But the lowest brick measured by A-hr rotates almost randomly every few hours or days. I've never seen a 1 in the balancing grid, and never seen a min/max delta of more than .02v.

- - - Updated - - -

vfx said:

I imagine our firmware upgrades have battery management improvements based on 10 million miles of gathered data.

Henry, we may not want to use the word "brick' in describing the battery parts. Fortunately they use the word sheets.

Click to expand...

There are actually 9 bricks in a sheet. Those are Tesla's words but I hear you. What word should we replace it with? It's a grouping of 69 cells that are sometimes connected in parallel, and sometimes (we think) in series with the matching cell in each of the other 98 cell-groups. Blocks? Or maybe "pillows" to go with sheets. :tongue:

hcsharp said:

I have noticed that my lowest brick for voltage is #1 and has stayed the same for a long time. Several months in fact. As long as I've been watching. This concerned me at first. But the lowest brick measured by A-hr rotates almost randomly every few hours or days. I've never seen a 1 in the balancing grid, and never seen a min/max delta of more than .02v.

Click to expand...

I attempted to balance this weekend by charging in range mode at 110V. For the first time a saw a few 1's for balancing. It appeared to me that only if the delta is more than .03 does the balancing occur. I can see them ignoring a .02V difference as that can easily be in the range of measuring error. For what it is worth I maybe got back one mile in range. Probably not worth the stress on the pack.

My battery is severely out of balance right now, I guess I have to leave it plugged in for awhile to get it to rebalance - the drive down to 20-30% SOC and recharge in range mode, sit for a couple of hours trick didn't work
but I'll try letting it sit for a couple of days. I used to get 230/174 range/std on a charge but suddenly dropped down to 174/128 range/std - I'll let it sit for a week and see what happens.

ChadS said:

BryanW, below is an explanation that Tesla sent me when I asked about it (my local service manager forwarded it; I think it had been written by somebody else in the company).

By way of example, over 3 years my Roadster had "lost" capacity so that a Range charge only got about 210 miles. But that was largely because I rarely charged in Range Mode, and when I did I timed it so I took off as soon as it was full - the batteries never got a chance to level. So I charged it in Range Mode, then unplugged it and let it sit for a week. Then I took it for a drive and charged again - and this time I got 224 miles.



BTW, I have heard several people say that charging on 110V increases the number of ideal miles they get. I have never charged on 110V; I have always used 240V as that's the outlet next to my car. I tried dialing the amps way down on my 240V connection (which typically is over 240V, like Vger's), figuring the responsible factor was low power over a long period of time, but didn't notice any benefit so I forgot about it. Vger's note above indicates that it may really be low voltage, rather than low power, that matters. I have noticed that I get different ideal miles after charging at different chargers, but never thought to try to see if that correlates with voltage. So I'm going to move some stuff around in the garage and give 110V a try on my Roadster.

Click to expand...