Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Loss of Range thoughts

This site may earn commission on affiliate links.
Hey there..

Many don't want to partake in the figuring out of the range loss conversations and trying to see if there is anything we can do about it. Which is fine, we all have that choice. For those that are interesting in playing around with things like myself, I wanted to throw these thoughts out there to see what you think.

What if a lot of this range decline was because we were consistently on the bad end of rounding error?
What if the end of a charge where it says "Charging Complete" was really important in their algorithm?

If the end point of the charges in the Model 3 are important, that could explain why I was able to manipulate my "Estimated Range at 100%" number.

If we've all been charging and the point where it stops on its own which may generally be on the wrong end of the rounding error in general, I think we'd see an overall decline in these numbers (like we are seeing, for most).

Hypothetical question: What if we were stop our charging on the other side of the rounding error so that the ending "Estimated Range at 100%" is always moving upward a bit every time we charged?

Some background on why I feel this way.. I shouldn't be able to manipulate these numbers simply by just starting a charge and stopping it 1-2 minutes later. I was able to gain range and lose range just by doing this so I'm going to play around with it some more, specifically stopping it on the good side of the rounding error to see if I can manipulate this to my advantage.

Of course, I won't be able to erase any actual/real battery degradation but if this really is an rounding error, I should be able to move it upward like I've seen.

Another thought, when I monitored the car when the batter was low, the rounding error was HUGE. When I was down to 3% SOC left one time, I took some screenshots of Stats for iOS. I was at 10 estimated miles left with 3%. Stats said my max rated range was 334 miles. I was shocked - which is when another user pointed out the rounding error. 10 / 0.03 = 334.

What if Tesla was taking that rounding error into account when the charging stops and this is what a large part of the problem is? Interesting things to think about.

After reading this, what are your thoughts?
 
Some additional thoughts... Since the rounding error grows as the SOC goes lower, this may be why those who have been charging to lower % numbers have experienced a rise in their range. For example, I've been at 288-290 for a long while but just this morning but stopping my charge at just the right spot (on the right side of the rounding error), I was able to realize an uptick in my max estimated range.

Now that I know this, I want to try it again when the SOC is down in the 20s like it was last night. Looking at the rounding errors from my charge last night, in theory, I could have stopped at this point (arrow below points out the right spot) and driven up my estimated range at 100% up from 284 to 295 in the screenshot below. You can bet that I'm going to be trying this next!

Then when I have a good jump like that, I'm going to let that sit so the car takes the readings over night and see if that practice helps drive the range up over time. I think the key is that the car needs to sit so that the algorithms see those numbers overnight and that's what Tesla records. Will keep this thread posted as I'm able to drive the car and try this theory out.

My theory: Stopping charging at lower SOCs on the right side of the rounding error will have a more dramatic effect on the "Estimated Range at 100%" number. The lower the SOC, the larger the rounding error and it presents more (in theory) opportunity to raise the "Estimated Range at 100%". If this theory is right, it is nearly impossible to increase range by ending charges at higher SOCs. The rounding error flattens out as the SOC goes up so the real opportunity is when SOC is low. Man, I wish I could have a lower SOC right now to try this. Working on it, will report back when I get more data.

upload_2019-11-27_8-19-17.png
 
Last edited:
Hey there..

Many don't want to partake in the figuring out of the range loss conversations and trying to see if there is anything we can do about it. Which is fine, we all have that choice. For those that are interesting in playing around with things like myself, I wanted to throw these thoughts out there to see what you think.

What if a lot of this range decline was because we were consistently on the bad end of rounding error?
What if the end of a charge where it says "Charging Complete" was really important in their algorithm?

If the end point of the charges in the Model 3 are important, that could explain why I was able to manipulate my "Estimated Range at 100%" number.

If we've all been charging and the point where it stops on its own which may generally be on the wrong end of the rounding error in general, I think we'd see an overall decline in these numbers (like we are seeing, for most).

Hypothetical question: What if we were stop our charging on the other side of the rounding error so that the ending "Estimated Range at 100%" is always moving upward a bit every time we charged?

Some background on why I feel this way.. I shouldn't be able to manipulate these numbers simply by just starting a charge and stopping it 1-2 minutes later. I was able to gain range and lose range just by doing this so I'm going to play around with it some more, specifically stopping it on the good side of the rounding error to see if I can manipulate this to my advantage.

Of course, I won't be able to erase any actual/real battery degradation but if this really is an rounding error, I should be able to move it upward like I've seen.

Another thought, when I monitored the car when the batter was low, the rounding error was HUGE. When I was down to 3% SOC left one time, I took some screenshots of Stats for iOS. I was at 10 estimated miles left with 3%. Stats said my max rated range was 334 miles. I was shocked - which is when another user pointed out the rounding error. 10 / 0.03 = 334.

What if Tesla was taking that rounding error into account when the charging stops and this is what a large part of the problem is? Interesting things to think about.

After reading this, what are your thoughts?
There have been some comments that Tesla should display an additional unit, ie 90.1% not 90%, because rounding is affecting the range numbers. I think I've seen my rated range on the Stats app change 7 miles, just because of rounding, as I drive on the highway.

Now, you're going to make me look for some screenshots:
Screenshot 2019-11-27 14.05.22.jpg

So, while driving, Stats is showing me that at 45% SOC, while the odometer goes from 141 miles to 138 miles, the Rated Range calculation goes from 314 miles to 307 miles. That's all due to the lack of precision in the SOC number. If they only added one more unit, the Rated Range calculation wouldn't vary, in this case 7 miles. So, some cases of "deg" are just rounding errors. Some is just BMS drift errors. And some is true deg, but it's hard to know which is which.

For those who don't use Stats, the Odometer and SOC% numbers will match those in the car.
 
Last edited:
I went back and looked at some old charges where I lost range in a single charge. It appeared that I lost range because the charging stopped on the wrong side of the rounding error. Here is a chart I made that shows the rounding error in a single charge. If I would have stopped the charging at the high points, that would (in theory) driven my "Estimated Range at 100%" upward. If I waited a few more minutes and it happened to stop on the low side of the rounding error, I would have lost a big chunk of miles.

I'm really starting to think that where your charge ends and on which side of the rounding error has a big effect on this. I think this may be what the problem is for a lot of us. If this is true, we should be able to manipulate it back up but it has to be done in steps. Can't keep moving it up one after another, the car needs to sit and that charge needs to be the last charge so it's "realized" by the car's algorithms.

upload_2019-11-27_11-14-42.png


Here is the data that shows where you stop charging, is what sets the range and I think proves my theory on paper. The charging stopped on the line in the blue square. See where the rounding error is? It's 291.776 miles. If I would have manually stopped the charging a few minutes earlier on the one in the pink, I could have ended with 293.387 miles.
upload_2019-11-27_11-2-52.png


Here is where you can see that the ending charge (and the rounding error) is taken into account when your charge ends:
upload_2019-11-27_11-6-57.png


If the charging would have ended a few minutes earlier, it would have actually increased the range to 293.387.
 

Attachments

  • upload_2019-11-27_10-58-11.png
    upload_2019-11-27_10-58-11.png
    154.9 KB · Views: 35
Last edited:
There have been some comments that Tesla should display an additional unit, ie 90.1% not 90%, because rounding is affecting the range numbers. I think I've seen my rated range on the Stats app change 7 miles, just because of rounding, as I drive on the highway.

Now, you're going to make me look for some screenshots:

Oh when I was down in the sub 10% range, the rounding error is HUGE. This is what I'm saying, I think they're using that rounding error in their data and that is a problem. This would explain the such randomness that people see, up and down. It's all over the place.

If what I think is true, we should be able to control it by strategically stopping our charging. The lower the SOC, the wider swings we get and more opportunity to stop on a huge jump up. If that's the case, that's a massive error in their calculation. If I'm then able to get this way up, I'm then going to do a charge way up to see if it sticks. I'm not done with this yet!! lol
 
Oh when I was down in the sub 10% range, the rounding error is HUGE. This is what I'm saying, I think they're using that rounding error in their data and that is a problem. This would explain the such randomness that people see, up and down. It's all over the place.

If what I think is true, we should be able to control it by strategically stopping our charging. The lower the SOC, the wider swings we get and more opportunity to stop on a huge jump up. If that's the case, that's a massive error in their calculation. If I'm then able to get this way up, I'm then going to do a charge way up to see if it sticks. I'm not done with this yet!! lol

Question is how do you know when is that exact point in time to stop charging it at that peak?
 
If what you say is correct, the problem is that the SOC isn't precise to the tenths. If the charge stops at 74.6%, the car displays 75%, and if it stops at 75.4%, it also displays 75%. So, you're saying that if the car stops at 74.6%, but thinks it's 75%, then the next day does it again and again and again, the Rated Range estimate becomes lower and lower and lower, which is BMS drift.
 
If it's modestly cool and you still have some regen most loss is the heater.
If pretty cold and no regen, it's a combination of lost regen, depending on where and how you drive plus even more heat.
If it's extremely cold the car may spend some energy heating the battery.

Not really much mystery here.

The battery itself does not lose energy because of (normal) lower temps. It just slows down how much it can push in or pull out as it gets colder.

S/X does behave a little different as it activates the battery heater much earlier than a Model 3 might use motors to heat battery.

I'm not sure if the Model 3 can "raise" the rate of battery heating with the motors while you are driving.
Well, I guess it can because that's what it must be doing with new precondition when navigating to a Supercharger.
Oddly enough the Model S/X does not use it's battery heater (more) when routed to Super Charger, but it has that heater on way more often than the Model 3 does without routing to a Super Charger.
 
Well, I plan on using Stats for iOS since it gives live updates from the car during charges. Since the rounding errors are greater at lower SOCs, I'm going to try to start charing at 20%, then literally watch Stats and after a few minutes when I hit a high point, stop the charging for the night and let it sit so the number can be realized by the car for the night, then let the car charge up normally before we go to work.

You can see what this number is live with Stats.. Here are two screenshots from it minutes apart:
upload_2019-11-27_11-39-29.png


(For the below, take note of the values in the red boxes in the screenshots above which show the Estimated Range at 100% value)
First screenshot (left side): 3% and 10 miles left. Math is 10 / 0.03 = 333.3333 Estimated Max Miles at 100%
Drove a few more miles..
Second screenshot (right side): 3% and 7 miles left. 7 / 0.03 = 233.3333 Estimated Max Miles at 100%

So, my theory is, while it is charging, it sees these same rounding errors and they're much more pronounced at lower SOCs as I showed above. So, when it's charging, stopping at these higher points, should help us increase this range. I think that once we stop on a high point, we need to let the car sit for a long while so it takes that reading into account and takes it as truth. Do this over and over again and I think we'll be set. I'm going to work on the method next now that I'm honing in on the cause and how we can manipulate it to our advantage. This will obviously not fix actual, real battery degradation - just the error in the estimated range at 100%.
 
Oh when I was down in the sub 10% range, the rounding error is HUGE. This is what I'm saying, I think they're using that rounding error in their data and that is a problem. This would explain the such randomness that people see, up and down. It's all over the place.

If what I think is true, we should be able to control it by strategically stopping our charging. The lower the SOC, the wider swings we get and more opportunity to stop on a huge jump up. If that's the case, that's a massive error in their calculation. If I'm then able to get this way up, I'm then going to do a charge way up to see if it sticks. I'm not done with this yet!! lol

The rounding error is only due to the limited display in your particular app. It has nothing to do with, or any affect on how the BMS calculates capacity. Tesla actually gets a very precise calculation internally, and the BMS will use that data, so the rounding errors you are observing have no affect on what the BMS is calculating.

That said, the BMS does have its limitations, and there is always some uncertainty in determining what the exact capacity of the battery is at any particular time.
 
The rounding error is only due to the limited display in your particular app. It has nothing to do with, or any affect on how the BMS calculates capacity. Tesla actually gets a very precise calculation internally, and the BMS will use that data, so the rounding errors you are observing have no affect on what the BMS is calculating.

That said, the BMS does have its limitations, and there is always some uncertainty in determining what the exact capacity of the battery is at any particular time.

I understand what you mean but I'm referring to the "Estimated Miles at 100%" which is used in TeslaFi's "Battery Degradation Report" which seems to be also reflected in the car.

If what you say is true, I should be unable to change the miles shown in the car when you have it set to show miles, versus %. Based on what I am seeing, I am able to manipulate this number - if I'm supposed to be able to or not, it is appearing that I am able to do so. I am working on confirming this now.
 
The rounding error is only due to the limited display in your particular app. It has nothing to do with, or any affect on how the BMS calculates capacity. Tesla actually gets a very precise calculation internally, and the BMS will use that data, so the rounding errors you are observing have no affect on what the BMS is calculating.

That said, the BMS does have its limitations, and there is always some uncertainty in determining what the exact capacity of the battery is at any particular time.
Why can't the 3rd-party APIs pull the data with the same precision?
 
Also for those following, when you use the car and the battery level and miles drop, you can see this number jump around in Stats for iOS. The lower the SOC gets, the bigger it jumps around. This is what I'm trying to target the next time I charge. I'm going to run down to 20% which will provide a good jump. I'll let it sit overnight in hopes that it will stick when I charge to 90% the next time.
 
Why can't the 3rd-party APIs pull the data with the same precision?
The API will provide more precision in rated miles; my Remote Tesla app gives it to 2 decimal places, but the API only provides whole numbers for SOC.
If you can use TM-Spy or Scan My Tesla from the diagnostics port, you can get very accurate calculations for SOC, but for some reason, it will still only provide whole numbers for rated miles.
That is why I use both the Remote Tesla app, and TM-Spy.
 
  • Like
Reactions: KenC
Honest question (not trolling, I swear).

OP suppose everything you say is accurate. Suppose its a rounding error, because "energy" filled in a battery being displayed is an algorithm. How is that going to change what you do when you charge your car, if it changes your behavior at all?

Are you going to somehow try to chase a method of charging that produces the display you want, or are you going to just charge and forget it, because its just a display of a number that doesnt mean a lot?
 
Honest question (not trolling, I swear).

OP suppose everything you say is accurate. Suppose its a rounding error, because "energy" filled in a battery being displayed is an algorithm. How is that going to change what you do when you charge your car, if it changes your behavior at all?

Are you going to somehow try to chase a method of charging that produces the display you want, or are you going to just charge and forget it, because its just a display of a number that doesnt mean a lot?

Thanks for posting - and a great question.

I talked with a SC the other day for a long while about this and he told me that when the car sits for some amount of time (the amount of time wasn't disclosed) that the battery pack will stabilize and the car will take readings at that time which essentially "makes it so".

My thought process is to charge it up sort of slow so I can catch it at the right moment and stop the charge when at a high point by watching Stats for iOS. I think I've already shown that this is what directly feeds the "Battery Degradation Report" in TeslaFi.com. I plan on letting the car sit on these states for a while and then charging to my 90% in the morning. This is all new so I'm not sure how I will do it exactly. But the theory is that if I can get these high points to stick, we should be able to do this over and over increasing the values over time.

This could all burn up and I could prove nothing and that I'm just chasing a ghost but I won't know unless I try. I want my battery to get down around 20% so I can then charge it up a bit and stop it when it's on a high point. Then let it sit overnight and see what the miles says in the car when I switch it to range. I really think that the point where you stop a charge has something to do with the range number and this rounding error things what makes these things bounce all over the place. I've seem the rounding errors in the data I see on TeslaFi.com. The miles I get on the chart, seems to be the miles I see in the car so it looks to me like everything has the rounding number in it but who knows.. Like I said, this could turn out to be nothing. I'd like to find out and if it seems like I was successful, you can bet that I'm going to share the process to see if others can duplicate it.

When I charged to 90% last just two nights ago, 90% SOC got me 261.55 miles and an estimated range of 290.61miles at 100%.

When my car gets low enough I'm going stop it at a high point to take advantage of the rounding error and we will see what happens. I'll let it stay overnight and see if that helps. All discovery here so hard to say what will happen.

So over time, if this fixes things, I may use it to reset the range after a while. That also may help me understand that it really doesn't matter either. lol. I wonder if that has any effect on when the limitations of brake regent comes into place. Another thing to pay attention too.

Sorry for all the text, hope that answers your question.
 
It does, yeah. Tesla attracts a lot of engineering types so lots of owners like to dig into the numbers. For a lot of people the number being 7-8-15 miles lower seems to generate a ton of stress.

I personally would not be doing a bunch of extra charging cycles just to try to get the rated range number to show higher. We know for a fact that batteries degrade with extra charging cycles. We know that "shallow charges are better that deep discharges".

for me, I would not be chasing a rated range number by increasing the number of charge sessions, but like I said thats just me. Hopefully you find the information you dig up to be stress relieving.
 
  • Love
Reactions: Rocky_H
It does, yeah. Tesla attracts a lot of engineering types so lots of owners like to dig into the numbers. For a lot of people the number being 7-8-15 miles lower seems to generate a ton of stress.

I personally would not be doing a bunch of extra charging cycles just to try to get the rated range number to show higher. We know for a fact that batteries degrade with extra charging cycles. We know that "shallow charges are better that deep discharges".

for me, I would not be chasing a rated range number by increasing the number of charge sessions, but like I said thats just me. Hopefully you find the information you dig up to be stress relieving.

It's all good, I'm doing most of the observation from the usual cycles I do on a daily basis. Our car gets driven around 115-150 miles a day so it's easier for me to see this stuff over a shorter amount of time. I've maybe done two smaller cycles for this which isn't much.

My wife does the driving most of the time so I get a lot of time to look at the numbers.
 
Last edited:
  • Informative
Reactions: jjrandorin