This is why you can't get 'rated range'

[David99]

So how did you get back up and running again? Any clue on why it didn't let you use more of the 4kWh?

I had a generator in the back. It was a planned event. I doubt it will ever let you use the entire 4 kWh buffer. But when the car was new, once drove 9.5 miles at aprox 40 mph after it showed 0 which is a lot more than what I got at this test. At this test it was just 2.6 miles and driving 20 mph. Again, I believe it comes down to how 'healthy' the BMS thinks the battery is. When I drive a lot (on road trips) I have been able to go beyond 0 many times and always more than 2.7 miles. For this test I let the battery go down to 2% over a period of 2 days before starting this test. That's my experience in general, when the battery is heavily used it seem to deal with low SOC much better than when it's been sitting there for a while. It's not just temperature. The battery was pretty warm for this test.

 

[David99]

Correction: According to the CAN data I was able to get only 0.4 kWh from the buffer. Usable remaining showed -0.4 when it shut down.

 

[supratachophobia]

Do you think the shut down was because of one of the cell packs hitting 3.0v or lower and were you watching it on TM-SPY or the dash?

This is the big question. I will look at the CANbus on my own if this happens. I bet you are right.

 

[supratachophobia]

Out of curiosity, when you were at 0% displayed, what was TM-Spy or Scan My Tesla showing you for "SOC UI" and "Nominal energy remaining"? Thanks again for testing this stuff.



While I agree with you the car is supposed to shut down with voltage left, it's not what we're necessarily talking about. I've already explained it once with sources but I have yet to see an explanation that says otherwise. I'll explain it again. In general, lithium-ions can not go below 2.5V, else they become unusable because you can't charge at those levels without compromising safety. This is where the term "brick" comes from, your battery turns into a brick. Lithium-ion batteries typically have a protection circuit from 2.6-2.9 volt where the battery will stop "working" (can't be used), which is this "anti-bricking buffer" so you don't get near the 2.5V level mentioned earlier. The higher your anti-bricking buffer is, the more "pad" you have to preserve the battery if it isn't charged for an extended period of time because you will slowly lose energy over time.

As it's already been proven, Tesla's appear to shut down the vehicle around ~3.0V. The range from 2.5-3.0V is what I would consider the anti-bricking buffer; you can't use it... because it's an anti-bricking buffer...

What we're talking about is another 4 kWh buffer that the BMS is obviously tracking because 1) it reports "nominal energy remaining" which has consistently shown is around 4 kWh at 0% displayed SOC, 2) it also reports "SOC UI" which corresponds to the amount of "nominal energy remaining" and 3) we have several examples of vehicles actually shutting down (or extremely close to) at 0.0 kWh remaining or 0% "SOC UI", and dozens of examples of vehicles driving 10-17 miles past 0% displayed.

We also have examples of vehicles shutting down a mile or two after 0% displayed, and this is what I'm curious about and David99 is providing some insight on with actual BMS data.

In no way am I suggesting to drive below 0%. I think it actually makes perfect sense you would put a pad on the bottom end, on top of the anti-bricking buffer, because the car becomes so power limited now safety becomes questionable. For example, driving 20 mph on a 65 mph freeway... probably not smart. Not being able to go up a hill, probably not good. Strong head winds or very cold weather with high energy draw... not good as well. So this buffer makes sense such that folks get a fairly consistent driving experience down to 0% displayed (still going to be slightly power limited as you get to those levels).

I would agree with you and even go farther to suggest that in Tesla testing things, 4kwh left in the battery is the point where the lowest voltage module has the highest chance at being at or near 3.0v. Which, if you think about it, is pretty smart. It's lying, but it's smart.

Edit: Or 2.9v since that is what was shown int @David99's data.

 

[shred86]

I had a generator in the back. It was a planned event. I doubt it will ever let you use the entire 4 kWh buffer. But when the car was new, once drove 9.5 miles at aprox 40 mph after it showed 0 which is a lot more than what I got at this test. At this test it was just 2.6 miles and driving 20 mph. Again, I believe it comes down to how 'healthy' the BMS thinks the battery is. When I drive a lot (on road trips) I have been able to go beyond 0 many times and always more than 2.7 miles. For this test I let the battery go down to 2% over a period of 2 days before starting this test. That's my experience in general, when the battery is heavily used it seem to deal with low SOC much better than when it's been sitting there for a while. It's not just temperature. The battery was pretty warm for this test.

Some folks have used the entire 4 kWh buffer while driving, and this guy did it while stationary. I think you're right about how recent or frequently the battery has been used (especially if it's recently been close to 0%) and battery "health" will play a big factor in how much of that 4 kWh you can get into.

I would agree with you and even go farther to suggest that in Tesla testing things, 4kwh left in the battery is the point where the lowest voltage module has the highest chance at being at or near 3.0v. Which, if you think about it, is pretty smart. It's lying, but it's smart.

Edit: Or 2.9v since that is what was shown int @David99's data.

Yeah, I've noticed the cells are usually very balanced (+/- 5-7 mV) when you're not moving. When driving around, the voltage difference seems to fluctuate quite a bit. Seems directly related to how much energy draw there is (higher load = bigger voltage difference), which makes sense.

My theory is you can probably almost always "use" all of the 4 kWh buffer (assuming no imbalance issues), but it just depends on how well your BMS is in-tune, for a lack of a better term, with your battery cells or how "healthy" your BMS thinks the battery cells are (like David99 mentioned) which I'd imagine is how well your cells are balanced. For example, if the battery has been used recently and has been closer to the 0% levels or cells are very well balanced, the power limit wouldn't be as aggressive initially but would still significantly reduce as you get closer to 0.0 kWh. In the opposite scenario (haven't used the battery in a while, haven't been close to 0% in a while and/or cells not as balanced), you could still utilize all 4 kWh but you would have to do it at a very slow rate, like the guy in that YouTube video or crawling at a slow speed. If you're cruising at say, 20+ mph, I wonder if BMS just decides to shut down prior to 0.0 kWh to prevent going below a desired voltage level, especially considering when there's load on the batteries, the voltage is going to be lower, so higher load = lower voltage and a bigger difference in voltage across the cells (factoring in cell balance). In the extreme cases when the cells are very unbalanced (beyond what's considered "normal" for BMS), you'll see shut down before even 0% displayed, which we have seen often results in Tesla identifying a battery issue and replacing the pack or folks having to "re-sync" their BMS by doing low SOC to 100% SOC charges a few times.

Interesting stuff for sure. Makes you appreciate all the complexities involved with designing these cars.

 

[N..8]

Back full circle is that the 4kw is part of the pack Rate Range. Since it appears yes your can use it, Tesla says it's part of EPA Rated range. In the end it's really 12 miles that you shouldn't and everyone needs to understand that if you want it, it comes with a risk. So they scale the SOC from full battery down to having it with 4kw left, without the owner even knowing it. Pretty sneaky.

 

[supratachophobia]

Some folks have used the entire 4 kWh buffer while driving, and this guy did it while stationary. I think you're right about how recent or frequently the battery has been used (especially if it's recently been close to 0%) and battery "health" will play a big factor in how much of that 4 kWh you can get into.



Yeah, I've noticed the cells are usually very balanced (+/- 5-7 mV) when you're not moving. When driving around, the voltage difference seems to fluctuate quite a bit. Seems directly related to how much energy draw there is (higher load = bigger voltage difference), which makes sense.

My theory is you can probably almost always "use" all of the 4 kWh buffer (assuming no imbalance issues), but it just depends on how well your BMS is in-tune, for a lack of a better term, with your battery cells or how "healthy" your BMS thinks the battery cells are (like David99 mentioned) which I'd imagine is how well your cells are balanced. For example, if the battery has been used recently and has been closer to the 0% levels or cells are very well balanced, the power limit wouldn't be as aggressive initially but would still significantly reduce as you get closer to 0.0 kWh. In the opposite scenario (haven't used the battery in a while, haven't been close to 0% in a while and/or cells not as balanced), you could still utilize all 4 kWh but you would have to do it at a very slow rate, like the guy in that YouTube video or crawling at a slow speed. If you're cruising at say, 20+ mph, I wonder if BMS just decides to shut down prior to 0.0 kWh to prevent going below a desired voltage level, especially considering when there's load on the batteries, the voltage is going to be lower, so higher load = lower voltage and a bigger difference in voltage across the cells (factoring in cell balance). In the extreme cases when the cells are very unbalanced (beyond what's considered "normal" for BMS), you'll see shut down before even 0% displayed, which we have seen often results in Tesla identifying a battery issue and replacing the pack or folks having to "re-sync" their BMS by doing low SOC to 100% SOC charges a few times.

Interesting stuff for sure. Makes you appreciate all the complexities involved with designing these cars.

I've always wondered if they got smart enough to include Navigation Destination into the equation. Like, if you are driving to a supercharger 5 miles away and it's a route destination on the screen, but you are in the 4kwh anti-stupid buffer, it does all it can to make sure you get those 5 miles.... It would be like your car "cared" about you.

 

[David99]

Yeah, I've noticed the cells are usually very balanced (+/- 5-7 mV) when you're not moving. When driving around, the voltage difference seems to fluctuate quite a bit. Seems directly related to how much energy draw there is (higher load = bigger voltage difference), which makes sense.

That's actually an artifact. While the cells might drift apart a little more under load, it is not as much as it appears looking at the CAN bus data. The issue is that the measurements are taken in sequence, not at the same time. Both TM-Spy and Scan my Tesla show the sequence of measurements coming in. It takes about a good second for all voltages to come in. In normal driving, it is rare that you have the exact same power draw over 1.5 seconds hence the greater voltage differences. It's not actually the cells drifting apart that much.

The other issue is noise on the data. For example when you supercharge, the power should be pretty constant. When you look at the cell voltages, you see a zigzag pattern. It's like a wave going along the cells. Interestingly, when you drive and keep the power draw as constant as possible you see the exact same wave pattern, except reversed. The nature of the wave is very obvious and it is extremely unlikely that under load the cell groups suddenly drift apart in voltage to create a wave pattern going through groups 1 through 96. I don't know where the wave pattern is coming from, but it's obviously an artifact as well. It is not the true voltage difference between cells under load.

IOW, reading the CAN bus data you can't measure cell balancing under load (charging or discharging). The data stream has too many artifacts. Static measurement is fine, though.

Here is a screen shot showing the wave pattern.

 

[shred86]

That's actually an artifact. While the cells might drift apart a little more under load, it is not as much as it appears looking at the CAN bus data. The issue is that the measurements are taken in sequence, not at the same time. Both TM-Spy and Scan my Tesla show the sequence of measurements coming in. It takes about a good second for all voltages to come in. In normal driving, it is rare that you have the exact same power draw over 1.5 seconds hence the greater voltage differences. It's not actually the cells drifting apart that much.

The other issue is noise on the data. For example when you supercharge, the power should be pretty constant. When you look at the cell voltages, you see a zigzag pattern. It's like a wave going along the cells. Interestingly, when you drive and keep the power draw as constant as possible you see the exact same wave pattern, except reversed. The nature of the wave is very obvious and it is extremely unlikely that under load the cell groups suddenly drift apart in voltage to create a wave pattern going through groups 1 through 96. I don't know where the wave pattern is coming from, but it's obviously an artifact as well. It is not the true voltage difference between cells under load.

IOW, reading the CAN bus data you can't measure cell balancing under load (charging or discharging). The data stream has too many artifacts. Static measurement is fine, though.

Here is a screen shot showing the wave pattern.

View attachment 379402

I had a feeling that was the case just based on watching the voltage of the various cells during acceleration and deceleration. I know TM-Spy samples pretty fast when recording (~100fps is what I’m seeing), but I wonder if it’s just a limitation of how quickly cell voltage is being reported? Anyways, good point as that better explains the big jumps in groups of cells during accel/decel.

 

[ran349]

That's actually an artifact. While the cells might drift apart a little more under load, it is not as much as it appears looking at the CAN bus data. The issue is that the measurements are taken in sequence, not at the same time. Both TM-Spy and Scan my Tesla show the sequence of measurements coming in. It takes about a good second for all voltages to come in. In normal driving, it is rare that you have the exact same power draw over 1.5 seconds hence the greater voltage differences. It's not actually the cells drifting apart that much.

The other issue is noise on the data. For example when you supercharge, the power should be pretty constant. When you look at the cell voltages, you see a zigzag pattern. It's like a wave going along the cells. Interestingly, when you drive and keep the power draw as constant as possible you see the exact same wave pattern, except reversed. The nature of the wave is very obvious and it is extremely unlikely that under load the cell groups suddenly drift apart in voltage to create a wave pattern going through groups 1 through 96. I don't know where the wave pattern is coming from, but it's obviously an artifact as well. It is not the true voltage difference between cells under load.

IOW, reading the CAN bus data you can't measure cell balancing under load (charging or discharging). The data stream has too many artifacts. Static measurement is fine, though.

Here is a screen shot showing the wave pattern.

View attachment 379402

Is that a screen shot from Scan My Tesla? I have a question about the charge and discharge numbers on the top of the graph. Let's say for example, you charge your car and after your charging is finished, the car dash says it added 50 kWh to the battery. Does that charge total go up by exactly 50 kWh, or something slightly different?

 

[N..8]

Mine goes up each time it's charging, it does run up and down.

 

[ran349]

Mine goes up each time it's charging, it does run up and down.

Those look like lifetime numbers, so I would think they would only increment. Are you talking about something different?

 

[David99]

Is that a screen shot from Scan My Tesla? I have a question about the charge and discharge numbers on the top of the graph. Let's say for example, you charge your car and after your charging is finished, the car dash says it added 50 kWh to the battery. Does that charge total go up by exactly 50 kWh, or something slightly different?

Those numbers are lifetime totals but unfortunately not correct. It's a longer story. Some older cars just don't report it correctly.

 

[ran349]

Those numbers are lifetime totals but unfortunately not correct. It's a longer story. Some older cars just don't report it correctly.

Ok, thanks, but what about this question:
Let's say for example, you charge your car and after your charging is finished, the car dash says it added 50 kWh to the battery. Does that charge total go up by exactly 50 kWh, or something slightly different? My guess is the BMS would report a number slightly more.

 

[shred86]

@David99 - Out of curiosity, does the "SOC" percentage in Scan My Tesla match up with what you're seeing displayed in the vehicle, or is it slightly off sometimes? If it's exact, that further supports a lot of what has been discussed in this thread. It looks like Scan My Tesla is calculating "SOC" by taking (nominalRemaining - buffer) / (nominalFullPack - buffer) * 100.0, so it's not a value reported directly from the CAN bus. Same with "Usable full pack" and "Usable remaining", both are simply the nominal values minus the buffer. Initially, I thought they were. However, nominalRemaining, nominalFullPack and buffer are reported via CAN bus (0x382).

 

[ran349]

@David99 - Out of curiosity, does the "SOC" percentage in Scan My Tesla match up with what you're seeing displayed in the vehicle, or is it slightly off sometimes? If it's exact, that further supports a lot of what has been discussed in this thread. It looks like Scan My Tesla is calculating "SOC" by taking (nominalRemaining - buffer) / (nominalFullPack - buffer) * 100.0, so it's not a value reported directly from the CAN bus. Same with "Usable full pack" and "Usable remaining", both are simply the nominal values minus the buffer. Initially, I thought they were. However, nominalRemaining, nominalFullPack and buffer are reported via CAN bus (0x382).

From the data I have seen, including a lot from David99, the dash SOC % correlates precisely to the usable percent remaining, which is defined as you noted in your post. This is the case from 100% down to 0%.

 

[David99]

@David99 - Out of curiosity, does the "SOC" percentage in Scan My Tesla match up with what you're seeing displayed in the vehicle, or is it slightly off sometimes? If it's exact, that further supports a lot of what has been discussed in this thread. It looks like Scan My Tesla is calculating "SOC" by taking (nominalRemaining - buffer) / (nominalFullPack - buffer) * 100.0, so it's not a value reported directly from the CAN bus. Same with "Usable full pack" and "Usable remaining", both are simply the nominal values minus the buffer. Initially, I thought they were. However, nominalRemaining, nominalFullPack and buffer are reported via CAN bus (0x382).

SOC and the percentage shown in the car is the same. On the CAN bus there seem to be three different battery percentages. SOC, SOC UI and 'SOC min'. At 100% all are identical.
SOC goes down to 0% when 'usable remaining' is 0. That's what the car shows and that's when it also shows 0 miles left.
SOC min seems to be the true battery level. It includes the buffer.
SOC UI is somewhere in between. Not sure how it's calculated It doesn't make much sense to me.

 

[shred86]

I'm fairly certain only "SOC UI" and "SOC Min" are reported via the CAN bus. These bytes are reported under CAN ID 0x302 (source).

Scan My Tesla computes "SOC" by using the nominalRemaining, nominalFullPackEnergy and buffer, all of which are reported under CAN ID 0x382. You can see how it's being calculated for Scan My Tesla on the Github repo, specifically:

p.AddValue("SOC", "%", "br", (bytes) => soc = (nominalRemaining - buffer) / (nominalFullPackEnergy - buffer) * 100.0);

The fact it aligns perfectly with what's displayed in the vehicle suggest to me that's probably how the car is computing it as well.

TM-Spy supposedly uses "SOC UI" for the SOC that's displayed in the app. It appears to match exactly with nominalRemaining divided by nominalFullPackEnergy, thus I've always thought this was the capacity of the battery pack with the buffer (it indicates 0% when nominalRemaining is 0.0 kWh).

The same applies for "Usable full pack" and "Usable remaining" in Scan My Tesla. It's not reported via CAN bus, they're just calculated values which take nominalRemaining or nominalFullPackEnergy minus the buffer. Again, you can see this in the Scan My Tesla source code. If "Usable remaining" shows 0.0 kWh when the display hits 0% SOC or 0% rated miles, I'd imagine this is exactly how the car is calculating it as well (nominalRemaining minus the buffer).

Edit: Since I can't use Scan My Tesla (no android devices), I was curious what some of the other CAN bus reported values were (really just trying to figure out what TM-Spy displays) so I created a simple script in python to decipher raw CAN frames. I created a repository on Github for anyone that might be interested as well. Basically lets me convert raw CAN frames from TM-Spy recordings. Only supports a few CAN ID's right now and "SOC UI" isn't converting correctly for some reason... always shows a value too high so disregard that for now.

 

[ran349]

I'm fairly certain only "SOC UI" and "SOC Min" are reported via the CAN bus. These bytes are reported under CAN ID 0x302 (source).

Scan My Tesla computes "SOC" by using the nominalRemaining, nominalFullPackEnergy and buffer, all of which are reported under CAN ID 0x382. You can see how it's being calculated for Scan My Tesla on the Github repo, specifically:

p.AddValue("SOC", "%", "br", (bytes) => soc = (nominalRemaining - buffer) / (nominalFullPackEnergy - buffer) * 100.0);

The fact it aligns perfectly with what's displayed in the vehicle suggest to me that's probably how the car is computing it as well.

TM-Spy supposedly uses "SOC UI" for the SOC that's displayed in the app. It appears to match exactly with nominalRemaining divided by nominalFullPackEnergy, thus I've always thought this was the capacity of the battery pack with the buffer (it indicates 0% when nominalRemaining is 0.0 kWh).

The same applies for "Usable full pack" and "Usable remaining" in Scan My Tesla. It's not reported via CAN bus, they're just calculated values which take nominalRemaining or nominalFullPackEnergy minus the buffer. Again, you can see this in the Scan My Tesla source code. If "Usable remaining" shows 0.0 kWh when the display hits 0% SOC or 0% rated miles, I'd imagine this is exactly how the car is calculating it as well (nominalRemaining minus the buffer).

Edit: Since I can't use Scan My Tesla (no android devices), I was curious what some of the other CAN bus reported values were (really just trying to figure out what TM-Spy displays) so I created a simple script in python to decipher raw CAN frames. I created a repository on Github for anyone that might be interested as well. Basically lets me convert raw CAN frames from TM-Spy recordings. Only supports a few CAN ID's right now and "SOC UI" isn't converting correctly for some reason... always shows a value too high so disregard that for now.

Do you know if TM-Spy provides the readout on the phone for those parameters, or just the charts? I think Scan My Tesla does report all those values like SOC, nominal full pack, usable full, etc. I have the cable and Bluetooth attachment and am wondering if TM-Spy will give me what I am looking for. One thing I want to determine is what my designated buffer size is. Since I have a 70D, I think it is supposed to be 2.4 kWh, not the usually reported 4 kWh.

 

[David99]

Ok, thanks, but what about this question:
Let's say for example, you charge your car and after your charging is finished, the car dash says it added 50 kWh to the battery. Does that charge total go up by exactly 50 kWh, or something slightly different? My guess is the BMS would report a number slightly more.

There is a little difference. The car only shows whole numbers so it's hard to make an accurate statement. When charging 60 kWh at a Supercharger, the BMS reports aprox 1 kWH more than the car shows on the UI. TeslaFi.com reports the same thing. Not sure where the 1 kWh is lost.