Speak EV - Electric Car Forums banner

1 - 6 of 6 Posts

·
Registered
Joined
·
2,234 Posts
Discussion Starter #1
I've taken to accept the battery meter in a LEAF is not linear. A quick probe with LEAFSpy seems initially to agree with this, the gauge drops slower than SOC initially, and then faster than SOC once it gets down below 50% or so.

However I noticed something on friday while driving to work with LEAFSpy running.

LEAFSpy shows "kWh remaining" which on my car with a full charge starts at 25.0 (or sometimes 25.1 dependnig on how it feels)

I happened to glance at the readouts at one point, and noticed that the dash was displaying 80% and LEAFSpy was showing 20.0kWh remaining. The fact that it lined up so perfectly piqued my interest. So I started paying close attention, and discovered that its almost perfectly linear, every 4% of the battery meter, i would lose 1kWh. This held true from 80% right down to 36% when i arrived at work.

So it appears it IS infact linear, but its linear based on GID's/KWH available at a full charge. It doesnt track SOC linearly, because the car keeps 10% SOC in reserve at the bottom.
 

·
Registered
Joined
·
4,426 Posts
Don't forget to take into account the decline in voltage as the battery discharges when considering the linearity of the SoC meter.

Raw SoC in a BMS is usually based on remaining Ah capacity vs full charge Ah capacity, (since the BMS fundamentally tracks Ah in and out of the battery to track state of charge) and each Ah has less usable energy when the SoC is low because the voltage is lower. energy = time x voltage x current, and at low SoC voltage is reduced.

So if the meter was a perfectly linear representation of remaining Ah capacity, it would seem non-linear because you could go further on the first 50% than you could on the last 50%, even if there was no top/bottom buffer because the voltage is lower at the lower SoC.

BTW, we don't really know what GID's are - they're a reverse engineered unit which seem kinda arbitrary. While Leafspy assumes they are kWh using a very strange unit of measurement (which you can adjust in the settings, making it clear that nobody is completely sure what they are) they could just as easily be a measure of Ah, and without closely analysing energy in/out taking into account voltage changes it would be difficult to tell for sure whether it represented kWh or Ah.

My hunch is that the SoC meter on the dashboard is based on Ah not kWh, and that GID's are actually a measure of Ah not kWh. This would inherently make the SoC appear non-linear.

I've noticed on my car that for a constant miles/kWh I can't drive nearly as far on 20% charge near the bottom end as I can on 20% near the top end, and the difference is quite significant. This is consistent with the SoC meter reporting Ah capacity not estimated kWh.

Your last comment that a 10% reserve buffer at the bottom would make the SoC meter non-linear doesn't make sense by the way. If you reserve 10% at the bottom so that raw SoC of 10% = dashboard SoC of 0% and (for the sake of argument) 100% Raw = 100% reported, then you can do a simple linear interpolation between them. There is no inherent reason for this to become non-linear.

I think you'll find the non-linearity is simply Ah vs kWh. The meter probably reports Ah linearly (with some linear scaling due to top and bottom buffers) which gives a non-linear approximation of kWh.

Energy to propel the car is measured in kWh while state of charge of a battery is based on Ah and there is not a fixed linear relationship between them. (This non-linear relationship between kWh and Ah is also load and temperature dependent)

One further thought about "non-linearity" of the SoC reading on an EV is that over most of the middle portion of the SoC range the BMS relies on Coloumb counting (Ah in vs Ah out) to keep track of the SoC, but near the bottom (and top) it will start using voltage estimation to re-calibrate the starting point for the coloumb counting to correct for drift.

If there is a large discrepancy between the Coloumb counting Ah remaining figure and that which voltage estimation calculations this may cause a fairly sudden adjustment in the reported SoC to occur.

I used to see this on my Ion once the battery had degraded and had some dodgy cells - I would arrive at a rapid charger at say 35% SoC (raw reported SoC using Canion) turn the car off, plug into the charger and suddenly the reported SoC is only 22% or similar. This is because turning the car off then "on" again (during rapid charging) caused the BMS to do a voltage estimation and decide it wasn't happy with it's previous SoC estimation. It would do something similar after charging where it under reported SoC at the end of the rapid charge but after unplugging and turning the car on there would be a 10-15% jump in claimed SoC.

So at those points where voltage estimation starts to come into play (above about 80-90% and below about 30%) there can be discontinuities in reported SoC if the BMS is not very smart or there are some faulty cells. Measuring and tracking SoC and SoH is not an exact science.
 

·
Registered
Joined
·
121 Posts
Don't forget to take into account the decline in voltage as the battery discharges when considering the linearity of the SoC meter.

Raw SoC in a BMS is usually based on remaining Ah capacity vs full charge Ah capacity, (since the BMS fundamentally tracks Ah in and out of the battery to track state of charge) and each Ah has less usable energy when the SoC is low because the voltage is lower. energy = time x voltage x current, and at low SoC voltage is reduced.

So if the meter was a perfectly linear representation of remaining Ah capacity, it would seem non-linear because you could go further on the first 50% than you could on the last 50%, even if there was no top/bottom buffer because the voltage is lower at the lower SoC.

BTW, we don't really know what GID's are - they're a reverse engineered unit which seem kinda arbitrary. While Leafspy assumes they are kWh using a very strange unit of measurement (which you can adjust in the settings, making it clear that nobody is completely sure what they are) they could just as easily be a measure of Ah, and without closely analysing energy in/out taking into account voltage changes it would be difficult to tell for sure whether it represented kWh or Ah.

My hunch is that the SoC meter on the dashboard is based on Ah not kWh, and that GID's are actually a measure of Ah not kWh. This would inherently make the SoC appear non-linear.

I've noticed on my car that for a constant miles/kWh I can't drive nearly as far on 20% charge near the bottom end as I can on 20% near the top end, and the difference is quite significant. This is consistent with the SoC meter reporting Ah capacity not estimated kWh.

Your last comment that a 10% reserve buffer at the bottom would make the SoC meter non-linear doesn't make sense by the way. If you reserve 10% at the bottom so that raw SoC of 10% = dashboard SoC of 0% and (for the sake of argument) 100% Raw = 100% reported, then you can do a simple linear interpolation between them. There is no inherent reason for this to become non-linear.

I think you'll find the non-linearity is simply Ah vs kWh. The meter probably reports Ah linearly (with some linear scaling due to top and bottom buffers) which gives a non-linear approximation of kWh.

Energy to propel the car is measured in kWh while state of charge of a battery is based on Ah and there is not a fixed linear relationship between them. (This non-linear relationship between kWh and Ah is also load and temperature dependent)

One further thought about "non-linearity" of the SoC reading on an EV is that over most of the middle portion of the SoC range the BMS relies on Coloumb counting (Ah in vs Ah out) to keep track of the SoC, but near the bottom (and top) it will start using voltage estimation to re-calibrate the starting point for the coloumb counting to correct for drift.

If there is a large discrepancy between the Coloumb counting Ah remaining figure and that which voltage estimation calculations this may cause a fairly sudden adjustment in the reported SoC to occur.

I used to see this on my Ion once the battery had degraded and had some dodgy cells - I would arrive at a rapid charger at say 35% SoC (raw reported SoC using Canion) turn the car off, plug into the charger and suddenly the reported SoC is only 22% or similar. This is because turning the car off then "on" again (during rapid charging) caused the BMS to do a voltage estimation and decide it wasn't happy with it's previous SoC estimation. It would do something similar after charging where it under reported SoC at the end of the rapid charge but after unplugging and turning the car on there would be a 10-15% jump in claimed SoC.

So at those points where voltage estimation starts to come into play (above about 80-90% and below about 30%) there can be discontinuities in reported SoC if the BMS is not very smart or there are some faulty cells. Measuring and tracking SoC and SoH is not an exact science.
Thanks for the explanation - i think i've a general grasp of what's going on. I've a simple question - how well can a simple readout/meter perform in estimating range ? Particularly a WattsLeft or LeafDD (or similar - but something simpler than LeafSpy for ease of use while driving...). These display CANbus (?) data from the battery, so one can see exactly how many Kwh are present. I was hoping that this would make range estimation simpler (i've a 1st gen with ageing battery) - but this may not be the case if the Kwh/distance travelled ratio changes significantly between the top 100% and the bottom 20% of the battery.
 

·
Registered
Joined
·
2,234 Posts
Discussion Starter #4
well as per my original post, the dashboard % reading follows the LEAFSpy reported "kWh available" perfectly. If you know how many kWh are reported when its full, then you know how much is remaining.

DBM's post calls into question the accuracy of Leafspys readings, but really we dont know either way. Personally my view is that theres been a lot of time and effort spent on LEAFSpy over the years, i would be inclined to trust it. If they are incorrect, then any of these similar tools will be equally incorrect, as they're all doing the same thing.
 

·
Registered
Joined
·
2,202 Posts
For my 13Reg 24kWh Leaf, I have a higher %SOC used per mile travelled at the start of a journey than later on (? due to changes in battery temperature?) and the GOM seems to know this as (starting at 100% SOC with the mileage trip set to zero) the trip mileage usually equals the GOM reading at around 45% SOC.

Also, if I set the car to charge to 80% SOC (actually displaying 79% SOC when finished), the %SOC used on a regular 7.5 mile journey is higher than if I set it to charge to 100% SOC.
 

·
Registered
Joined
·
2,234 Posts
Discussion Starter #6
How do those figures compare to what LEAFSpy reports for GIDs and energy available?
 
1 - 6 of 6 Posts
Top