Running the Battery to 0%

[MickeyAO]

I could write an app that logged data from the vehicle, recorded every available variable (both actual and derived) the truck is using\recording, while also recording data from various external instruments to measure the environment the truck is operating in.

The next step after that would be instrumenting the truck itself, giving an independent but parallel set of data to compare to.

Been there, done that, sorry I can't share the video and data
Now working on a Model Y with the 4680 cells.

Sponsored

 

[Box Cat]

If I could get a group of volunteers to gather the data from their vehicles and give me the maintenance records for their vehicles. It should be fairly direct to apply some predictive failure analysis to the data with the added dimension of being able to see some components degrade in real time.

You can count me in. I haven’t been able to play a lot with the truck but I have been able to collect several hours of data. This includes going down to 0% SoC, going from 0% to 100% in a controlled environment with battery temperature close to nominal values, and a voltage relaxation experiment. I have written a Python package to process the data imported from Car Scanner and put in a more manageable state for analysis and reporting. I was planning to post in the Forum but I don’t seem to ever have the time.

 

[SpaceEVDriver]

Start wi

I think I should take the time to pull the battery management software and other battery related software off the truck and spend a few weekends reverse engineering it. So much of this interpretation of the physical state of the battery is being done in software. I'd like to know why it took 15 miles at a steady 48mph for the battery to go from a displayed 100% to 99% but by the time I completed a 42 mile circuit, the displayed values matched up to a reasonable 3 m/kwh for that speed.

Anyway, this thread has been amazing if only to validate my questions about the behavior of the truck is caused by more than sparse data sampling.

I wonder how much @Ford Motor Company would be irritated if I posted a graph of what the software is doing at a range of charge levels and possibly a list of variables they are using in their calculations....

I could write an app that logged data from the vehicle, recorded every available variable (both actual and derived) the truck is using\recording, while also recording data from various external instruments to measure the environment the truck is operating in.

The next step after that would be instrumenting the truck itself, giving an independent but parallel set of data to compare to.

This paper won't tell you exactly what Ford is doing, but if you really wanted to dig into this, it's a start. There are likely better papers out there for the specific battery chemistry, but this will give you an idea of one of several of the possible algorithms that Ford is using to estimate SoC:

https://ris.utwente.nl/ws/files/6547139/MST_05.pdf

 

[Yellow Buddy]

sorry I can't share the video and data

Which is why we have to try to figure it out by posting wildly inaccurate information to get your reaction.

 

[Yellow Buddy]

I think I should take the time to pull the battery management software and other battery related software off the truck and spend a few weekends reverse engineering it. So much of this interpretation of the physical state of the battery is being done in software. I'd like to know why it took 15 miles at a steady 48mph for the battery to go from a displayed 100% to 99% but by the time I completed a 42 mile circuit, the displayed values matched up to a reasonable 3 m/kwh for that speed.

Anyway, this thread has been amazing if only to validate my questions about the behavior of the truck is caused by more than sparse data sampling.

I wonder how much @Ford Motor Company would be irritated if I posted a graph of what the software is doing at a range of charge levels and possibly a list of variables they are using in their calculations....

I could write an app that logged data from the vehicle, recorded every available variable (both actual and derived) the truck is using\recording, while also recording data from various external instruments to measure the environment the truck is operating in.

The next step after that would be instrumenting the truck itself, giving an independent but parallel set of data to compare to.

Decided to bump up the OBD reading from CarScanner against FORScan today and the values were exactly the same.

Doesn’t help a ton but it at least eliminates some of the potential concerns about app interpretation.

 

[Box Cat]

Start wi

This paper won't tell you exactly what Ford is doing, but if you really wanted to dig into this, it's a start. There are likely better papers out there for the specific battery chemistry, but this will give you an idea of one of several of the possible algorithms that Ford is using to estimate SoC:

https://ris.utwente.nl/ws/files/6547139/MST_05.pdf

Thanks for sharing. I suppose that if a non-proprietary solution is used, one (teardown guys?) could visually inspect the BECM PCB and identify the chipset used.

But then there would be no fun in reverse engineering !

 

[SpaceEVDriver]

Thanks for sharing. I suppose that if a non-proprietary solution is used, one (teardown guys?) could visually inspect the BECM PCB and identify the chipset used.

But then there would be no fun in reverse engineering !

I expect it's software defined rather than hardware defined.

We've seen changes to the battery SoH being reported after a software update on the Mustang. The reported SoH is now much more reliable and seems to make more sense than it did before that software update, but the initial change caused a sudden (and worrying to some people) drop in reported state of health.

 

[Box Cat]

I expect it's software defined rather than hardware defined.

We've seen changes to the battery SoH being reported after a software update on the Mustang. The reported SoH is now much more reliable and seems to make more sense than it did before that software update, but the initial change caused a sudden (and worrying to some people) drop in reported state of health.

I agree. There would always be a SW component anyway since manufacturers would not want to lose the ability of patching, reconfiguring. HW changes are far more involved.

 

[Helium]

I expect it's software defined rather than hardware defined.

We've seen changes to the battery SoH being reported after a software update on the Mustang. The reported SoH is now much more reliable and seems to make more sense than it did before that software update, but the initial change caused a sudden (and worrying to some people) drop in reported state of health.

I was just thinking about whether I should post about this very thing. A couple of weeks ago I took my truck in to do recalls. Of course, being an efficient service tech with a profit motive, he updated the software. In a before and after comparison my SoH went from 97% to 98.5% (still is as of this morning and the odometer is reading > 105K).

Do I believe the service tech magically healed my battery....nope. But considering Ford is gathering tons of data from thousands of vehicles in the real world, I'm willing to bet your paycheck they are able to make a better estimate with tighter tolerances now.

I'm convinced at this point, I'll be posting a 300,000 mile update in 4 years or so and the battery will be healthy enough (>95% maybe) to where I'll be debating whether I should keep it in the truck or pull it out and use it for solar power storage.

 

[SpaceEVDriver]

I was just thinking about whether I should post about this very thing. A couple of weeks ago I took my truck in to do recalls. Of course, being an efficient service tech with a profit motive, he updated the software. In a before and after comparison my SoH went from 97% to 98.5% (still is as of this morning and the odometer is reading > 105K).

Do I believe the service tech magically healed my battery....nope. But considering Ford is gathering tons of data from thousands of vehicles in the real world, I'm willing to bet your paycheck they are able to make a better estimate with tighter tolerances now.

I'm convinced at this point, I'll be posting a 300,000 mile update in 4 years or so and the battery will be healthy enough (>95% maybe) to where I'll be debating whether I should keep it in the truck or pull it out and use it for solar power storage.

A full discharge-recharge cycle: down to <10% SoCD and back up to 100% SoCD on L1 or L2, without driving once the recharge starts will recalibrate the SoH metric. I do this about once or twice a year. Best to do it at ambient temperature of 70-80 ⁰F.
The coulomb counting part of the algorithm drifts as small errors accumulate.

I expect to get 600,000-1.5M miles from both the Lightning and the Mustang.

 

[Helium]

A full discharge-recharge cycle: down to <10% SoCD and back up to 100% SoCD on L1 or L2, without driving once the recharge starts will recalibrate the SoH metric. I do this about once or twice a year. Best to do it at ambient temperature of 70-80 ⁰F.
The coulomb counting part of the algorithm drifts as small errors accumulate.

I expect to get 600,000-1.5M miles from both the Lightning and the Mustang.

I would be ecstatic to get 600,000 miles out of the truck. Even with periods of expensive maintenance, this would be amazingly economic for me. At my current usage rate, which will slow dramatically in a few months, I could keep it on the road for 20-30 years. Imagine have a working first gen electric lightning on the road in 2052. I better Rhino line the under body now

 

[SpaceEVDriver]

I would be ecstatic to get 600,000 miles out of the truck. Even with periods of expensive maintenance, this would be amazingly economic for me. At my current usage rate, which will slow dramatically in a few months, I could keep it on the road for 20-30 years. Imagine have a working first gen electric lightning on the road in 2052. I better Rhino line the under body now

To be honest, I'll probably be looking at the next new thing by 500,000 miles. Then I'll do what you mentioned: put the Mustang out to the PV pasture so it can provide battery backup for the workshop, put the Lightning out to PV pasture so it can provide battery backup for the house...

 

[Box Cat]

A full discharge-recharge cycle: down to <10% SoCD and back up to 100% SoCD on L1 or L2, without driving once the recharge starts will recalibrate the SoH metric. I do this about once or twice a year. Best to do it at ambient temperature of 70-80 ⁰F.
The coulomb counting part of the algorithm drifts as small errors accumulate.

I expect to get 600,000-1.5M miles from both the Lightning and the Mustang.

Another factor affecting SoC (and I assume SoH as well since it reduces the overall SoC) is cell/module balancing. If a module SoC is out of whack with respect to other modules it will drag down the whole battery SoC. According to the shop manual, module rebalancing is performed whenever trigger conditions are met: voltage variation detected, SoC greater than 5% and ignition off. If that happens modules (cell groups) with higher voltage are discharged over a 48h period of time. After that, a new cycle can be started by cycling on and off the ignition key. So effectively, this will be happening without you even knowing, unless you constantly drive .

 

[Helium]

unless you constantly drive .

Nearly everyday...

 

[Grumpy2]

According to the shop manual, module rebalancing is performed whenever trigger conditions are met: voltage variation detected, SoC greater than 5% and ignition off.

That is interesting. Any other similar entries in the shop manual? Is the cell rebalancing timing conditions addressed?

Sponsored