Battery capacity display that actually is accurate

[harrisonpatm]

I put this battery meter on my e-motorcycle build:
https://www.amazon.com/Battery-Capacity-Indicator-Monitors-Lead-Acid/dp/B08L8RDW17/ref=sr_1_4?crid=2G84J5SZBOV9A&keywords=72v+battery+capacity+gauge&qid=1662818320&sprefix=72v+battery+capacity+gauge%2Caps%2C97&sr=8-4

My battery is 24s LiFePO4. I read reviews for this product saying that it needs a few full charge/discharge/charge cycles to display accurate remaining battery life. Unfortunately this has not happened so far. I can say that the live voltage display is accurate, within .05V of my BMS reading. Problem is that as cheap as this meter is, it tries to reflect remaining capacity while also experiencing voltage sag during high-amp acceleration, leading to a wildly inaccurate reading. Even when not in use the percentage remaining is off. This hasn't been a dealbreaker, since I trust my BMS reading (which I have to access via my phone), I have a good idea of its range based on a couple of range tests, and I can do things like set the odometer when I charge 100%. I don't go on long excursions anyway.

Does anyone have recommendations of a battery capacity display that actually displays somewhat accurate SOC? Preferably under $50. Thanks!

 

[Chalo]

You can't read state of charge when the battery is under variable load. Sag will always be interpreted as lower SOC than what's actually remaining. But you can infer SOC from resting voltage, if you understand your battery's discharge voltage curve.

My point is, only read unloaded voltage, know what the numbers mean, and all you'll need is a simple voltmeter. This presumes the battery is in regular use and its voltage per state of charge is repeatable.

If you want to know state of charge while you're motoring, you'll need some kind of coulomb counter. I sometimes use cheap "watt meter" gauges from eBay, but they're wildly inaccurate and inconsistent one to the other, so they are only useful when comparing different readings from the same meter.

Here's one example (though this isn't the kind I've used before):
https://www.ebay.com/itm/281768704865

 

[harrisonpatm]

Sag will always be interpreted as lower SOC than what's actually remaining. But you can infer SOC from resting voltage, if you understand your battery's discharge voltage curve.

This I understand, it is what I am doing currently. It's just a bit more difficult with LiFePO4, since the discharge curve is nearly flat.

Let me ask in a different manner: what does a mass-produced EV use to show remaining battery life? If I had to guess, Id say it uses a more complex combination of current SOC, miles traveled since last charge, amps delivered on last charge, and estimated miles remaining based on this info, and not simply SOC.

 

[Chalo]

Let me ask in a different manner: what does a mass-produced EV use to show remaining battery life?

I'm sure most of them measure amp-hours in both directions.

 

[nicobie]

To get even a little close, the big boys (car manufacturers) take all the input signals including your usage habits and run them thru a algorithm controlled computer, which then makes a educated guess on how much range is left.

When it comes to ebikes, our brain is the 'algorithm controlled computer'. It's never wise to trust a BMS, even the smart ones aren't smart enough.

 

[amberwolf]

Sag will always be interpreted as lower SOC than what's actually remaining. But you can infer SOC from resting voltage, if you understand your battery's discharge voltage curve.

This I understand, it is what I am doing currently. It's just a bit more difficult with LiFePO4, since the discharge curve is nearly flat.

A wattmeter (even a cheap sub-$20 one off ebay/amazon/etc, like this random amazon link:
https://www.amazon.com/gp/product/B013PKYILS ) will help with that. It's not perfect, but if it reads charge and discharge current, you can use it to tell how much charge (Ah or Wh) you started with (assuming you start at empty), and read off the discharge Ah (or Wh) to subract that.

If you have a common charge/discharge port, you can simply reset the meter, charge up from empty, now it will show a certain amount of capacity. Don't reset the meter, but instead just ride, and discharging will decrease the amount you see. It's safe to assume it will take more Ah or Wh to charge than discharge due to balancing time if there is any, so you'll get a little less capacity (hundreds of mAh range usually) than the meter says it has left.

If you don't discharge fully to empty, then don't reset the meter, and it will then increase the capacity during charging, and you can use this to roughly (but more easily than by voltage) track available capacity. At some point the difference between charge and discharge amounts will grow enough to mess this up and you'd need to discharge to empty and reset and start over; that point can be determined experimentally if you need to.

Let me ask in a different manner: what does a mass-produced EV use to show remaining battery life? If I had to guess, Id say it uses a more complex combination of current SOC, miles traveled since last charge, amps delivered on last charge, and estimated miles remaining based on this info, and not simply SOC.

It's possible. The closest system for ebikes or other small-EVs I'm aware of, commonly available off the shelf, that does anything remotely similar is the Cycle Analyst v3 from Grin Tech http://ebikes.ca but it doesn't do everything listed.

I think there is at least one DIY wattmeter / bike computer with a thread here on the forum that was to do something like this, but I don't recall which one, or if it was ever finished.

There might be projects on Github and the like that could be used to do this, though you might have to merge more than one of them together into something to do what you want.

 

[john61ct]

Battery SoC% calculation is rarely accurate

Even with dedicated shunt based monitors costing hundreds of dollars, a 6-8% accuracy (well away from HVC 100% and 0% LVC) is considered state of the art.

Better to rely on the experimental method with YOUR unique rig and use case, and develop your intuition.

At rest 24hrs fully isolated is the only time voltage gives a good indicator, and you need to develop your own SoC correspondence chart,

bracketed by your definition of 100% and 0%

using an accurate shunt-based coulomb counting wattmeter and low C-rate loads.

Once you have a precise measurement of actual capacity at low rates

and then at your actual-usage rates (much lower)

You can get a feel for your SoC over a given ride route based on your throttle behaviour, hills involved etc.

So your nominal 20Ah pack may on some rides only deliver 11.5Ah before you should stop, on others maybe 14Ah

The only explicit objective indicator you can (maybe) rely on is, "how many Ah have I consumed so far since 100% Full?"

Of course realtime voltage display, ideally at 1S of the lowest cell/group

should act as your backup failsafe for when your intuition fails you.

 

[harrisonpatm]

Im building my intuition with everything stated above, so thank you for the reassurance that I'm doing it "right." Well, as least right for me. I also raised the low voltage cutoff levels in my bms. So if I ever were out and my controller shutoff, I can use my phone to readjust the bms and get another mile out of it safely. Reserve tank...

I'm finding the realtime voltage display quite useful to monitor voltage sag at high draw moments, like acceleration or steep hills. So at least it's somewhat useful.

 

[BareKuda]

Does anyone have recommendations of a battery capacity display that actually displays somewhat accurate SOC? Preferably under $50. Thanks!

If you don’t already have a smart BMS with Bluetooth, you can get one and it will not only show reasonably accurate SOC but also let you know whats going on as you charge it or use it on a trip.

I have a 20s LiFePO4 battery with ANT bms which has both Android and iOS apps and uses mainly coulomb counter since the voltage curve is too flat on LiFePO4 to use that.

I only charge mine mine to 3.44v/cell and told the BMS the battery is about 10% smaller than actual capacity so that it will have 10% left when the SOC shows 0%.

If it’s impossible to add a smart BMS you can buy a meter with a shunt that does coulomb counting for SOC. If the meter cant report current then no way it can report SOC accurately.


There are a few types, the type with shunt will be more accurate.