battery gauge accuracy.

[Burgerman]

Pilot plus system.
Odyssey batteries.

Yesterday, I went miles, all over the place, bank in town, to get parts for replacing light switches and wiring, etc covered about 8 miles but a lot of turning, shopping, cleaning up at home. Never lost a single light all day. The powerchairs gauge tells me the battery is all but full! When recharged, 41Ah returned. Thats basically a fully discharged battery with sod all usable (to a wheelchair) remaining. So the battery voltage would have just dropped low very rapidly in the next mile or so and then start flashing red... All at once.

Today, after full recharge I stayed home, did nothing much, sat about waiting for people to deliver things. Late evening I waterered the lawn. Lost all the battery gauge lights, except for one orange and the reds... So based on that gauge, no power left. Was that true? Nope!
Ah returned overnight? 13.9Ah... So in reality, the thing was reading the rapid voltage drop when I watered lawn on the grass due to the increased load. So it incorrectly thinks the battery was dead! While at least 2/3rds was actually remaining. Stupid things tell you almost nothing of any use.

Those gauges only tell you that the LEDs still work. They look official, pretty, but hey do not relate to state of charge. The R-Net one is not any better either. If you are attempting to figure out range left based on those things, good luck with that!

[Fedor]

My experience shows that it depends on the controller and the algorithm inside and type of battery also. I used DK Shark and the battery indicator helped to illuminate the joystick only. Beautiful, but useless. 90% of the time they showed a full charge, when the battery really weakens, one lamp disappeared and after 20 minutes of driving all the others disappeared.

With Curtis/Q-Logic the experience is better. There are only 7 indicators on the screen. They go out gradually and evenly, but this still does not show the actual state of the discharge of the battery. Somehow I left 2 indicators out of 7, should mean about 30%, then waited 24 hours and measured the voltage on the batteries (MK Gel), the multimeter showed 12.40v, which means more than 60% of the charge.

It is clear that this can be affected by battery aging, internal resistance, total capacitance, currents given, and much more. And in this case, the "normal" user needs to show "something" that talks about the state of the charge.

However, I would prefer instead of useless indicators the simplest built-in voltmeter on the screen. Still there will be no accuracy, but it is already more useful, i think.

[Burgerman]

However they work they cant be accurate.

Think about this. I can take 2 healthy relatively new batteries from a chair. Charge both fully.
I can use one to start a car. That takes about 1/4 of an Ah. But a huge current for about 1 second. Wait 5 mins. Measure voltage. It will be maybe 12.7V. It will be depressed, and will take around 12 to 24 hours to stop rising back to its 12.95V.

I can take a full 10Ah from the other battery. Slowly with a headlamp bulb over 2 hours. Its voltage will be HIGHER than battery A above, for the first 6 to 8 hours, then the other battery will overtake it... And the one with 10Ah removed will end up at 12.8V.

How can any battery meter know this, based only on voltage. Most average this voltage over a period, and some ignore the voltage anytime the joystick isnt central. Or they would be confused by the voltage drop under load. Even if they measured current taken out, they still cant know the battery history or temperature from yesterday... Or its age, its internal resistance since it cannot determine how much is resistance and how much is surface charge (peukert). The next issue is even more important.

If we decide a battery at 10.8V is fully discharged, which it is, and call 11V the last red, then that should work right? Well no. Because a battery that just climbs a ramp, or a short steep hill needs some amps to move the chair. The amount that this drops the voltage depends on 6 main things. Capacity, resistance, peukert, state of charge, battery cycle life, and recent discharge history. So what if its a standing chair, like the one on a different thread. Thats got just 36Ah high impedance UPS cheap batteries. It refuses to move the chair, when the battery voltage falls to 25.4V unloaded (almost fully charged). Due to the above affects. So the 11V (22v for a very discharged battery) is absolutely meaningless if the chair wont go! That would only work if the battery was massive. And had zero impedance. And zero peukert. So even knowing the actual state of charge isnt even helpful, unless you understand all the other parameters too. Because even a 90% charged battery may be inadequate to move a powerchair if that battery is small or high impedance. No battery meter knows this.

[Fedor]

It's true. But I think it's still not a rocket science and engineers can program presets for different batteries, know the manufacturer, discharge curve, battery capacity, current, internal resistance and more. In the controller, they can also install the simplest thermal sensor, and also store the history of all these indicators. Moreover, the controller even knows how and when the battery is charged. There will be a problem due to peukert and unexpected consumption of high current, but he can show a message about it and pause in the measurements to adjust the indicators. The whole range of works will significantly increase the accuracy and usefulness of these LEDs. But who needs it? Most do not even think about it.

[Burgerman]

Over the years hundreds of battery monitoring systems have tried. Theres too many variables. And even if they KNEW the EXACT state of charge, they dont know when the powerchair will stop.

E.G I once fitted some 100Ah brand new group24 lead acid deep cycle batteries by a company called elecsol. I was green. Saw 100Ah and my eyes lit up! Bright orange things. The battery gauge quite correctly told me they were fully charged. The chair didnt have the power to climb the ramp into my van when fully charged. Due to the voltage drop under load, triggering the controllers protection, which reduces current to prevent the battery voltage falling below that needed to maintain safe function. When they lost 1 light, about an hour later, and were still about 90% full, that was it. Out of range!!! With about 90Ah remaining! Chair wouldnt go back into the house. How did that battery gauge and knowing the state of charge help? The actual state of charge is often the least important metric.

The impedance of an MK gel, increases drastically as it becomes discharged. It was DESIGNED to run out of power long before it ran out of capacity. So at 12.2V its done for, with at least 25% of the Ah still remaining but not accessible by a powerchair. A so called starved acid design. Stops the user running it too low, since the electrolyte practically turns to water before all the Ah is used up, at the rate we need it. So no damage done by deep cycle use... Because you cant! The battery feels dead, long before it is. Increases cycle life.
The opposite battery type, the very low impedance 68Ah Odssey PC1500 runs the chair further with less capacity, and works when much lower in state of charge, and it has the power to climb a steep ramp even down to 11.5V or less where it is 99% discharged. So it keeps going long after the chair has stopped with an MK.

Too many variables for any battery fuel guage to ever make any sense of other than a approx guess and pretty lights.

[Burgerman]

Heres how I tell.

My chairs are programmed to respond. And I mean jump!
I can tell the difference between fully charged MKs and a set of Odyssey batteries in the first 10 feet of driving the chair. The MKs feel soggy, the edge that makes the chair jump isnt there. The steering is like elastic and isnt "solid". It loses its certainty, makes me slow down and take more care. Feels as if your cars steering wheel is connected by stretchy rubber bands. The fully charged MKs feel like a half discharged Odyssey. Soggy...

Now, I can feel the battery response sagging as I use it during the day. Starts good, deteriorates fast as the day goes on. When it gets to the point where I begin to really feel it to the point where it pisses me off and I have to plan my steering more, I know they are about 50% discharged. About 25 to 30Ah used up. And the guage usually shows full still. Or if I have been climbing a hill or darting left/right it shows low... This happens faster if I am zapping about all over, say pressure washing or leaf blowing the drive, or off road with the dog grabbing his ball or frisby. It takes a long time in straight lines covering distance. My seat of pants is way more useful and meaningful than the battery guage. And usually confirmed on top up. Since all my chargers have a way of counting Ah. (I may add, that lithium feels WAY more responsive and accurate and it does not change at all all day long.)

If my chair does not respond instantly, as this 90% turn accel test shows, I cannot feel this battery sagging. Because the delays introduced feel the same as a crappy battery! Set it to 100 turn acc /dec here instead, and it makes it far easier. Remember that this vid is an old 80A controller, and an old chair, very tired batteries, rear drive and a fat 20 stone user.
I drive like this most of the time for 2 decades. And if you do, you can FEEL the battery sag as it ages or discharges. Battery meters be damned - they tell me nothing useful. http://www.wheelchairdriver.com/gopro/control.mp4

[Fedor]

I agree, in your usecase it is impossible to know exactly the state of the charge in the battery. Nevertheless, my point is that the manufacturer can at least improve the accuracy of the indications. Or make a displayed voltmeter, which will give a "quiet" mode of use (driving along the road with the same route) at least an approximate state of the battery. And also useful when charging. Now, on most controllers, the state of these LEDs looks ridiculous.

[Burgerman]

Yes. They are a feelgood thing. Most of the day they say full. Once the battery is done for, they drop... Or not. Depending on a bunch of stuff. What they cant tell you is how much is left.

[Burgerman]

In R-Net you can display Volts. And Amps.

L Motor volts,
R Motor volts,
Battery volts,

L Motor Amps,
R Motor Amps,
Battery Amps.

And do so in real time on a laptop screen as you drive.

[expresso]

u can display pack volts on JS with r net ?

[Fedor]

Yes. They are a feelgood thing. Most of the day they say full. Once the battery is done for, they drop... Or not. Depending on a bunch of stuff. What they cant tell you is how much is left.

Not so dramatic. On Curtis for 3 years if I travel along the same route, then the LEDs disappear almost always in the same places. Day by day. With similar weather and time of year. This tells me that there is some kind of correlation if you use a daily route, without strong sudden changes in current consumption. And not old batteries.

[Burgerman]

Oh they corrolate with something! We just dont know what! Temp makes a big difference. Cold = less capacity, and increased resistance/peukert. And slower aging. And higher charge voltages needed.

If I go to the end of my street I meet a main road. Right = uphil slightly. Left = downhill.

If I go right, and go fast, I lose most of the lights in a couple of miles, looks like I cant get back. If I stop, have a . 2 hours later all the lights return. If I turn left out of my street, I can go 4 miles into town, and not lose a light. On the way back I lose most of them... But again, they come back a couple of hours later if I turn off and back on. Interesting but meaningless. As you say a volt meter is more helpful, but it also cannot tell you whats left. A hobby watt meter is more helpful. I know that I can get about 40Ah from a set of grp24s normally.

But not always. So at least you can count Ah. But a 74Ah gel, will give you 30Ah at very best going uphill at full speed before voltage drops to unusable, and you stop. And 50Ah going downhill because of peukert and the fact that 5A is enough to keep going. But that may be 3 miles uphill and 20 or better downhill...

[Fedor]

Yes, with a non-typical route, with a bunch of hills, falls, in a park with grass, on the ground, it is absolutely impossible to determine the actual state of the battery if you do not have a day to stand still staring at the ducks on the lake.

If I go right, and go fast, I lose most of the lights in a couple of miles, looks like I cant get back. If I stop, have a . 2 hours later all the lights return.

This interesting behavior is most likely due to the controller algorithm. On my Curtis with the new MK gel, if I lose 5 LEDs, they will never come back until I get up for recharging. In the experiments, I waited about 24 hours, and the battery indicator on the joystick was never restored.

[Burgerman]

Yet the battery volts inevitably do. Its trying to pretend it was right, and refusing to climb! Charging (exceeding float voltage or about 27V) tells it its safe to admit it was wrong!

[jeffreyclay]

This might give a lead user some sort of power-remaining indicator. It features a cumulative watt/hr reading with reset. You'd want to off-board the display from the shunt.
https://www.banggood.com/300V100A-Digit ... rehouse=CN

[Burgerman]

And cheap. But with with a few provisos. Every time you decelerate, or decend a hill/ramp the current goes the opposite way and puts power back into the battery. And it ignores that. And every time you turn on the spot or exceed 100A (An R-Net 120 can do 120A per motor, so 240A peak) it misses the extra power too. Which can be quite substantial. To get an idea how much is regenerated you can fit it the wrong way around. Can be a lot on some trips, or very little to non used indoors at low deceleration levels.

But those two things more or less cancel each other out. I use a Turnigy Watt meter that is basically the same thing. https://hobbyking.com/en_us/turnigy-180 ... tore=en_us
Its more useful than the usual battery meter to know Ah used obviously. Ignoring the two small things above. But it still cant tell you when the chairs going to stop moving. It does this say after 30A is used going across grass, or up a slight hill. And after maybe 45Ah is used going on the flat in straight lines... But its still more informative than the voltage based guage.

[jeffreyclay]

Good points all but if you knew your watt/hr capacity for a safe discharge limit a reset-able meter that can read cumulative watt/hr use should be close on predicting your last 20ft of travel. Then you could pause for a pint and don your PV sombrero for a self-rescue charge. There's a school of thought that says if you reduce your front caster diameter 25% to cause your chair's frame a forward tilt, you will fool it to think it's always going downhill and you recoup all your spent energy by regeneration!

[Burgerman]

Good points all but if you knew your watt/hr capacity for a safe discharge limit a reset-able meter that can read cumulative watt/hr use should be close on predicting your last 20ft of travel.

Thats the problem. There isnt one. Its not Ah remaining that stops the chair, its impedance and peukert which is variable depending on load and discharge history and battery *efficiency.

Start 8am and use in the house only, all day long watching TV eating etc. For 2 days. Chair stops, and you used 45Ah
Set off at 8am, at 8mph, and you may get 16 miles. takes exactly 2 hours. And chair stops. You used 30Ah.
Set off at 8am, and spend an hour working hard in the garden cutting grass, or clearing leaves with leaf blower back and forth, and battery dead as well as controller overheated and rollback... 25Ah used. Battery will recover some later on. But heavy peukert losses are not fully recoverable even a day later. *The charge level stratification inside the battery cells (peukert) wastes energy as heat and gassing/recombination as it re equalizes.

Whats worse is that its the increased dc resistance that stops the chair and not the amount of Ah remaining or already used. And that depends on load, and discharge history. The reason that you measure AC resistance (impedance) is because that is constant. But DC resistance increases to 10x or 20x higher in the first few seconds of discharge and continues to increase depending on load/history.