EricGhost

Thanks Andre’, but I do not know if it can be generalized. The core is a LLSS linear least square solution for a 1rst and 2nd order polynomial (V,A,Ah variables) of the mentioned batteries discharged curves I found in internet, which KS says are mounted on my wheel. Not sure How general and accurate are the LLSS found coefficients for other batteries, for my KS18L looks like they are quite ok. No the app is a DIY on Android studio with graphic just for my phone

I think that, after 3-4 30Km trips with different tyre pressure (from 2.8 bar to 1.2bar) to simulate different Wh consumption, I got my goal of finding a reliable computing for the remaining range of my wheel (within a conservative safety margin intrinsic in the model of a real battery) The model is based on the discharge curves for Sanyo NCR18650GA and Panasonic NCR18650GA both 3.500Ah nominal There are two situations 0-2.75Ah almost linear in (V=volt,Ah=ampere hour,A=ampere) and >2.75Ah non linear The wheel is defined by 4packs of 20cells with cut-off at 60V, nominal 84-60V =24V working range where -> 2V error is about 10% . Thank to the logs of wheellog which register the values of: time,Volt,Current,Speed and Km, I can analize my trip by splitting the data into different current (A) ranges, 0A-5A,5A-10A,10-15A,.....,>30A. To use the current as a reference has two main reasons: 1-battery energy drops significantly at high current 2-current is related to the motor torque and therefore to the wheel efforts during its usage: strong acceleration/deceleration, climbing at high speed, etc. The blue data below the "BatteryStart" tell the story of my recent trip as a combination of different currents, and therefore where I stand in the actual wheel usage. The next table after the blue one tells me how much is left if I stay in a single ampere range and what is left if I continue with the same wheel usage. There can be some apparent mismatch in the number (see 30% 0.7 Km and 2.0Km) that is because two statistic are involved one with weigths according to the ampere range and another without (brutal average) just to appreciate the difference, and also if wheellog log is missing. The data gives me a better level of confidence where I stand with my left battery compared with the simple app battery %

@palachzzz my mind was on my computing and when I said the app does not lie I intended the voltage at static, which is the only reference I take as almost good to know the left energy at zero ampere. so when you stop you do know what is left for small load. The problem is which loads are you going to use? I’m trying to elaborate a near future from the recent past through the recording of wheelLog logs As I wrote everything change under load and of course the KS app statement about future km is completely unreliable, that is the reason I felt the need of a more accurate prediction! BTW MANY CURVES https://electricbike.com/forum/forum/batteries/18650/52414-3500mah-league-sanyo-ncr18650ga-panasonic-ncr18650ga-lg-mj1-samsung-inr18650-35e In these curves at 3.4V you have 2.8Ah for 0.7A now the area of the trapezoid (4.2+3.4)*2.8/2=10.5Wh full charge around 12.5 then 10.5/12.5=0.84 not so far 16% left against 27%

I thank you for the correct reference, I re-installed the 2.0.13 If you look back some posts you will see that I did start some computing to find out a proper prediction of left battery and I'm still working on it. There is a non linearity in the 1rst 10-15% from the full charge (but that is not important as you are in a safe area) and another after around 70% of batteries discharge (to be conservative). The KS official app does not lie (on the battery, KM is another story) so much if you consider a current around the 3-4A which means <=1A per pack in a 4pack like my KS18L, the batteries behaviour is almost nominal, thank to @palachzzz from the wheellog logs you can analyze your trips and see where you stand. As far as I saw in the discharge graphs (but internet can mislead also) all LG,Sanyo,Panasonic NCR18650GA 3500mh are to be considered reliable up to 30% as they can still provide a reliable 10A discharge current which means for a 4pack a 40A. From 100% to 30% the main key of having high currents it is that you waste batteries energy [Wh]=[Ah]*[V]. At higher currents [A] the voltage drops and batteries release less [Wh] therefore they need to use more [Ah] (than at lower currents) to meet in the end the motor [Wh] request. After 70% of batteries usage the situation it is tricky as it depends a lot on how much current -> torque you ask from your wheel, so I prefer to have the definition of left capacity and left capacity at 100% performance.

Sorry if I got lost somewhere but today I tried the app 2.0.15 on my KS18L, and I was scared as my batteries dropped so fast! I cut my usual trip in a half and went back home with a 32% static and a 25% while moving. Consumption was almost one third more than usual from a 90% to a 30% instead of 50% or a bit more ? I was thinking my batteries were damaged, but then I connect with the KS app and found out a battery level 47%! so 15% more which is more consistent with my experience Which wheellog version shall I use for KS18L ?

I have a ratio wheel/gps of 1.18 which is quite a lot of skiing around for one like me that goes almost straight. I think they made an over consideration on the wheel radius, maybe infinite rigid tyre so the biggest radius the more km. I agree do not muddle the two take out the unuseful cheap wheel turns * radius and put a gps

@palachzzz uploaded the 7z EUCrange.7z

@Lars Heeneman As I wrote my app is customized for my phone graphic as I’m not professional developer, and because I cannot invest time on such an issue. I just wanted to build a math battery discharge model to predict the range based in historical data. Now the model (see post) integrates the V*I*dt=dWh thanks to the wheelLog logs and if there are high currents the penalty on the discharge batteries is taken into account. Today I tested it successfully I was few hundreds from my home when the degraded mode kicked in. I asked if it could be implemented in wheelLog (in the end the batteries are a constant), the app will not be useful unless your phone is a 6” with a 1080x2160 1:2, but I can give you the android studio project I just need big email b cause it is 6mb zipped file.

Yes that’s the reason there is a coefficient km real/km app =82% so for me is 18%. All numbers are related to compensated km

Done! one click on my "Use WheelLog" and from your log files I got everything I need, thanks!

@palachzzz you’re right I have to input but because I cannot access the wheel data. What I have to input is start voltage and km and then End voltage and km. But if I can read wheel Volt and total distance, I would push a button at start to memorize the Vstart and Kmstart and then after a while let us say every 5 or 10 min you can evaluate and update the current Wh/km and make predictions which will differ by person, drive style, kind of path but the person is the driver, the currently driving style is in progress and the path is the current one, mountain,flatland,hillside,seaside....., so yes I can estimate remaining distance for all riders in the world, they do for cars tank fuel why not for battery? Yes discharge current can be tricky but I’m working on it, 2.8Ah to 3.5Ah is 30% difference the same % that trigger KS degraded behaviour You have V(t)*I(t)=W(t) power function in time and the area described by the W(t) is the energy Wh that you have used to make your km range so far. Unfortunately a strong current I(t) reduce battery capacity so you loose more energy than your used one, but I think it can be managed All other params are almost a constant. Anyhow now that I have your log file I’ll try to do it by accessing your file data ?

Ok I did not know of your changes nor that wheellog was working with ks18L, so I kindly ask you if it can be improved. I think the remaining % does not give an useful information as the residual km, I’m 60% so what? However to know that I can ride more or less 10km (or whatever) at 100% wheel performance it’s a better info that I have 20km before being n the range on a critical %battery is also a good info It’s true that this residual km range can change but you have the recently done Wh/km or you can insert your Wh/h In my opinion even a couple of km error are better than a % You are in a better position to compute because you can monitoring the used current which impact on the cells usage My calculations quite sure were brutal compare to yours, now I’ll try with a log of yours plus your infos to see

@palachzzz as the ks official app is quite inaccurate in km range and prediction of left range I did a rough app that helps me on the left range based on battery voltage But I'm not an app developer, so I have to input data manually. If you think people will be interested to have such info in your app I can give you the math or the source

They should think about a firmware virtual black box registering what’s going on in the controller and periodically sending packets data to the app to keep memory usage low on the firmware. That’s would help much more than verbal descriptions

I told you so