Topic: Inaccurate SoC indication? (Read 1724 times)

Doug S · « **on:** September 04, 2017, 06:11:29 AM »

So yesterday I took my first semi-long trip with the Diginow charger (in 100 degree weather, excellent timing, eh?). I went to have lunch with my father and the rest of our family for his 86th birthday; a round trip of ~200 miles, so I'm not quite a member of the 250-mile club yet. I'll get there soon enough.

The charger worked like a champ and I can't say enough good things about it. What did not, distressingly, was the bike's SoC indicator, either on the dash or on the app. Long story short, after thinking through what I saw yesterday, and discussing it with Brandon, I'm quite sure that the SoC indication can be pretty wildly inaccurate.

I recharged three times on the trip. The first two times I didn't do a full charge, trying to take advantage of the fact that the CC charging phase ("bulk charging") is the fastest, and not get into the CV phase, which is slower. That means you only get back up to 80% or so, but it improves the ratio of time spent riding to time spent charging That worked fine, but afterwards, the SoC indication didn't seem right.

The third time, I was in no hurry and nobody was waiting to charge, so I just let it finish. The cell voltage got up to 116, charging current dropped to 8 amps, but the SoC indicator on the app still said 65%! That's just wrong -- 8 amps at 116 volts is not only CV mode, it's trickle charging. I walked over to the bike, and on the dash, it read 100%, which IS correct.

So my conclusion is the firmware doesn't track SoC very well through partial charges, especially multiple partial charges. But once the battery is FULLY charged, the SoC is reset to 100%, and of course that's accurate. Partial charging might be effective at reducing charging time, but it messes up the SoC indication.

I'm seeing now how it could be very useful to have a voltage indicator where you can see it 100% of the time. It's sad that the firmware doesn't work better, and I'm still a little concerned that it may make decisions like cutting back power when the SoC is still pretty high. But that's what we have to work with at the moment.

Is that what others have experienced? Am I misinterpreting something?

Keith · « **Reply #1 on:** September 04, 2017, 07:03:25 AM »

I'm doing a lot of data gathering based on voltage controlled partial charging. What I've seen is not unlike what you describe. I've found things like this in the log files that may help to explain it. There are two different SOC modes and when they change so do the numbers. I can say that voltage is not a perfect SOC indicator. I've seen the voltage increase when SOC is decreased, due to temperature I think. And the voltage changes over time after charging while the SOC is not changing.

Code: [Select]

 09060     08/29/2017 21:45:39   DEBUG: Switching to coulomb couning SOC mode. SOC: 60%, VSOC: 61%, I: -2943mA    
 09061     08/29/2017 22:41:57   DEBUG: SOC off by over 5%, adjusted from 70% to 79%    
 09062     08/29/2017 23:27:52   Charged To Full            003 AH, SOC: 89%,         L:4024,         H:4027, B:003, PT:027C, BT:037C, PV:112737
 09063     08/29/2017 23:41:30   DEBUG: SOC off by over 5%, adjusted from 91% to 97%

Code: [Select]

09283     08/30/2017 18:18:32   DEBUG: Switching to coulomb couning SOC mode. SOC: 80%, VSOC: 75%, I: -7892mA    
 09284     08/30/2017 18:19:30   DEBUG: Switching to coulomb couning SOC mode. SOC: 80%, VSOC: 78%, I: -22073mA    
 09285     08/30/2017 18:22:39   DEBUG: Switching to voltage SOC mode. SOC: 82%, VSOC: 81%, Ihall: -401mA, Iext: 0mA    
 09286     08/30/2017 18:23:40   DEBUG: Switching to coulomb couning SOC mode. SOC: 82%, VSOC: 80%, I: -5351mA    
 09287     08/30/2017 18:38:54   Charged To Full            002 AH, SOC: 89%,         L:4027,         H:4030, B:003, PT:031C, BT:042C, PV:112769
 09288     08/30/2017 19:09:33   DEBUG: Switching to voltage SOC mode. SOC: 100%, VSOC: 100%, Ihall: -1605mA, Iext: 0mA    
 09289     08/30/2017 19:09:52   SOC adjusted for voltage   old:   173418uAH (soc:100%), new:   0uAH (soc:100%), low cell: 4147 mV

BrianTRice@gmail.com · « **Reply #2 on:** September 06, 2017, 12:58:35 AM »

This is a great set of finds, Keith. I've never seen those events linked up that way.

I've updated the wiki with some information I gathered around this, partly from wikipedia:
http://zeromanual.com/index.php/Unofficial_Service_Manual#State_of_Charge

gyrocyclist · « **Reply #3 on:** September 06, 2017, 06:39:33 AM »

Quote from: BrianTRice on September 06, 2017, 12:58:35 AM

This is a great set of finds, Keith. I've never seen those events linked up that way.

I've updated the wiki with some information I gathered around this, partly from wikipedia:
http://zeromanual.com/index.php/Unofficial_Service_Manual#State_of_Charge

Quick anecdotal: I have two of the "older" elcron chargers, which I've used maybe a dozen times. When I unplug, the reported SOC from the Zero dashboard is up to ~10% higher than what was reported on the iPhone app.

togo · « **Reply #4 on:** September 06, 2017, 12:07:37 PM »

> Is that what others have experienced? Am I misinterpreting something?

Yeah. The 2014 SR are reportedly the worst. Partial charge and discharge seems to rachet the inaccuracy up or down, depending.

I've seen people have trouble charging because SoC and resultant battery capacity calcs were so far off that it did a 1C limit way low and would open the contactor. On my particular bike SoC tends to run high. I used to consistently run out at about 6% SoC and the firmware update fixed that, but if I do partial charges and partial discharges, the gets so innaccurate that I've seen it at 52% when actually empty. Zero's offered to replace the MBB, but I'm having too much fun riding it right now. Maybe when it's rainiy and cold sometime...

The awesome thing about rapid charging is that my Zero has become my first choice on almost every trip anywhere. It's so damn fun, and there's never an "out of power" or a real "you'll have to wait a long time somewhere" excuse.

So I've learned to live with not trusting SoC or estimated range, I have an idea how far I can go on a charge, and I tend to charge more from 40% or so to 116v, which registers as full, but really is more like 90% (i.e. do the constant current phase, skip the constant voltage top off, because I'd rather ride than charge slow). Connecting and disconnecting is fast, and there are things to do at the stops, so I just charge more often.

BTW, did you ever notice the battery charges faster at higher voltage? It's true.

Keith · « **Reply #5 on:** September 06, 2017, 05:53:26 PM »

Here's more data. Red is the same as before, 8 amp charge current, blue is 4 amp. I'll do 16 amp today IF it stops raining...

Doug S · « **Reply #6 on:** September 06, 2017, 08:14:47 PM »

Quote from: togo on September 06, 2017, 12:07:37 PM

Yeah. The 2014 SR are reportedly the worst. Partial charge and discharge seems to rachet the inaccuracy up or down, depending.

Based on my single experience, it seems to be cumulative. It seems like it read maybe 10-15% low after the first charge, 20-25% after the second, and 35% after the third, until I let it hit maximum cell voltage and it reset itself to 100%. I'm assuming the firmware is smart enough to do that when the cell voltage tops out -- one of the few guarantees we have is that when the cell voltage max'es out and charge current drops close to 0, you are by definition at 100% charge.

Quote

The awesome thing about rapid charging is that my Zero has become my first choice on almost every trip anywhere. It's so damn fun, and there's never an "out of power" or a real "you'll have to wait a long time somewhere" excuse.

I could see that, provided you really can tell when you have power and when you don't. I'm already planning trips I've put off too long. The only downside is that it is a lot slower than driving -- ride an hour, charge an hour isn't a very hard pace to beat.

Quote

So I've learned to live with not trusting SoC or estimated range, I have an idea how far I can go on a charge, and I tend to charge more from 40% or so to 116v, which registers as full, but really is more like 90% (i.e. do the constant current phase, skip the constant voltage top off, because I'd rather ride than charge slow). Connecting and disconnecting is fast, and there are things to do at the stops, so I just charge more often.

I think I'll adopt a more conservative approach (never thought I'd describe myself that way!). I'll just recharge until it hits max cell voltage and resets to 100%. There's a lot less uncertainty that way.

Quote

BTW, did you ever notice the battery charges faster at higher voltage? It's true.

Are you saying the delivered kW rating goes up until it hits CV mode? I hadn't noticed that, but it makes sense. If you're in CC mode @ say, 65 amps, as the cell voltage rises that will represent a higher delivered charging power, since of course power equals current times voltage. Some rack-mounted power supplies operate in CP/CV mode (Constant Power/Constant Voltage) rather than CC/CV mode. The difference is meaningless in power supplies, and I'll bet the change is delivered power is pretty small in our case, too.

Doug S · « **Reply #7 on:** September 06, 2017, 08:27:22 PM »

Quote from: Keith on September 04, 2017, 07:03:25 AM

I'm doing a lot of data gathering based on voltage controlled partial charging. What I've seen is not unlike what you describe. I've found things like this in the log files that may help to explain it. There are two different SOC modes and when they change so do the numbers. I can say that voltage is not a perfect SOC indicator. I've seen the voltage increase when SOC is decreased, due to temperature I think. And the voltage changes over time after charging while the SOC is not changing.

Keith, this is great stuff, keep it coming! I've always maintained that SoC is more art than science. I believe you're correct that temperature has a big influence on SoC indication -- I've parked my bike at work at 72% in the morning, went outside to ride home after the temperature has risen considerably during the day, and had it report 78% when I turn it on. Other factors that have a fairly large impact are battery life (not much we can do about that) and instantaneous current draw (positive or negative, due to internal cell resistance). So I'm sympathetic with the Zero engineers who have to try to make sense of it all...cell voltage says one thing, the coulomb counter says another, temperature is high or low at the moment, the bike is currently charging or discharging. It's not a trivial matter and there really isn't one true, god-blessed "State of Charge" number.

However, being sympathetic doesn't mean I don't reserve the right to complain that it isn't better than it is!

Doctorbass · « **Reply #8 on:** September 06, 2017, 09:50:17 PM »

I really miss my cycle analyst install on my 2012. it was counting the Ah and Wh and also taking account of the battery real time Interna resistance to correct the calculated SOC.

but for all my long trip the Wh count was the best reference i have evre had!

it cost about 200$ to buy and couple hours install with the shunt.

I also measure seperatly the regen Ah witch is interesting data.

I highly recommand that!

I will install one on my 2017 for sure.. no more doubt abot SOC reading. it also give you real time power, voltage, amp,internal resistance and semi-instantanous wh/km witch is very nice too!

Doc

Doug S · « **Reply #9 on:** September 06, 2017, 10:39:15 PM »

Quote from: Doctorbass on September 06, 2017, 09:50:17 PM

I really miss my cycle analyst install on my 2012...

I finally looked up the CA, and I'm seeing how it could be very valuable and add a lot of peace of mind. But I'm not thrilled with the idea of trying to find a spot to mount and install the low-resistance shunt they're peddling -- it's over five inches long!

Is there some reason we couldn't just find a convenient cable on the negative side of the battery and use that as our "shunt"? http://www.daycounter.com/Calculators/AWG.phtml says 4AWG wire has 0.8 mOhms per meter, so there should be some high-current cable somewhere that's well within range of the CA's calibration value (see attached user manual screenshot). Yes, there will be some temperature variation, etc., but all shunts have that problem. And it might not be exceedingly accurate, given variations on wire diameter, copper purity and the like, but it should be consistent and minor tweaks should be able to get it as accurate as anything can be.

Doctorbass · « **Reply #10 on:** September 06, 2017, 11:00:51 PM »

On my 2015 witch had the 2012 powertrain, i have installed a quality 500A shunt that measure 3 inches c 1.75inches. it was a 75mV version. I located it under the "Y" shape aluminum part under the seat. it worked fine. i also used 1/0 cable with crimped terminal so i had minimal loss. the display was mounted under the OEm zero disolay so i lifted the original display a bit. https://youtu.be/CKMVWBrHhyU?t=35s

Doc

togo · « **Reply #11 on:** September 07, 2017, 09:47:58 AM »

Cycle Analyst?

Doug S · « **Reply #12 on:** September 07, 2017, 07:55:10 PM »

Yes: http://www.ebikes.ca/shop/electric-bicycle-parts/cycle-analysts.html

There's a revision 3 available, but it doesn't look like the new features would apply to us. I'd stick with the 2.4.

Keith · « **Reply #13 on:** September 08, 2017, 02:50:56 AM »

More data, more confusion. I have many different charging cases with my FX packs and two Quiq chargers. Here are four runs, done with great precision and care that don't behave very nicely. I guess I'll do them all a second time just for fun. The 8 amp curves are very different, unexpected but it was a lot cooler today. 8 amp is one pack with one charger and two packs with two chargers. I'll look at logs too. My purpose is to "calibrate" my arduino partial charge controller so I know how to use it in various configurations. It's getting close but there is a lot of noise in the data, apparently due to the SoC readout inconsistency versus voltage.

togo · « **Reply #14 on:** September 12, 2017, 12:28:02 AM »

SoC inaccuracy seems especially bad when you don't charge to 100%

ElectricMotorcycleForum.com

News:

Author Topic: Inaccurate SoC indication? (Read 1724 times)

Doug S

Inaccurate SoC indication?

Keith

Re: Inaccurate SoC indication?

BrianTRice@gmail.com

Re: Inaccurate SoC indication?

gyrocyclist

Re: Inaccurate SoC indication?

togo

Re: Inaccurate SoC indication?

Keith

Re: Inaccurate SoC indication?

Doug S

Re: Inaccurate SoC indication?

Doug S

Re: Inaccurate SoC indication?

Doctorbass

Re: Inaccurate SoC indication?

Doug S

Re: Inaccurate SoC indication?

Doctorbass

Re: Inaccurate SoC indication?

togo

Re: Inaccurate SoC indication?

Doug S

Re: Inaccurate SoC indication?

Keith

Re: Inaccurate SoC indication?

togo

Re: Inaccurate SoC indication?