Discuss data from the LEAF Battery app, and Comparisons

[TomT]

I dunno... I'm up around 55 and I lost mine a long time ago.

Wow.. so 54Ah seems to be the trigger point for losing the 12th bar? Do you have access to the health % in the group 1 data? It would be interesting to see if Ah or health is a better indicator of battery degradation..

 

[DaveEV]

Seems to me that the 'Gids are high for being at VLBW, no?

Hold on a second, you've got 14 percent SOC and the app says that you have almost 3 kwh left. Seems to me that's high for VLBW. If you ran it down to Turtle at 4.3 mi/kWh, you'd have about 11 more miles of range. It seems to me that your pack is saving quite some reserve below VLBW.

Any of our battery gurus agree?

Yes, I noticed this on my car a while back - when I checked it it hit LBW just under 23% and VLBW around 13%. I suspect that the P3227 reflash moves these thresholds lower - will be interesting to see how much. LEAFfan posted percentages on his '13 - don't recall what they were but they were more in line with what you might expect (GIDs).

 

[GregH]

Seems to me that the 'Gids are high for being at VLBW, no?

Simulated Gids in the app appear to run slightly high compared to real Gids. At least when I compare the app to data coming out of my original Gidometer.

Interesting.. I think the simulated Gids are similar to Ah*SOC on the LeafDD... I believe Jim multiplies by a voltage constant to get simulated kwh then another constant to get simulated Gids... So they would both track similarly. If SOC is analogous to an Ah counter this would make sense. At a very low SOC you might expect to have more Ah but less kwh because at that point your working voltage is much lower.. So perhaps good old Gids are still the best energy counter because it tries hard to estimate true kWh remaining. YMMV.

 

[surfingslovak]

Seems to me that the 'Gids are high for being at VLBW, no?

Simulated Gids in the app appear to run slightly high compared to real Gids. At least when I compare the app to data coming out of my original Gidometer.

Interesting.. I think the simulated Gids are similar to Ah*SOC on the LeafDD... I believe Jim multiplies by a voltage constant to get simulated kwh then another constant to get simulated Gids...

Yes, Jim was using SOC * Ah * constant = Battery Wh, which was later changed to SOC * AHr * Nominal pack voltage = Battery Wh per discussion in the other thread.

So they would both track similarly. If SOC is analogous to an Ah counter this would make sense. At a very low SOC you might expect to have more Ah but less kwh because at that point your working voltage is much lower.. So perhaps good old Gids are still the best energy counter because it tries hard to estimate true kWh remaining. YMMV.

I'm a bit baffled at the discrepancy between Gids and 'Gids. The only explanation I can offer is that the SOC we are reading from the CAN bus is not the real battery SOC, but some other sanitized value. Per Cliff's plothttp://bit.ly/cliffsoctogidshttp://bit.ly/cliffsoctogids, the SOC value appears to be another expression of stored energy, perhaps something that was meant to be displayed on a gauge. If true, this would explain the discrepancy. A more accurate approximation for Gids could be the following formula: Raw Gids = "SOC" x 3.1 -17. Although Jim never got around to implementing it, I think it's worth a shot.

 

[nogajim]

Does anyone else have cell pairs that are always weak sisters? Number 21 is always lowest or next to lowest for me. Anyone else got one like this? I wonder if certain locations in the pack would tend to be low, or if this pair is just not keeping up with the rest. Spread is typically in the low 20s on mV.

 

[TomT]

Mine seem to move around so I'm not sure what to think...

Does anyone else have cell pairs that are always weak sisters? Number 21 is always lowest or next to lowest for me. Anyone else got one like this? I wonder if certain locations in the pack would tend to be low, or if this pair is just not keeping up with the rest. Spread is typically in the low 20s on mV.

 

[TickTock]

Does anyone else have cell pairs that are always weak sisters? Number 21 is always lowest or next to lowest for me. Anyone else got one like this? I wonder if certain locations in the pack would tend to be low, or if this pair is just not keeping up with the rest. Spread is typically in the low 20s on mV.

My lowest is #32

 

[Stanton]

Does anyone else have cell pairs that are always weak sisters? Number 21 is always lowest or next to lowest for me. Anyone else got one like this? I wonder if certain locations in the pack would tend to be low, or if this pair is just not keeping up with the rest. Spread is typically in the low 20s on mV.

My lowest jumps around a lot, but my highest is often (not always) #1.

 

[JeremyW]

Note that cells 1-48 are in the large "block" in the back. I'd suspect some in the middle of the block would be the weakest, or possibly cell #1 as it is closest to the shunts (extra heat).

It's getting harder and harder to get five TB's (or under 25c) in the morning here, summer is here! Very frustrating seeing ambient at 18c, and the pack be at 28c because I charged during the night. And I park outside! :?

 

[GregH]

I'm a bit baffled at the discrepancy between Gids and 'Gids. The only explanation I can offer is that the SOC we are reading from the CAN bus is not the real battery SOC, but some other sanitized value. Per Cliff's plothttp://bit.ly/cliffsoctogidshttp://bit.ly/cliffsoctogids, the SOC value appears to be another expression of stored energy, perhaps something that was meant to be displayed on a gauge. If true, this would explain the discrepancy. A more accurate approximation for Gids could be the following formula: Raw Gids = "SOC" x 3.1 -17. Although Jim never got around to implementing it, I think it's worth a shot.

Ever since finding the precision SOC in the Group 1 data (and considering the Ah*SOC energy display option on LeafDD) I've wondered what exactly it was. How do you define State of Charge anyway? If it's an Ah counter that's really easy.. Ah in = Ah out and as long as you've got a decent current sensor you're good. Btw, we don't.
But true remaining energy (ie kWh) is a bit more complicated. While you're in the higher SOC range and resting battery voltage is 370-390V you can more or less use a constant for the voltage multiplier to get kWh from Ah. But when you get to lower SOCs and the pack voltage starts dropping like a rock, those remaining Amp hours MEAN a lot less actual energy because the kWh you can get from them at the dropping voltage is a lot less. To truly estimate remaining kWh you'd have to integrate over an assumed voltage curve. And even then the remaining energy depends on how you use it! But even if you assume a more or less constant power draw, the true remaining kWh would be the remaining Ah * (the voltage curve expressing all future voltages as the pack discharges). Even if you get that right, if the driver decides he's going to drive really hard then HE depresses the voltage further reducing his remaining kWh. If he decides to drive really mellow then the voltage stays a little higher and he gets more kWh per remaining Ah. It could be that the SOC is just an Ah counter and the BMS internal calcs for Gids are trying to estimate future pack voltages to estimate as true an estimated kWh remaining as possible. ??

<edit> looks like simulated Gids = Ah CAP * SOC * 4.5... Can anyone confirm?

 

[GregH]

Does anyone else have cell pairs that are always weak sisters? Number 21 is always lowest or next to lowest for me. Anyone else got one like this? I wonder if certain locations in the pack would tend to be low, or if this pair is just not keeping up with the rest. Spread is typically in the low 20s on mV.

My lowest jumps around a lot, but my highest is often (not always) #1.

At lower SOCs I almost always find #1 the lowest voltage cell.. I should pay closer attention at higher SOCs as well.. if #1 is also the highest at high SOCs then that's absolutely my lowest capacity cell! And I'm not at all suspicious that it's right next to the BMS which must generate a good deal of heat when shunting power from all the other cells during cell balancing. Nope, not at all.

 

[surfingslovak]

Ever since finding the precision SOC in the Group 1 data (and considering the Ah*SOC energy display option on LeafDD) I've wondered what exactly it was?

Thanks for responding, Greg. I will give your post some thought. This has been a really busy week, but LEAF hacking and MNL are always a welcome distraction. That said, I'm under the impression that the SOC value you described above is another expression of stored energy determined via coulomb counting. It just appears to be formatted differently. Kadota-san's slides at the Google meeting in December 2011 meeting indicated that Nissan uses coulomb counting as primary method, but adjusts for voltage and temperature as well. We don't know if the SOC value we have can be simply multiplied by nominal pack voltage and the Ah value to get estimated battery kWh. Based on what Cliff noticed, it looks like we need an adjustment to the multiplier and a linear offset. I cold be wrong, but that tells me that the SOC value is sanitized or formatted in some way. I would expect the simple formula we discussed above to work with raw SOC. Anyway, let me think about this for a bit. Look forward to reading anything else you might want to share on the topic.

 

[GregH]

Ever since finding the precision SOC in the Group 1 data (and considering the Ah*SOC energy display option on LeafDD) I've wondered what exactly it was?

Thanks for responding, Greg. I will give your post some thought. This has been a really busy week, but LEAF hacking and MNL are always a welcome distraction. That said, I'm under the impression that the SOC value you described above is another expression of stored energy determined via coulomb counting. It just appears to be formatted differently. Kadota-san's slides at the Google meeting in December 2011 meeting indicated that Nissan uses coulomb counting as primary method, but adjusts for voltage and temperature as well. We don't know if the SOC value we have can be simply multiplied by nominal pack voltage and the Ah value to get estimated battery kWh. Based on what Cliff noticed, it looks like we need an adjustment to the multiplier and a linear offset. I cold be wrong, but that tells me that the SOC value is sanitized or formatted in some way. I would expect the simple formula we discussed above to work with raw SOC. Anyway, let me think about this for a bit. Look forward to reading anything else you might want to share on the topic.

Typically electric vehicles use Ah (coulomb) counting to track SOC by integrating current measured from a shunt or sensitive hall effect sensor. Unfortunately the Leaf has neither. With a precision of only 0.5A and a potential error offset of 1A or more it's a bit more challenging. In our the BMS we (EnergyCS/EDrive) did for the Prius we used the cars existing hall effect which was a bit more sensitive (Maybe 0.1A? Prius handled less current than the Leaf) Hall effects usually have an offset error. We tried to compensate for this by checking current when the contactors were open (no current flow) with varying results. The Leaf could do the same but doesn't seem to. My black Leaf has about a 1A offset from my blue Leaf (Joulee3). Joulee3 would show about 1A draw when the car was on and not moving with no AC. But the new car shows zero and sometimes a positive (charge) current! Clearly not correct and it's odd they don't do a similar offset error correction. Maybe they do internally but don't show us. Sometimes I've seen the precision SOC walk UP slowly while the car is sitting still.. not sure if it's a slow correction or if it's Ah counting with my offset error on my current sensor.
Anyway.. It would seem the SOC is an Ah (coulomb) counter with probably some internal checks to recalibrate based on voltage occasionally. I've seen this happen when I've accumulated enough error and get to a low SOC.. I can see the low pack voltage and I also see the Gids quickly drop off (about 1 Gid every 2 seconds when sitting still) trying to correct the Gids (and probably SOC.. didn't check). My point though was that since the voltage drops off a lot below LBW and especially VLBW that simply multiplying remaining Ah (ie SOC) by a "nominal" voltage will NOT net in accurate kWh remaining. Perhaps the car IS using the right math to better estimate remaining kWh at these lower voltages and thus present a better picture of actual kWh remaining.. ie Gids.

 

[TomT]

My car typically shows an offset of about negative .7 amps...

My black Leaf has about a 1A offset from my blue Leaf (Joulee3). Joulee3 would show about 1A draw when the car was on and not moving with no AC. But the new car shows zero and sometimes a positive (charge) current! Clearly not correct and it's odd they don't do a similar offset error correction.

 

[Boomer23]

kWh used to recharge from Turtle was 25.3.

CAP% was up to 96.16% and AH was up to 63.71.

It looks like the BMS and the pack are getting to know each other better and the car is learning. I'm encouraged.

I went away on vacation for five days and left the 2013 LEAF in the garage with 7 bars charge. Came back and checked the LB App. AH is now up to 65.33 and CAP at 98.61%. About 40% of the cell pair histogram bars were red at the time of today's reading. Lots of balancing happening in the pack, even without having charged or being driven for five days.

I'll be interested to see the True Gids at 100% charge tomorrow morning.

Recall that the first AH reading that I got at the dealer when I picked up the car was 60.89. It's really interesting (and, of course reassuring) to see the pack gaining this much AH over the week and a half that I've had it. The 25.3 kWh measured from the wall to recharge it from Turtle is also reassuring news that I seem to have full capacity.

 

[garygid]

State of CHARGE (SoC) usually measures the amount of charge (amp hours)
put into the battery. It is not a State of Energy (a fuel gauge), because
the voltage of the LEAF's battery changes as CHARGE is added to the battery.

The ENERGY recovered, per amp hour of current, is higher when
the battery is at a higher voltsge, and lower per Ah when the
battery voltage is low.

Some discussions about the various meanings of SOC here:
http://www.mpoweruk.com/soc.htm" onclick="window.open(this.href);return false;

 

[DaveEV]

I wanted to collect some data before I get the 3227 fix applied (scheduled tomorrow), so I drove down to VLBW tonight.

Started with an 80% charge:

LBW: 49.4 mi 4.5 mi/kWh 24.2% SOC
VLBW: 58.3 mi 4.4 mi/kWh 12.4% SOC

Percentages may be off a tiny bit - have to pull over to grab a screenshot and record data safely.

Shortly after LBW:

Right after VLBW:

 

[GregH]

Some discussions about the various meanings of SOC here:
http://www.mpoweruk.com/soc.htm" onclick="window.open(this.href);return false;

"towards the end of the cell's life its actual capacity will be approaching only 80% of its rated capacity"

 

[garygid]

drees,
When at the dealer for the firmwre update, you might
ask them to check for out-of-spec battery pack cells,
as long as it is connected to the Consult 3 Plus machine.

If you do, plan to have the pack down at lower SOC, like
above, where call-pair 24 looks low.

I believe the cell-pair check requires the average Cell Voltage
to be under 3.7 volts, or some such. GregH will remember.

 

[DaveEV]

When at the dealer for the firmwre update, you might
ask them to check for out-of-spec battery pack cells

Too late for that already dropped it off - and besides, aside from the pack being somewhat out of balance (probably due to me charging to 80% 98% of the time and rarely letting the car sit for any amount of time at 100%) the relative SOC of cells remains similar from a full charge to empty.

Kind of interesting to compare last night's VLBW screenshot with one from almost 2 months ago. Last night's on top, 2 months ago below.

Differences (last night vs 2 months ago):
* Max-Min is 20 mV less, pack better balanced (did charge to 100% earlier this week and sat for nearly an hour, then almost charged to 100% again later, SOC was 95% so I assume it was just about done)
* SOC% is 1% lower
* Avg CP Voltage is 0.08V lower
* Max CP Voltage is 0.11V lower
* Min CP Voltage is 0.09V higher
* Pack is about 5F warmer