Looking at charging our car from the inside out ...

[anko]

We've seen lots of pictures from charge boxes showing us how charge current is high for a long time and drops off at the end. Also, we've seen pauses for balancing cells. Here is a picture from how the car sees it:

Interesting how SOC makes a little jump during the balancing process .... That jump is 6.5% :shock:

 

[gwatpe]

Looks very similar to a picture I posted of the recharging power from the outside AC logging. Interesting that your PHEV turned the charging OFF for a short period as well. I do notice that sometimes mine does the same thing. My PHEV is doing the turn OFF the recharge for a bit right now. Is back recharging again now. about 5 mins OFF. I was looking at the rate of balancing being an indicator of the battery health. Now have OBD2 data that gives battery health as well from the inside of the car.

I have not seen the steps that Anko shows. The measurement resolution of the PHEV is not that good, typically 8bit resolution, compared to the 10 & 12 bit resolution I use for logging my own data. Note the period my PHEV stopped recharging and then continued. I have data and graphs for every time my PHEV has been recharged at home. Automated systems that work are great. Just need to sort out the PHEV comms, so my logging of it is as good.

 

[Sunder]

Interesting how SOC makes a little jump during the balancing process .... That jump is 6.5% :shock:

Makes sense though right? The BMS will cut the power from the entire pack, when the lowest cell hits LVC. So if you're measuring the capacity, you figure it out from the cell most likely to shut down first. Literally, the chain is as strong as its weakest link.

To be honest thought, 6.5% difference between the lowest cell, and the next lowest cell would indicate to me that a cell is faulty. If it were on my electric bike, I'd be cutting it out and replacing it if the pack was close enough to new that a new cell wouldn't be substantially stronger, or ditching the entire pack if the pack was older. Obviously can't do that on a battery pack someone else built

On that note, does anyone know how many cells we actually have in our pack?

 

[anko]

Interesting line of thinking. Good to be aware of. Will try to check with EvBatMon beta app tonight.

But if one cell was faulty, how would the BSM be able to pull up the capacity during balancing? I mean the faulty cell would still be faulty. Couldn't it just mean that during charging the BSM doesn't do a very good job at keeping the pack balanced? I mean, the balancing phase is there for a reason. And my PHEV is definitely not the only one doing this.

BTW: There are 80 cells.

 

[Neverfuel]

I don't think that looks like a faulty cell, I think it is just normal balancing activity.

Each of the 80 cells has its own Cell Monitoring Unit, Temperature Sensor and Balance Driver.

If you think about it like filling an ice cube tray. First you will just put the tray under the tap (just to get the volume of water in there). Then you will stop filling the almost full holes and individually fill up the less full ones, then go back and try to get the water level to the same height in every one of the holes in the tray.

That's how cell balancing works - first it dumps in a shedfull of charge. It doesn't matter where it goes. Then as the battery approaches full charge the CMU in each cell detects the highest voltage and stops that cell from charging using the balancer driver, this gives the other cells time to catch up. As a new cell reaches a limit above the others, it is also stopped from charging, to bring the level of charge in the other cells up. It continues in this way until all cells are within a tolerance limit set by the PHEV ECU then charging is stopped.

The time gaps are possibly down to allowing the temperatures on each cell time to cool to a specific level (by starting the cooling fan and possibly the air conditioning) so that accurate voltage measurements can be made.

If Anko has a deterioration in one cell over the others, the ECU would tell him that a battery service was required, because it would never reach 100%. Bear in mind that the 100% actual value will decrease over time as the cells degrade through age! but they should all degrade at roughly the same rate due to balancing.

This is why fast chargers only go to 80% (although you can force them to go to 95% by charging a second time! but the second charge is at a much lower ampage and takes longer for the extra 15% than it does for the original fast charge to 80%). 15% extra is a long way in a Tesla, but for us PHEV drivers it's only about 3-4 extra miles so not worth the extra half hour wait.

Interesting to see that Anko's battery charged for longer than Gwatpe's before shutting down the charge and restarting again. This might validate my point about temperature above as Gwatpe's battery will naturally be hotter at this time of year than Anko's, so needs to let the cells cool down earlier in the charging cycle.

 

[anko]

I've seen this shutdown happen as early as within 5 minutes of beginning the charge process. I got home from work with an almost empty battery. I had to go out for some shopping and thought a 15 minutes might be enough to get me back and forth to the shop in EV mode. After 15 minutes, I noticed SOC had hardly increased. Later on I saw the gap, which started almost immediately after me hooking up the car to the grid.

Not knowing the gap was purposely introduced for balancing, later I tried if I could 'close the gap' by unhooking and hooking up again. Turns out you can. But doesn't seem a smart thing to do ....

 

[Sunder]

Maybe "faulty" is too strong a word. From what I understand, the purity of the conductive solvent is the main contributor to the internal resistance of the battery. If the internal resistance of each cell within the battery isn't very close, during the bulk charging Neverfuel described, its a bit like trying to fill the ice cube tray with some of them partially covered. Its not going to be even. In an ideal situation, filling is not random, its even. Uneveness is indicative of a non ideal battery.

Likewise, when the power is being used, the discharge profile is different.

No battery is perfect. But a good battery for my bike requires balancing every 6 months. A cheap battery needs balancing every 3-4 charges when the cell capacities are > 5% off.

6.5% is not by any means cell death, but unless the BMS is programmed to balance on a cycle basis, or a time basis and not an "as needs" basis, 6.5% isnt all that normal either.

 

[Neverfuel]

Maybe "faulty" is too strong a word. From what I understand, the purity of the conductive solvent is the main contributor to the internal resistance of the battery. If the internal resistance of each cell within the battery isn't very close, during the bulk charging Neverfuel described, its a bit like trying to fill the ice cube tray with some of them partially covered. Its not going to be even. In an ideal situation, filling is not random, its even. Uneveness is indicative of a non ideal battery.

Likewise, when the power is being used, the discharge profile is different.

No battery is perfect. But a good battery for my bike requires balancing every 6 months. A cheap battery needs balancing every 3-4 charges when the cell capacities are > 5% off.

6.5% is not by any means cell death, but unless the BMS is programmed to balance on a cycle basis, or a time basis and not an "as needs" basis, 6.5% isnt all that normal either.

Hi Sunder

My point was that the temperature and voltage are the inputs to the calc for battery capacity. During the stopped period, the battery will be cooling down. As each cell is measured independently by its CMU, when the highest temp cell has cooled to within the ECU tolerance zone, there will be a recalculation of the voltage of ALL the cells, and a decision made by the ECU as to whether to switch on charging again or wait until the next hottest cell comes within tolerance.

You mentioned that your bike requires balancing every 6 months. The PHEV balances on EVERY full recharge AND full discharge cycle. 6.5% is not the fault of one cell, especially during a cooling period! as calcs are only done in real time when current is flowing. MIT is simply a recalculation of voltage x temp once the temp has changed to give the PHEV-ECU the proper indication of % SOC, thereby forcing charging shutdown or reduced ampage balancing.

But then it's all guesswork unless you slaved away for 3 years designing the system, so what do I know

 

[Sunder]

Fair enough. I wouldn't have thought that temperature came into it that much with such a low C rate chargebut as you mention, I'm guessing, and you've spent 3 years designing this.

I'm curious as to what the algorithm is that could "find" 6.5% of battery capacity without any additional input from the charger though. Genuine question. There would need to be quite a substantial change in temperature or voltage in the cell to find that much power by rebalancing.

 

[Neverfuel]

Fair enough. I wouldn't have thought that temperature came into it that much with such a low C rate chargebut as you mention, I'm guessing, and you've spent 3 years designing this.

I'm curious as to what the algorithm is that could "find" 6.5% of battery capacity without any additional input from the charger though. Genuine question. There would need to be quite a substantial change in temperature or voltage in the cell to find that much power by rebalancing.

Sorry Sunder, I didn't mean that I had spent 3 years working on the system, I was saying that we are all in "guesswork" territory unless we have design experience from inside the Mitsubishi production team (which I certainly don't). I just voice my opinion based on my experience of related subjects. I think that's why the forum is interesting, because, more often than not, there is a debate on HOW the technology hangs together. I don't know about you but I have never had as much fun talking about a car, as I have driving it (and I have driven some beautiful dream cars in my life - none of which I owned, I would like to add). I wouldn't even spend the amount of time that I spend on the forum pursuing my favourite hobby (which I refuse to mention on the grounds that it may identify me!!!)

 

[gwatpe]

Interesting to see that Anko's battery charged for longer than Gwatpe's before shutting down the charge and restarting again. This might validate my point about temperature above as Gwatpe's battery will naturally be hotter at this time of year than Anko's, so needs to let the cells cool down earlier in the charging cycle.

I was not able to read the time data on the X-axis, so I cannot say how long the balancing phase took, or the first part of the charge took from anko's chart.

For my data, there is no times on the axis, and the portion of a day, that I have shown has no time reference. I have checked my data and the balancing phase typically takes 30min, and the first stage was 90min. I have other records when there was no interuption of charging for a full recharge.

As I mentioned in my post above, I believe that the measurement of the AC charging power is very basic, and the graph is really only showing about 3bits of resolution. The 7 steps that are shown, I would not interpret the steps as actions by the balancing systems. This is just the measurement resolution giving a + / - 1 value and this has been recorded with time.

My PHEV draws about 170W when it has mad a decision to suspend the charging for a bit, and this could be power used to run cooling pumps and such.

Anko's chart, drops to zero, and this may be that the resolution of the measurement gives a zero or that this is power only going to the battery.

I no longer scrutinize the recharging data as I only recharge from my solar systems when I have surplus, and this may not be every day, and I don't recharge unless the battery is showing under 1/2 on the gauge, and seldom from empty.

 

[Sunder]

Sorry Sunder, I didn't mean that I had spent 3 years working on the system, I was saying that we are all in "guesswork" territory unless we have design experience from inside the Mitsubishi production team (which I certainly don't). I just voice my opinion based on my experience of related subjects. I think that's why the forum is interesting, because, more often than not, there is a debate on HOW the technology hangs together. I don't know about you but I have never had as much fun talking about a car, as I have driving it (and I have driven some beautiful dream cars in my life - none of which I owned, I would like to add). I wouldn't even spend the amount of time that I spend on the forum pursuing my favourite hobby (which I refuse to mention on the grounds that it may identify me!!!)

I'm with you there. I used to be very keen on car forums, back in the day before Fast and Furious ruined the scene. That's when we would debate how different methods of grinding manifolds give you different swirl and efficiency at different RPMs... I walked away when someone came onto the car forum asking "I saw a Nissan Pulsar with a silver radiator and thought it looked great, where do I get one from?". We were all puzzled, until she posted a photo of a front mount intercooler...

These days, I spend a fair bit of time on electric bike forums. With people trying to build motorcycle speed electric bikes (Not me, I'm happy with a 50km/h bike that looks indistinguishable from a push bike), and really digging down into the various technologies (We've had electrical engineers build and sell custom field oriented controllers, mechanical engineers machine us improved stators with thinner laminations, even chemical engineers telling us how to better treat batteries to get more out of them), it's been a very fun journey.

This is my first foray back into a car forum after 10+ years out, and I am having a lot of fun finding about our car, and how to get more out of it. If someone asks how to get a silver radiator on our car though, I'm out. (Kidding)

 

[gwatpe]

I normally don't leave the OBD adapter plugged IN when I power OFF my PHEV, but I had been testing a new screen layout on a tablet, and happened to power OFF with it still connected. Torque continued to be connected and was still logging. I reconnected the charging cable, and yes, the current going back to the battery was still being displayed and recorded. Was just a test of the last phase of charging with hopefully same cell balancing going on. Had checked with EvBatMon, and the cells were within 1mV, so don't expect to see much balancing. Watched the battery power drop in about 0.3kW steps, this is around 1A steps into the battery, starting from 1.9kW. The external logging had the current drop smoothly from about 1950W, over a 15 minute period when recharging was terminated at about 500W.

here are comparison charts.

raw plotted data from the PHEV

Data from external logging

PHEV logging was at 0.5sec intervals, while external logging was 30secs.

Note the big step jumps in the top graph. This is NOT any balancing effects, but purely the low resolution of the measurements in the PHEV. We need to be careful what we read into the data we collect.

Will be repeating this experiment for an empty battery, and comparing it with external logging, with some data filtering to hopefully clean the data.

 

[anko]

Let me clarify a thing or two:

First of all, the value for battery charge is not measured in an 8 bit resolution.

Both the readings for amps as the value for total voltage are 16 bit values. So, although not all 16 bits may be used, the resolution is definitely more than 8 bits for both values. But it doesn't mean a whole lot to us. For Volts, there actual value is derived by taking the value read form OBD and dividing it by 10. So, the precision for Volts is one decimal after the decimal point. For Amps, the actual value is derived by subtracting 400 from the value read. So, it is in effect an integer value.

With voltage hovering around 300 and a bit all the time, you can expect that the (dis)charge power (as it is derived from Volts and Amps read from the OBD port) goes up and down with steps of approx. 320 - 330 volts (the 0,3 kW that we see often these days).

We also have to be aware that the fact that the OBD port presents Amps as an integer value does not necessarily mean that the internal measurements are that crude too.

Further, when I started this post I mentioned possible cell balancing. But when I did this, I was not referring to the small 0.3 kW drops offs at the end of the charging process. These were not a surprise to me per the above. I was referring to the huge 3 kW drop off in the middle of the charge process, where charging effectively stops.

Finally, it seems that via the OBD port you can access any active ECU. So, ECUs that are active when the car is switch off can still be accessed via the OBD port. As Gwatpe noticed. This would at least apply to BMU and OBC. But also to EVMiEV (the ECU that controls REMOTE access and what not. But it does for instance not apply to Front and Rear Motor or Generator. Kinda makes sense.

When you start up Torque Pro, it does not immediately start talking to any ECU but it first wants to 'establish a connection to the car' (blue car symbol blinks). In this phase, it tries to talk to the engine ECU (7E0) asking it, amongst others, which PIDs are supported by that ECU. As the engine ECU does not respond, TP believes the adapter is not connected to the car properly and it will not send any other commands. Depending on how things are programmed, I totally expect EvBatMon not to have issues with this phenomena.

 

[gwatpe]

Have a little bit clearer picture of the recharging process from the cars perspective.

The Battery voltage, and the battery power as well as the battery %SOC are shown. The vertical line is the transition from constant current to constant voltage.

This was during the last phase of recharging.

The %SOC rose in a negative exponential fashion, as expected.

The steps in the battery power reflect the low resolution of the measurement due to the small magnitude of the measurement. The resolution is only 0.3kW, so for 2kW, we only get a total of 6 steps. Even if the power measurement has 16bit resolution at a maximum value, at the point we are measuring, only 3bits are seen.

It is likely that the current measuring element is external to the battery, so any individual cell balancing within a pack would not be seen in any current measurements during the constant current or the constant voltage phase of the recharging. If cell balancing was going to happen, it is going to be difficult to see when it happens during the recharging. It is possible that the battery in my PHEV does not need any balancing. If just stopping the current was associated with cell balancing, I would expect to see some evidence, at least every recharge. There was none in this recharge, or the last 5 recharges. I have seen the recharging stop for periods of about 10mins during a recharge, and it is unlikely that this is related to cell balancing. This may be just to allow the PHEV computers to make internal adjustments, or compensations that would possibly affect the calculation of completion of recharging.

The recording of the systems when a rapid recharge is done, may shed some light. Rapid recharging is most likely to show cell differences due to the additional stresses place on the battery cells. Only the EU shipped PHEV has the rapid recharge port. My AUS shipped PHEV is without the necessary equipment, so I cannot test this.

 

[Kim]

Yes, a CHademo charge chart would be interesting also...!