Page 1 of 4

Traction Battery/Cell Voltages

Posted: Thu Jan 24, 2013 6:17 am
by JoeS
Perhaps one of our friends working at deciphering the CAN Bus can answer this question: what exactly is the North American iMiEV cell (and thus pack x88) voltage vs SOC curve?

Re: Traction Battery/Cell Voltages

Posted: Thu Jan 24, 2013 12:26 pm
by peterdambier
Sorry, cell voltage and SOC are not related. That is why some people, including me, have developed an allergy against battery management systems.

Between 30% and 70% the cell voltage does not change at all. While charging, cell temperature is an indication. It suddenly rises dramatically at about 95%. Same goes for the lower end with sudden voltage drop giving another indication. Between 30% and 70% our battery behaves more like a capacitor than something chemical. There are no losses and no heat. At the ends heat indicates chemistry and destruction.

The art of bottom balancing

Discharge each cell individually until it gives no more energy. That is done discharging through a resistor for a certain time, waiting about 4 hours and looking for idle voltage. Cell voltage will close in at something below 3 volts. Now select the cell with the highest voltage and bring all cells to that same voltage. That means charging and discharging and waiting for maybe one week. With all cells balanced put them together in one column with no taps. Keep them this way for the rest of their lives.

Should you ever discharge this battery completely all cells will break down at the same time with no chance for cell reversal. Cell reversal would be death. Bottom balanced batteries have proven to come alive again after deep discharge. Reversed cells never come back.

When is the battery charged?

Split your battery in two halves and compare voltages. At the end of charge one cell will be the first to be filled to the rim and rise its voltage. Now the two columns will differ in voltage. Stop charging. And dont dare balancing at the top end or you'll lose cells for sure. A typical fiery incident occurs when you discharge a top balanced battery until one cell reverses. You have lost one cell but that is 3 volts only and the BMS does not tell you. You can still drive your car. There is no indication something fatal has happened. Go home and charge your car.

You have got 3 volts missing and those 3 volts will be distributed among the surviving cells overcharging them. The BMS will not notice the battery is full and continue charging - there will be a fire in the early morning. That has been proven time and again with BMS priests prosecuting everybody who dares telling the truth.

It is mostly lithium cobalt that will end in fire our lithium iron phosphate take a lot of torturing to ignite but ignite they will when you overcharge them long enough.

Our turtle hides one bar from us if not more, with reserves at both ends and it does bottom balancing when below 2 bars. No need to worry but we can help her when charging into the final hour unplug before the turtle does. We can easily calculate how long charging will take. The charger will ramp down one hour longer charging very little if anything at all. That is the time to unplug. Unplugged you cannot overcharge - no risk of fire.

Nevertheless I have started playing with arduino and I'll try to hack the batteries saying hello to each of the cells. Any idea how to give each of them a name? I dont even know whether they are male or female. Imagine we could breed lithium iron phosphate cells. Dont know what to feed - electricity of coarse.

Re: Traction Battery/Cell Voltages

Posted: Thu Jan 24, 2013 5:44 pm
by JoeS
Peterdambier, wow... :shock:

Perhaps let me start off by saying where we agree -

Temperature is a critical parameter
Voltage reversal is death for a cell
Regulating only the voltage across an entire pack can result in murdering individual cells within that pack

If I understand you correctly, you are suggesting that, after careful bottom-balancing, a string of cells can be assembled "... put them together in one column with no taps."

Without getting into the merits or mechanics of top- vs. bottom-balancing, that's where we disagree.

I personally would never dream of putting a lithium chemistry pack together without the ability to both monitor and control each individual cell's (or paralleled cells') voltage and temperature and hopefully rate-of-change of these parameters, regulating the amount of current going into not only the pack but having the ability to shunt current away from each cell as needed to control its voltage and temperature, as well as the ability to shut down the input should any parameter misbehave. Monitoring rate-of-change of voltage and temperture could provide some pre-emptive warning and corrective action.

As an example, for charging, my relatively simple Revolectrix PowerLab 8 (model airplane battery charger which I use on my electric scooter and electric outboard LiFePO4 packs) does a wonderful job of individually managing individual cells in an 8s string, never allowing any individual cell to not only not exceed its maximum voltage rating but also limiting its voltage (via current limiting) so all the cells in the string maintain the same voltage as the overall pack is being charged (they all increase in voltage together).

Where we also disagree - with all due respect, while I recognize that, despite the voltage vs. SOC curve being quite flat and varying for both charge and discharge currents as well as temperature, I believe the voltages ARE measurable and can be used as a simple first-order SOC indicator (which is all I'm trying to do here with the iMiEV). Are you suggesting that coulomb-counting is the only way of determining SOC? If so, what do you use as a reference point?

Finally, based upon the iMiEV's charger power draw, when fully charging the pack the last hour or so is a decaying curve which leads me to believe that they are indeed top-balancing the pack, the blue curve in this snapshot (the red voltage curve shown is only one leg of the actual 240vac input):


Mitsubishi's letter specifically asked us to discharge down to two bars and leave the charger alone to fully charge up the pack until it automatically shuts off - and do that periodically. I believe they WANT us to top-balance the pack periodically.

Thus, back to my original question: anyone have any specific voltage vs. SOC curve for the cells used in the North American iMiEV?

As a point of reference, if I read this correctly, 100%SOC is 3.96v/cell (348.48v pack):

I really don't want to get into a debate, but just wish to know what would be a reasonable and safe voltage to program a regulated dc-dc converter output to inject some current into our battery pack at approximately our iMiEV's 50%SOC point. :geek:

Re: Traction Battery/Cell Voltages

Posted: Thu Jan 24, 2013 5:58 pm
by Don
peterdambier wrote:Sorry, cell voltage and SOC are not related.
That's not exactly true - At either end of the scale (nearly fully charged or nearly fully discharged) cell voltages can give a pretty accurate idea of the cell's SOC. We do need to combine cell voltage with coulomb counting to get a more accurate measurement for the mid-range though


Re: Traction Battery/Cell Voltages

Posted: Thu Jan 24, 2013 6:17 pm
by JoeS
Don, thank you for the link. I have Davide's book, but keep forgetting the excellent references he has on his website.

Re: Traction Battery/Cell Voltages

Posted: Fri Jan 25, 2013 9:16 am
by peterdambier
Coulomb counting, calibrated at the bottom end comes very close but must be calibrated time and again maybe once every season.

At the ends we can see when a cell is fully charged or discharged. Cell temperature is an indication too. With monitoring cell temparature accidents like this one

should not have happened but they are using lithium cobalt chemistry in the first place.

Battery management is kind of a religion but we do have a chance to canbus into our batteries and look at every cell. And dont worry, I have seen more ICE cars burning than electric cars. Dont mention the war about A/C refrigerants:

Re: Traction Battery/Cell Voltages

Posted: Fri Jan 25, 2013 10:49 am
by JoeS
peterdambier wrote:At the ends we can see when a cell is fully charged or discharged.
We agree, by measuring cell voltage. With my own batteries (be they lead-acid, LiFePO4, or iMiEV's), I simply never venture into the discharged territory, and carefully control the upper end.

peterdambier wrote:Cell temperature is an indication too. With monitoring cell temparature accidents like this one should not have happened but they are using lithium cobalt chemistry in the first place.
Overcharging has evidently been ruled out
(Ref: ).
We are all following this one with interest, as I have to believe the BMS on these batteries is extremely robust. I'm looking forward to their solving this puzzle, but am fearful that the evidence may have been sufficiently damaged to preclude a root-cause determination. Talk about forensic engineering with the whole world looking over your shoulder!

peterdambier wrote:Battery management is kind of a religion
Yep! Having spent the first seven years of the last decade cruising on a self-contained solar-recharged sailboat, I didn't take battery management lightly and never had any problems keeping all the onboard systems happy.

I view temperature and rate of change of temperature monitoring and controls responsive to temperature as backup safety devices for Lithiums, whereas if I recall with NiMh chemistry (at least for small-format batteries) they are primary chargepoint indicators. One of the beauties of Lithium-chemistry batteries is that simple voltage management keeps them happy, making charge control even simpler than lead-acid. What I'm trying to say is that the thermal mass of our large-format Lithiums is such that the batteries may be overcharged for some time before the battery gets warm enough to indicate a problem. With good cell-level control the only thing we then should need to worry about is manufacturing defects or damage. The 787 battery investigation report should be fascinating.

Sorry for both of us getting off-topic. I'm still looking for an iMiEV cell voltage around 50% SOC. :geek: I guess the other alternative is to stick a voltmeter onto the traction pack and see what it says at around 8 bars. :roll: I'd much rather use the CAN bus data...

Re: Traction Battery/Cell Voltages

Posted: Sat Jan 26, 2013 8:02 am
by garygid
Some of your battery chemistry "facts" seem to be pulled from
some very different battery chemistries.

Measuring the individual cell voltages is a "must-do" to WELL
regulate charging, or discharging, of almost any Lithium
cell chemistry.

Also, as you charge a capacitor, the voltage goes up linearly.

There is always a relationship between cell voltage and
SOC of the cell, even if the "curve" is more-nearly-flat
with some chemistries. With Lithium Iron Phosphate,
the curve is usually not flat, giving substantial warning
at both ends of the SOC curve.

Many simple pack chargers try to regulate their charging
based on pack voltage, making the ASSUMPTION that the
cells in the pack are nearly identical, and well balanced.

Re: Traction Battery/Cell Voltages

Posted: Sat Jan 26, 2013 8:19 am
by garygid
Just "listening" to the iMiEV's CAN bus, one gets all the cell
voltages, twice a second as I recall.

So, with the CAN bus data, it is fairly easy to tell when
it is safe to charge a non-moving car.

However, the comparatively large cell-voltage swings
of typical Lithium Iron Phosphate cells while driving
would likely make it more difficult.

What are you trying to do, add a "Range Extender"
generator? Handling the gas fumes, and exhaust,
are the most difficult part.

How much power/current are you trying to "inject"?

Re: Traction Battery/Cell Voltages

Posted: Sat Jan 26, 2013 9:40 am
by JoeS
gary, I'm looking at the possibility of a range-extender 48v LiFePO4 stand-alone pack (with its own separate charger and BMS) using a dc-dc converter to feed the iMiEV traction pack, similar to what Enginer had been doing. I would never dream of contaminating my emissions-free EV with an onboard ICE generator! :evil:

I recognize the voltage variability as a function of load or regen, and I think my question has been answered here:
Basically, at almost no load (only 1.1A), jjlink measured a US-spec battery pack at 339vat 34.5% SOC. Thus, I'm looking for a dc-dc converter to go from a nominal 48vdc to something around 340v (adjustable), this upper voltage being very conservative to ensure no possibility of intruding on the iMiEV's BMS, with diode isolation as well to ensure the battery doesn't zap the output stage of the dc-dc. Not looking for much current (1A-10A?), which I'm sure is a function of $$$. Haven't found any dc-dc converters in this range yet.