Share →


Ed Clausen provides a seasonally appropriate description of a somewhat advanced application of postive temperature coefficient heating to his BMW 320iS electric car. The 320iS is actually a somewhat rare and very interesting BMW and may be the perfect electric vehicle car to start with as a donor. It was the precursor to the e30 series that has been extremely popular for many years now. We’re starting to get a little excited as this is a four door all weather car in a cold weather region of the country. It features a nice manual transmission that is mated to one of our Siemens 1PV5135 motors, an Azure Dynamoics DMOC645 inverter, and our Generalized Electric Vehicle Control Unit.

Ed drives a Leaf and found it sufficiently engaging that he very nearly gave up the idea of doing a conversion. I’m kind of confident that when he completes this BMW, he will see things rather differently, having a longer range from his 400v 180Ah pack of CA series cells. As the 180’s almost always come in at 200Ah, we’re looking at an 80kWh pack – larger than the Escalades. He is a bit critical of the Leaf’s short range. If he remains on the good side of 4000 lbs, we think he’s going to be very surprised when he drives this thing 200 miles on a single charge the first time. The battery weight will actually IMPROVE the BMW ride, and the Siemens will drive it very pleasantly. His dreams of competitive drag racing may take a bit of a blow with this power plant in a car that size, but with the manual transmission, I think he’ll find it pleasant. And with his 1500 watt PTC heater, apparently quite warm. His location in Methuen Massachusetts probably doesn’t get a great call for air conditioning actually.

My own project this week stems from an infuriating change in the CALB specification sheets. In fact, as I review much of the information online I’m finding that the available info on LiFePo4 cells is actually losing ground, with myth and conjecture creeping into all but the Scientific journals and papers, which are increasingly becoming pay per view.

The heart of the issue is that they have reduced the maximum charge rate to 1C for not only the CAM series cells, but oh so quietly for the CA series as well, which HAD been rated at 3C.

Something must be afoot and so I spent some timing talking with our counterparts at CALB on how we got there. They noted two problems, temperature and cycle life. I asked for the data on both several times and it gradually came to light that they don’t have any. But they really like this American custom of just typing yourself smart. It eliminates a lot of testing expense.

And so they noted that my test of 3C in 40Ah cells was not persuasive because it is the same materials as the 72Ah or even 180Ah cells, but the actual current rate goes up and these higher current rates cause an increase in temperature. When I inquired as to what level of temperature, they became a bit vague. And as it turns out, they haven’t actually tested anything to make that determination.

The cycle life position is much more assured though. I pestered them repeatedly for the “test” data that shows decreased cycle life in high current charge cycles. They finally admitted they had no “test data” in the cycle sense, but had firm REPORTS from one of their customers in China who used the CAM72Fi cells in a BUS. After about 800 cycles in a bus, regular all day daily service, they were losing cells. The fast charge rate was a little vague. They were doing a couple of hundred amps of REGEN charging – not at a charge facility at all.

I have six or seven hundred problems with all of this. If they are doing a couple hundred amps of regen on a bus on CAM72FI cells, that implies to me that they are DISCHARGING them routinely at the 700 or 750 amps level during accelerations.

We have demonstrated that these cells can produce 10-12C of discharge current for periods up to 30 seconds. This is of interest if we want to take advantage of the maximum power capabilities of any particular vehicle. That is not to say we advocate USING the maximum power capability of your vehicle in any sort of continuous sense. It means if you needed, for some reason, maximum acceleration for 15 seconds, these batteries would provide the power. You would not be power “limited” by the batteries.

Now I’m picturing a single string of CAM72 cells trying to drive a BUS. Where ALL accelerations are more like 20 and 30 seconds and at max current.

I use 400Ah cells in the Cadillac Escalade. Indeed, I have the motors and controllers to do 2000 amps at 170 volts in this beast. And the 400Ah cells will easily do it. But generally in driving it it is an EVENT to do more than 750 amps on any particular acceleration. I CAN do 2000 amps. I HAVE done 2000 amps. But that’s just not how we roll almost all of the time.

If you are constantly tapping your cells for the max power they can put out, it will certainly decrease battery life. And it is almost certainly not because of excessive regenerative braking.

So do size your battery pack to your application. If you are doing 600-700 amps dozens of times per drive, I would want to see you in a 180Ah size battery cell. It will simply last longer.

So I was feeling a little bit like I’d been thrown under the bus on the maximum charge rate. So this week we ran a basic test to see what the temperatures would be on a 3C charge of a 72Ah cell – a 216 ampere rate. In absolute terms, this is of course much higher than the 120A we tested the CA40FI cell at.


Take a look at the resulting charge curve. In our normal charge process, we might have a fully discharged pack to 2.75 volts per cell. When we apply current to it, it immediately rises climbing quickly to 3.00v and perhaps a bit beyond. But by 3.05 or 3.10 volts, it has passed the “knee of the curve” and enters a much flatter area of charge where large amounts of charge current result in very tiny increases in voltage. This continues to about 3.45 volts where it again turns up and starts to increase in voltage at an accelerating rate. By 3.65 volts it has gone pretty nearly verticle.

If you examine the graph of the 40Ah cell at 3C or 120 amperes, the CURVE is identical. But it is DISPLACED in voltage. Now we rapidly increase to about 3.40 volts and THERE we find the first charge knee. Then a long flat period to 3.65 volts where it is not only not straight up, but it is JUST starting to bend upwards.

This is ENTIRELY consistent. At a higher current level than normal, our voltage will be higher than normal as well.

We talk about charge voltages a lot. I have repeated over and over and over that they are not real and in fact do not matter at all in and of themselves. In the CONTEXT of a procedure, given a specific set of circumstances, we use voltage because it is EASY to measure. But it is NOT state of charge and you might think of it as more of a PROXY for state of charge – all other things being equal.

Which of course in this case they are not. Our procedure assumes a standard charge rate (1/3C) and a standard temperature (25C). If you change that, the meaning of the voltage goes out the window.

The manufacturer recommends 3.65 volts. Now understand that historically, these same sorts of LiFePo4 cells recommended for a long time 4.2v per cell. At some point, after we determined that the KNEE was pretty much done at 3.65v, this was dropped to 4.0v per cell. Still later, they came to 3.65v per cell.

But even that is as an example. And it again assumes 1/3C current rate. At 126 volts, we were lucky to have 30 amps from our little chargers into 180Ah cells. That’s half of 1/3c. More like 1/6th C. And at 1/6th C, you are going to result in a higher state of charge than you will at 1/3C – charging to the same proxy voltage of 3.65.

And so you find Jack the Wide One recommending 3.55v per cell as the target voltage. Extra safety? Well actually a bit of extra reality. We’ll be at about the same point charging to 3.55v at 1/6C as we would be at 3.65v and 1/3c. It is easiest to explain this by saying that we are already at the steep end of the curve and there is no real range to be had here. The CURVE matters. The voltage really never did.


Here’s another example. This is the DISCHARGE rate comparing the discharge curve of a CA180Ah cell at 88F and at 0F. In this case, the curves are NOT identical, but more importantly, the voltage is again displaced. Normally, we would consider 2.80v as being very nearly 100% discharged. But we can see from the chart that we are pretty much there at the BEGINNING of discharge if we are at a temperature of 0 degrees.

All of this long wind is to introduce you to the result graph of our 3C testing of CAM72Fi cells. These are the SAME chemistry as the CA series, but manufactured to optimize for size and weight. And while 3C at 40Ah is 120 amps, for a 72Ah cell this is 216 amperes. And so it displaces the VOLTAGE further. But again it does NOTHING to the charge curve itself.

In this event, we begin charging at a known SOC based on a very measured 2.76volts per cell. In fact all three carefully balanced to 2.76v. By ONE MINUTE of charging, putting in about 3.5 Ah of energy, we are already above 3.60 volts. We see the usual notch as the cell warms up a bit, and then we do in fact turn the knee and see a long flat period of around 3.55v per cell. We should be done charging right?

Not quite. We cut it off at about 3.7v. But we had put in about 69.1 Ah at that point. And as you can see from the curve, we had JUST started to bend the upper knee of the charge curve. Indeed, after charging, we disconnected the cells and within just a few minutes they were done to 3.33v. We normally looke for 3.37 or 3.38 static in a cell we consider “fully charged.”


We actually have a lot of head room to play with here. Recall that we can charge these batteries to 4.2v.

We have recently been discussing CHAdeMO fast charging for the DIY custom conversion cars and we are bent on doing it. But it is a different world at 125 amperes than it is at 11 amperes, and this test illustrates why.

We recently had a forum discussion where a couple of live Leaf drivers in Europe were reporting that some CHAdeMO charge stations would give them 80% charge, while others seemed to cut off early. And they were certain the EVSE was doing it. No matter how I explained that the charge process was entirely a function of the software in the car, they “proved” me wrong by noting different charge times at different charge stations despite pushing the button a lot. When I inquired as to voltages and CAN message data, they actually accused me of demanding that they do my work for me. I was actually hoping for some data rather than grandmother observations of the weather and what it means. I don’t like reading messages from God in cloud formations.

I think if they check again, they will find that the short charging stations are putting out a higher level of current (perhaps 50kw) than the stations that appear to give 80% charge (at 25kw). I’m just guessing.

The issue is that the Nissan Leaf uses LiMN2O4 cells. They are nominally 3.7v charged to 4.2volts – identical to our Renault Influenza packs. Again, the software in the car advises the EVSE of the target voltage. The EVSE actually DOES advise its MAX current. But if the firmware in the car does not take that as an input and calculate the charge voltage based on the charge current, you will get a different termination at high currents than you will at more modest ones.

And I get the idea that the top balancing adherents are devoted to this concept based on LiMNO and LiCoO2 cells, not LiFePO4. They appear very afraid of going much over the 4.2v. I really just don’t have the direct experience with cells of that chemistry to advise what happens when you exceed 4.2v, but we are just BARELY turning the knee there at all. Really we are NOT on the knee at that point. So I suspect it won’t hurt them actually. But I don’t know that with any authority.

So in devising our software for CHAdeMO fast charging, we need to map what the charge voltage should be based on the charge current. As we can see from the CAM72FI cell, we needn’t bother at all with a tapering current. Just figure out what the voltage is and cut it off there – quite close to 100% charged in 20 minutes.

Of course teh maximum CHAdeMO current is going to be 120 amperes – not 216. And more likely, the maximum current will be something like 80-100 amperes.

I’m actually considering throwing out voltage as a determinant of SOC for fast charging. We could go for a deltaV/deltaT scheme where we monitor how quickly it changes in voltage. For example, when we first go on charge, IF the voltage is LESS than some absolute, we note the change in Voltage compared to the change in time. If the voltage quickly goes over the absolute, we know someone has just plugged in a fully charged vehicle and we terminate.

If it does not exceed the absolute voltage, we simply calculate what the deltaV/deltaT IS and save it. This will diminish as we cross the knee and go onto the flat part of the curve. We simply keep timing/calculating it in a running total, and when it again goes back UP to equal the rate of change we first encountered, we’ll know we are fully charged and terminate the charge process. We can further count Ah to confirm, and check for absolute voltages for example. But if we had a bad cell that went through the roof, it would show up on this deltaV/deltaT and we could detect it before the whole area burned to the ground.

As to the CAM72 cells, we reached a maximum temperature of 42C. As the non-aqueous organic solvents in the cells start gassing at about 80C, I’m pretty comfortable here. The normal advise is to keep below 65C in temperature. The manufacturer’s spec sheet says charging is ok between 0 and 45C and we are within. True, we started at a 56F ambient, and we are not packed with 50 other cells in a box. But there is a lot of leeway here. I do NOT believe from this test that temperature is a limiting issue at charge rates up to 3C.

As to cycle life deterioration, I have to say that while I’m unconvinced, we have also proven nothing so far here and it remains an open question. But in any event, given our use patterns here at EVTV, an 800 cycle life would be more than adequate. Indeed, we have several cars I would LOVE to upgrade to take advantage of these new cells as the ones in the car are so serioiusly obsolete. But how can you do that when the car still runs at very nearly 100% of the original range capacity?

Perhaps I can persuade a couple of our more attentive battery testers to take on the challenge of 3C cycle testing.

Jack Rickard