Saturday, September 19, 2009

More Battery Boxes and a Breakout Test Box




Haven't posted for a couple of weeks. We've been busy. We're STILL working on the battery boxes.

We have the toughest one pretty much done. This is essentially a battery "module" that sits on rails where the rear passenger seat was. It slides fore and aft about 18 inches to reveal two 12 cell boxes beneath the seat area where the gasoline tank was.

There is also a 33 cell box aft where the spare tire area was below the cargo floor. By mounting it on rails, we're able to access the lower battery boxes with relative ease.




SInce it is going to be in the passenger compartment with us, there are a number of considerations we don't normally have to deal with.

First, it's fully enclosed with a lid that screws down onto the box.


Second, because it is fully enclosed, and because it is in the passenger compartment quite removed from the passing air stream, some form of cooling was required. As described earlier, we installed a 235 CFM 12volt COMAIR fan on the aft end of the box using a thermostatic switch with a probe located in the center of the pack. In this way, if the internal temperature of the module exceeds about 100F, the fan will automatically kick in and blow air INTO the box. A series of small holes drilled in the bottom of the box will vent beneath it to the battery boxes below and out of the car. The location of these drilled holes will regulate the air flow more toward the center of the box.

I've wrestled quite publicly with the issues of battery management. This is the largest string we've yet built at over 40 kWh and a total of 112 cells. We're waiting on the motor/transmission assembly to be returned from VAC motor sports and much of our front end assembly rather depends on placement around that drive train. So we have had some time. So I decided to build the biggest ball of spaghetti on the planet and wire an access to each terminal of each cell in the entire battery pack.

I had lucked onto some 19-pin Amphenol connectors on eBay. These cost about $35 each but I got a sack of 50 brand new ones for $65. These were all bulkhead mounted females and no male plugs at all. I bought four of the male plugs on Digikey for $35 each.

So we decided to wire amphenol plugs to the battery packs. In this way, we could later add a balancer circuit, monitoring system, or whatever just by mating in with the plugs.



It has been a lot of work. True it will let us test a good bit. But it is entirely unnecessary to operate the car, and it has turned into a huge amount of work. I don't recommend it.

Still, it gives us access to all voltages at the cell level. We used 16 gage wire to do the wiring, and little 16 gage 5/16 loop terminals to attach to the cells. The wires are soldered into the amphenol plugs. I basically build a plug with wires at the bench, and then install it in the pack routing and trimming each wire in turn.

Each amphenol plug will do 18 cells. You need to connect to both the positive and negative terminal of the first cell, and then the negattive terminal of each succeeding cell in series. The cells are strapped in series by small copper bus bars made of several individual copper plates bent into a flex shape with a 5/16 hole in each end. They come with the batteries along with M8 bolts to bolt into the cell terminals.

Along the way, I decided to build a test set with a simple 5 digit voltmeter, and a rotary wafer switch out of a Soviet submarine. This switch is a monster two wafer 35 position switch.




In this way, I can connect the wire from pin A of the Amphenol to SW1-A, and pin B to SW1-B and the voltage across the cell can be read on the voltmeter. If you cycle to the next switch position, you will read pin B to pin C, and so forth through the 18 cells.

As long as we're reading the voltage, it is pretty trivial to provide a little bit of a resistor circuit to bleed off some energy. I used a single pole 3 position toggle switch to switch two 0.5 ohm 25 watt power resistors across the terminals. In this way, I can put either a 1 ohm load on the cell or a 1/2 ohm load on the cell. At 3.3 volts, a 1 ohm load would be about 3.3 amps. But the load is so small, that the resistance through the switch comes into play and I get about 2.5 amps. On high bleed through the 1/2 ohm resistor I get about 5 amps.

As long as we're doing that, it is pretty easy to add an old time analog ampmeter to the circuit so I can see it bleeding current.

I found a small 3 amp 12vdc power supply for about $39.




It was easy enough to add that into the box with a 2 pole single throw (2PST) toggle switch. This connects a 12 volt 3 amp source across the cell which will charge it at about 3 amps. In this way, I can ADD a little energy to a low cell.


The result is a sort of breakout box/test set for the battery modules, with an Amphenol plug that mates the Amphenol bulkhead connectors on the battery packs. By connecting it to these connectors, I can cycle through 18 cells and if one of them is a little low or a little high, I can either charge it a bit or bleed off a little energy.

I know this is terribly low tech from the 1970's, but it gets me in touch with what's actually happening with the cells - at least in the garage. It would be easy enough to add an Arduino microcontroller and display system later of we wamted to battle EMI issues for the rest of our life. We simply connect it using the same connectors.






The voltmeter selected was a little 5 1/2 digit 20vdc unit from LightObject. This unit will show me the cell voltage to the millivolt - 3.266 for example. It's probably a 0.5% accuracy, but that's close enough for government work. And we're more concerned with RELATVE voltage from cell to cell than we are to absolute voltage. If all cells are at 3.266 vdc, or say ranging from 3.264 to 3.267, that is what we are looking for mostly.

The 0.5% on a 20 volt meter would be within 0.1volt. I measured with 0.02 volt using a good Fluke multimeter.



This voltmeter requires a 5vdc supply that is isolated from the voltage it measures. Fortunately, they had a little 12vdc to 5vdc isolated converter the VB1205S for about $10. It will actually work on 9-18v. I supply connected it to the 12vdc 3 amp supply and the meter. If you connect the charger with the CHARGE switch, the voltmeter goes dead, but it comes right back when you finish charging.

That brings up a couple of other items that I almost hate to go into. But I've seen so much discussion about "voltage sag" on these batteries that I guess I had better address it. Although mostly nonsense, there are a couple of issue you can control that are important.

So, let's start with the question of how we can measure voltages through these amphenol plugs, across a terminal strip, through a comically ancient wafer switch from a Soviet Submarine, and get any sort of accurate reading at all? Aren't there little voltage drops across each of those connections?

Well, in answer to the question, YES. There are. If there were any current flowing at all. But there isn't. The input impedance on the voltmeter is actually unknown in this case, because the Chinese manufacturer isn't given to a lot of expense on documentation and so forth. But typically, it would be on the order of 10 Megohms or roughly 10 million ohms. The resistances through all those connections might be on the order of 0.25 ohm. And so if you drop 3 volts across a series resistive circuit with 10 Megohms on one end, and a quarter of an ohm on the other, the voltage drop across the quarter ohm is less than the granularity of the digits on the voltmeter. With essentially NO current draw, we can measure these voltages quite precisely.

How does this relate to voltage sag? Well, if we cut in our load of 0.5 ohms and draw 5 amps or so, you will observe the voltage dropping to about 2.5 volts. This would indicate these batteries aren't much good. They "sag" to 2.5 volts on just 3 amps of current draw.

Actually, not. We are measuring the voltage out at the terminal ends, and indeed we DO drop voltage across the connectors, the switch, etc at that current level. If we had 0.25 ohms and 5 amps of current, we would see about 0.25x5 or 1.25 volts of drop.

And that's rather the point. Battery cell "sag" is very much dependent on where you measure it. The only place that gives you any indication of battery voltage under load is DIRECTLY at the terminals. If you go through anything else with the current, and measure the voltage at the end, you aren't measuring anything useful. We could for example, use our breakout box to measure voltage under load while driving the car. That's because we have essentially ZERO current through the measuring circuit, with potentially 300 amps through the drivetrain.

But that brings up another point. It MIGHT be useful to measure voltage at the controller under load, not to tell if the BATTERIES sag, but to tell if your connections are any good. ANY resistive in the current path at 300 amps is going to give off a lot of heat, and diminish power to the drive train.

Probably more than you wanted to know about voltage measurements. But take everything you read on the web about "VOLTAGE SAG" relating to LiFePo4 batteries not just with a grain of salt, but just basically ignore it. There are entire conversations that have been going on in forums for MONTHS that are so much nonsense I can't begin to sort them out or offer them any useful information. But this is what it has to deal with. Voltage measurements change dramatically under current load, depending on where and what you are measuring. NO LOAD, it doesn't matter. The voltages appear very accurately everywhere. UNDER LOAD with a current draw, and you can only measure accurately at the source.

Jack Rickard
mjrickard@gmail.com