Thursday, March 6, 2008

2000 miles and counting

I broke 2000 miles with the 914 EV this week. The car has been running smoothly with no complaints. One of the "features" of the 914 is that the turn signals come on (but don't flash) if you leave the turn lever in the "left" or "right" position, even when the ignition is turned off. This presents the dilemma of running down the 12V battery while the main pack is charging. If the 12V battery is dead, you need a small 12V battery (9V would probably work) to activate the DC-DC converter relay to bootstrap the 12V charging process.

The PakTrakr with the series 100 ohm noise-reduction resistor has eliminated all the spurious error alerts. I took a serial log of all the batteries during my commute (graph below) and found battery #11 (the second battery in the front compartment) to be drooping slightly more than the others. I noticed that it tended to have more acid around the battery caps on top of the battery and it's fluid level was lower. The accuracy of the PakTrakr is only 0.1V so the voltage droops look like sharp steps on the graph.


UPDATE: Roger asked me to explain the graph above in a little more detail. The PakTrakr logs voltage for each battery and then dumps it out a serial port in .csv (comma separated) format for MS Excel. I captured the above graph from the chart that MS Excel generated. The batteries are 8 volts apiece and the serial port dumps tenths of a volt (scale on the left edge). This graph displays voltage for all 18 batteries (each is a "series) in the system. As I drove to work with acceleration and regeneration, the voltage on the batteries varied. The batteries with reduced health or capacity will tend to droop lower during heavy acceleration and spike higher during regenerative braking due to higher internal resistance or lower energy capacity. As I mentioned earlier, I found that series #11 on the chart was consistently at the lowest value during heavy acceleration periods, so I suspect that battery will have some issues sooner than the others. I topped off its water to increase its storage capacity a bit. You'll also notice that the "steady-state" voltage at the end of the chart is lower than at the beginning, showing that the pack has gone through some discharge during the 15-mile commute.

I'm getting more involved with the OEVA club. If I don't get elected co-chair next Thursday, I'm going to start an open-source EV conversion kit sub-group within the OEVA. I even have a lawyer friend who is willing to assist with some of the open-source gotchas involved in such a project. Researching my motivations and how much I'm willing to throw myself into this will definitely be important.

Have a great weekend everyone.

11 comments:

Roger Daisley said...

Tim: If you have the time, would you explain what we're seeing on the chart? How was the test hooked up?

TimK said...

Hi Roger,

I've added an update to describe the chart a bit better. The PakTrakr has a wire tied to each node within the series battery pack to monitor voltage.

Roger Daisley said...

Thanks, Tim.

Regarding the open source idea: I think it’s a great idea! I assume what you are talking about is a kind of "general store" for EV conversions, where anyone can supply parts and/or kits … a central clearing house. (I might even be interested in the VW side of it.) One feature I'd love to see is an "eBay style" feedback system, so you could determine the reliability of the particular vendor and attempt to avoid the "EA effect."

pjorg said...

Tim,

You might want to consider doing a separate equalization charge on the one battery with the lower voltage.

Paul

TimK said...

Paul, thanks for the suggestion. I measured the lower voltage at the end of discharge. At the end of equalization charge, all batteries sit at the same voltage (within 20mV), so I'm guessing the lower battery simply has slightly less capacity. If the battery has slightly lower capacity, then I shouldn't charge it up more than the others because it will reach maximum charge and boil more than the other batteries. Ideally, I'd like to match its capacity to the others.

TimK said...

Roger: Actually I had something slightly different in mind. I wanted to create a kit like the voltsrabbit, voltsporsche or S-10 kit, but have all the designs available for free.
I would pick a practical but still cool car to make the kit for, possibly a Saturn but more likely a Honda Civic. This would primarily contain circuit diagrams, detailed designs for the battery boxes and explicit step-by-step instructions like the ones from ElectroAuto. This way people could work with their local welder and plastic supplier to make the custom parts for their vehicle and follow the instructions for installation.

I think there are already man EV resources out there (like EVAlbum) that contain places to buy and trade parts. This would be a set of easy instructions for people who had some mechanical ability but didn't know where to start.

As I mentioned this before, this has already been done for the 914, the VW Rabbit and the Chevy S-10; however, I don't see any "easy" instructions for a more modern practical vehicle that is widely available for conversions.

pjorg said...

Tim,

That may be but you may also want to consider that each battery is made up of four cells. It is possible to measure each cell and if necessary equalize each cell separately.

Paul

pjorg said...

Tim,

Here is a link about Lead Acid battery performance that you may be interested in.

http://www.nrel.gov/vehiclesandfuels/energystorage/pdfs/apesaran.pdf

Tim said...

Tim, you might consider building a little 9-12V pack out of AAA batteries with a plug on the end, and having a matching plug underneath the dash. Use alkalines so they won't self-discharge, and throw the module in your toolbox. That way, if you ever have to "jump" the 12V accessory battery to get the controller and DC/DC started, you'll have the backup on hand.

ianaudio said...

Tim, I've just been checking the voltage readings from my PakTrakr for the 24 NiMh batteries. I had noticed that at least one channel was reading suspiciously lower than I expected and I was concerned that one the indicated battery was in trouble. Turns out that the PakTrakr is not measuring the battery voltage very accurately. With my truck sitting in the garage and no load on the batteries, I took a snapshot of the serial stream into my Palm Pilot, then measured the voltage of each battery. I've found that the one channel in question is about 0.4v lower than my Fluke meter measures. In fact, I'm seeing as much as +/- 0.3 volt difference in PakTrakr vs Fluke on most batteries. I'm thinking that the resistors used for voltage scaling in the remotes may not be of tight enough tolerance to actually give the +/- 0.1v accuracy that you refer to. I'm thinking now that I'll need to create a voltage correction table for the program I'm going to write to display the PakTrakr voltages.

TimK said...

Hi Ian,

My own measurements concur with your own. The PakTrakr modules really aren't that accurate. I did a really low reading on one, but that was due to a corroded connections between batteries. When all batteries are balanced, I get readings that are +/- .2 volts off the Fluke meter readings. For now, it's great at pointing out bad connections and gross problems, but I probably won't rely on the accuracy. Cheers, Tim