Saturday, February 6, 2010

Analyzing PakTrakr Files with PrestoPlot

I took the Civic out for a spin today with my neatly installed PakTrakr and captured some data while pulling some reasonably high currents.

I programmed the Synkromotive controller to limit the battery amps to 250 to help prolong the life of the batteries. It takes a bit of the acceleration edge off while on the freeway at 60mph, but otherwise, the power is just fine for commuting.

After loading the PakTrakr data into Excel 2000 (yeah, I know it's old...), I experience much frustration with trying to zoom into areas to analyze the data further, especially peak areas.

I turned to a freeware plotting program that we use at work for quickly analyzing performance data called PrestoPlot. After loading the comma-delimited-data from the PakTrakr log file into Excel, I removed all but the battery voltage columns and added an incrementing list of numbers as a first column. I can now export this as a tab-delimited textfile from Excel and import it into PrestoPlot for easy viewing.

Here's an exported image (click to enlarge) from PrestoPlot showing the whole dataset gathered from the PakTrakr. The data seems very spikey and I was concerned that some of my batteries were dipping lower than other.

Upon quickly zooming in with PrestoPlot, I was able to see that the big droops on one battery were accompanied by positive spikes on the adjacent battery. Note the green dashed lines in relation to the red dashed lines in the picture above (click to enlarge). Hmm..., sounds like the PakTrakr isn't filtering out noise quite as well as I'd hoped.

In areas where the system is rather quiet, the batteries seem quite well balanced. Perhaps my pack isn't as dead as I thought it was, but it has just lost a bit of capacity. Perhaps putting some 0.1uF capacitors on the adjacent signal lines going into the PakTrakr might help with this.

Not too sure what to do at this point, but at least I get a good clean sampling stream on the serial port under heavy acceleration.



Ross Cunniff said...

"Perhaps my pack isn't as dead as I thought it was, but it has just lost a bit of capacity. "

I bet you are not charging your pack to full capacity if it is cold. I have the same problem (which is why I'm spending so much effort on battery warmers). See the Chairman Technical Manual - if your average temperature is 50 degrees F, you should be charging your 144v pack to 177 volts at constant current. If, like me you set the voltage based on the 70 degree specs, you are only going up to 172 volts.

Combine that with the double whammy of reduced discharge at low temperature, and yes, range is severely impacted. With the old floodies and the Zivan charger, I never noticed this effect because it always went into a balancing cycle, well above the needed charge voltage. You don't want to do this with sealed batteries, of course.

PS - I decided to remove the PakTrakr from the volt914. On the ElectroJeep, the draw from the first battery in each remote group was enough to cause serious battery imbalance.

TimK said...

Hi Ross,

Good to hear from you. You know what, I'm seeing the same thing from the first battery on the PakTrakr. It constantly charges last. My charger is based on the rectified 120VAC, so sometimes it just can't get over the 172 volt level even though the shunt balancers are temperature compensated.

Perhaps I can add a "PakTrakr" power-switch to only enable it when I need to check things out...


Ross Cunniff said...

Sadly, the warmers did not work out. Still looking for a solution. Probably the battery heating pads or something similar.

David Harrington said...

I have had great success with Farnam batery heater pads from KTA-EV. I created my own temperature control for them, but they keep my pack at 76F and with insulation, they only cycle on for about 20 minutes every 3-5 hours.