Tuesday, May 1, 2007

Measuring State-of-Charge on Lead Acid Batteries

Since I'm still waiting on parts from ElectroAuto, I decided to take the plunge and do some extensive research on Lead-Acid battery technology. The 914 AC kit uses 8V golf cart batteries and I'm wondering if different batteries like Hawkers or Optima Yellow-Top batteries would be better.

After researching different battery types, it looks like the Hawkers and Optimas can put out more amps for more torque and are more rugged, but the golf-cart batteries still rule in terms of capacity per weight and size. With the AC system, I'll probably be pulling less amps and am looking forward to the extended range. Both US-Battery and Trojan have 8-volt golf-cart batteries. The kit was specificially designed for the US-Battery model and the equivalent Trojan battery is just 1/8" of an inch larger in one dimension, so I'm guessing it won't fit in the battery cases provided by ElectroAuto (grrr...).

Given that I'm using lead-acid technology, one of the big challenges is knowing just how much charge is left in the batteries so I don't run them down below 20% SOC (or state-of-charge). This turns out to be a very difficult problem because drawing current from lead-acid batteries, especially at different rates, causes a widely varying voltage at the terminals. The non-linear dynamic nature of the battery makes accurate SOC measurements difficult, if not impossible.

I feel very fortunate in that the company I work for (HP) has a subscription to IEEE Xplore, so I downloaded literally a hundred scientific papers today on battery management and state-of-charge measurement of lead-acid cells.

The simplest method that seems to have the most promise takes the average of voltage and current over the past few minutes and uses that with an interpolated lookup table of discharge curves to determine the approximate state of charge. A more complex method actually forces a small (25mV) AC voltage back into the battery at a variety of frequencies to extract the complex impedance of the battery structure. If you take this measured complex impedance along with some measurements of the battery when no current is drawn for an hour, you can supposedly derive the average concentration of the sulfuric acid in the electrolyte, and thus the current SOC. The electrolyte concentration is one portion of the complex impedance model.

I've been in contact with Victor at Metric Mind here in Portland who is developing the state-of-the-art EVision battery monitoring system. We might possibly work together on a project after chatting at the next OEVA (Oregon Electric Vehicle Association) meeting.

This week is packed, so I might get to hooking up the back-up light wiring on the EV if there's time on Friday. Off to some interesting bedtime reading on lead-acid modelling technology...


brad said...

I have enjoyed reading about your EV conversion. I hope to do the same one day.

I thougt that this guy's concept of SOC montoring was interesting and that you might enjoy looking at his page:

Best of luck as you continue your project.

TimK said...

Hi Brad,

Thanks for the comment and the link. I like centralized display. Individual chargers definitely solve a bunch of problems with equalization. I'm researching some recent algorithms on SOC that I hope to implement. Have a great week.


Tom Eberhard said...

Hi there,
just discovered your blog while looking for people who have used an AC motor in their conversion. What has your experience been like with the motor from Azure? I'm planning on using one of their motors. Thanks, Tom.

sparc5 said...

So what did you end up doing to monitor SoC. I'm in the same boat. Also read hundreds of EE Spectrum papers. Texas Instruments has a "Impedance Track™" chip that does all the work for you with great accuracy but only for Li-Ion type batteries. If only they could make one for lead acid too :(

TimK said...

I have a PakTrakr installed in the car and that gives me a reasonable SOC. For a really accurate SOC you can get an amp-hour counter or an EVision system from MetricMind.com

It still would be nice to have a circuit that drives the gas gauge, but I'm focusing on the Civic-EV right now at civic-ev.blogspot.com

Cheers, Tim