Re: (ET) FW: Fully charged pack voltage

["David Roden" <etpost drmm net>]

On 28 Aug 2009 at 23:29, harry landis wrote:

> With sufficient rest and stable temperature, voltage measurements
> provide an amazingly accurate SoC estimation for lead acid batteries. 

Sorry, as posted earlier, I disagree.  In support of this position I quote 
here the flat statement of electrochemist Nawaz Qureshi, Vice President of 
Engineering, US Battery Mfg Co : "Voltage is not an accurate indicator of 
state of charge (SOC)."

I'm not an electrochemist myself, but I think the problems with sensing 
SOC 
with voltage are that :

1. Fully-charged voltage varies with the battery's design;

2. Fully-charged voltage varies with the battery's age;

3. Fully-charged voltage varies with the battery's temperature; and

4. The difference between fully charged OC voltage and flat OC voltage is 
fairly small, less than half a volt for a 6 volt battery, so a small error 
in measurement creates a significant error in SOC reporting.

SG is still the most reliable measurement.  

Alas, I don't know of any chargers which sample the electrolyte's SG.  So, 
many smart chargers use _on-charge_ voltage, compensated for battery 
internal temperature and combined with charge current sensing.  This is a 
convenient and relatively easy way to estimate state of charge while 
charging, to control a charge algorithm.  A seldom used improvement on 
this 
algorithm takes battery age into account, since the peak on-charge voltage 
usually falls as a battery ages (but see below).

Some other charge control schemes have been tried with varying degrees of 
success.  For example, the term "valve regulated" for semi-sealed 
batteries 
is a leftover from a scheme which used the gassing pressure of a closed 
cell 
to regulate the charge.

One rather good charge control scheme counts the amp hours withdrawn from 
the battery in use.  When charging it replaces those amp hours, plus a 
percentage for charge efficiency.  This scheme is often combined with some 
kind of voltage and current sensing as a way to improve confidence in its 
accuracy.  I think this is (or at least once was) the way most laptop 
computers charge their lithium batteries.  Someone please correct me if 
that's wrong.

Dv/dt is a fairly elegant algorithm, a variant on voltage sensing (dv/dt 
== 
derivative of voltage with respect to time).  Instead of calculating and 
waiting for for perfect "I'm full" magic voltage and current numbers, this 
approach watches the voltage rise as the battery charges, relying on the 
fact that the voltage rise normally flattens at the "full" point.  Dv/dt 
automatically compensates for temperature and battery age.  It usually 
includes a minimum voltage to avoid terminating early on a long, slow 
charge, and also detection of inverse dv/dt (thermal runaway protection).  
I 
think Lester Electrical may still hold the patent on dv/dt.


David Roden - Akron, Ohio, USA

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = 
Note: mail sent to the "etpost" address will not reach me.  To send 
me a private message, please use the address shown at the bottom
of this page : http://www.evdl.org/help/
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =