From: Josh (Norm) Audette <"Josh>
Date: Sat, 12 Oct 2002 07:54:25 -0400 (EDT)

<NORM>
        from firmware/powermgmt.h:
        #define BATTERY_LEVEL_SHUTDOWN 450 /* 4.5V */
        #define BATTERY_LEVEL_EMPTY 465 /* 4.65V */
        #define BATTERY_LEVEL_DANGEROUS 475 /* 4.75V */
        #define BATTERY_LEVEL_FULL 520 /* 5.2V */

        Research of NiMH cell characteristics reveals the following bits of
        information:

        o NiMH cells usually end their useful life around 1.1-1.0V. The
        recommended cut-off voltage on the discharge cycle is 1.0V, though the
        minimum recommendation I've seen is 0.8V. The real danger in
        discharging batteries comes from overdischarging them. This is an
        effect commonly seen with a series of cells, where one cell is weaker
        than the others, and discharges down to 0V or lower (reversing its
        polarity), resulting in a further weakening of that cell.

        o A fully charged NiHM battery usually has a voltage of 1.4-1.5V. I
        think if you watch your jukebox debug screen during its charge cycle,
        you'll see its voltage peak around 5.7V (a little over 1.4V/cell).

        With this in mind, and also keeping in mind two other factors: a) the
        voltage read by the jukebox is "amortized" or averaged over the four
        cells and b) there are slight variations in jukeboxes and batteries
        everywhere, I'd recommend the following configuration:

        #define BATTERY_LEVEL_SHUTDOWN 420 /* 4.2V */
        #define BATTERY_LEVEL_EMPTY 445 /* 4.45V */
        #define BATTERY_LEVEL_DANGEROUS 455 /* 4.55V */
        #define BATTERY_LEVEL_FULL 540 /* 5.4V */

        The 5.4V for full allows for variations in devices plus for the fact
        that a fully charged NiMH battery drops a little faster from 1.5 to
        1.4V before levelling off to a more constant dV/dt.

        The 4.2V shutoff allows cells to reach a low of around 1.05V per cell
        before shutting off the device. This is well above the 0.8V minimum
        recommendation I've seen. Leaving a larger difference between the
        EMPTY and SHUTOFF values accounts for the greater negative dV/dt slope
        at the end of a NiMH cell's useful life.

        What advantages does this offer us?
        The increase in the FULL value gives us a better range for the useful
        life of the battery. For example, we won't see 100% battery for the
        first three hours of operation. Instead, it'd be closer to 1 hour at
        100% and then the percentage would start changing.

        The decrease of the SHUTDOWN cutoff value gives us a little bit more
        time out of our batteries, which a) gives us more music time d:) and
        b) is a little better for the batteries, resulting in longer battery
        life (more useful charge/discharge cycles).

        The increase in the difference between SHUTDOWN and EMPTY gives us a
        little bit of extra time to run on empty.


        I apologize for the length of this message, but I find it is easier to
        consistently convey my ideas to people of different mindsets when using
        more words.

        PS: Refining these values becomes more important as people start moving
        from the 1500mAh cells that come with the jukebox to 1800mAh and
        2100mAh cells (and greater) cells that are now becoming available on
        the market.

        Another idea we might want to consider is determine the shutoff point
        of the unit based on dV/dt, just as we determine the cutoff point for
        charging. Graphs (available from NiMH manufacturers like Panasonic,
        Energizer, Duracell, Rayovac, etc.) of a cell's discharge
        characteristic show a sharp decrease in dV/dt as the cell reaches the
        end of its useful life, and it should be fairly easy to detect. Using
        this scheme you'd want to keep an absolute minimum voltage (somewhere
        between 4.0V and 3.6V) that you never go below, but the idea is to cut
        off based on the battery's unique cutoff point, rather than a set
        voltage.


        Again, I apologize for the long-winded email. I just hope the ideas
        are useful enough to merit the time it took y'all to read it.
</NORM>

-- 
  Why, oh why,
  Do my telnet sessions die?
  			- Norm, "Ode to a bad firewall"
- - -.-- .-..  - - -.-- .-..  - - -.-- .-..  - - -.-- .-..  - - -.-- .-..
Josh "Norm" Audette <norm_at_gnurple.net>       AIM: normmonkey ICQ: 25260232
Systems and Network Administrator (Contract)          <norm_at_linuxcare.com>
              Linuxcare.  Simplifying Server Consolidation.
10101001010100101100110011000100000101010010101001011001100110001000000000
. .-. - .--  . .-. - .--  . .-. - .--  . .-. - .-  . .-. - .--  . .-. - .-

Received on 2002-10-12