The current directory contains the latest versions of the following
files:

*****************************************************************

14 May 2001         Current Version: V1.03

    setup.exe       installation progrom for PC application software
    ss_anc.001      
    ss_anc.002
    ss_anc.003

    SHARC.IHX       Intel Hex file of SHARC eprom code for EZANC-II.
                    Build 15.

Changes and fixes since previous version:

1.  the length of FIR filters for modeling increased by a factor of 2. 

2.  the hardware sample rates have been corrected to reflect the correct
    frequencies of crystals in production units.  In V1.02 the sampling
    rates were 5208Hz to 32kHz in 10 steps.  In V1.02 the sampling rates
    are 5513Hz to 32kHz in 10 steps.

*****************************************************************

.\OBSOLETE\BUILD167:    V1.02, Build 167

24 March 2000

    setup.exe       installation progrom for PC application software
    ss_anc.001      
    ss_anc.002
    ss_anc.003

    SHARC.IHX       Intel Hex file of SHARC eprom code for EZANC-II.
                    Build 15.

Problems fixed since previous version:

1.  It is entirely possible that the system wouldn't work with 
    downsampling on.  It would have depended upon the group delay of 
    the physical system.  The problem was that SDS forced a set of 0's at the
    start of the transfer function model, and the previous number (12) may
    have been too many in some instances.  SDS was able to make it work with an
    acoustic system, but it wouldn't with a straight connection.  Anyway, SDS
    have put in a sample rate dependent zeroing function, which works well.

2.  Regarding the losing of weights when the config is saved to 
    flash, again this was correct.  SDS actually saved them, and then 
    wrote over them when SDS flushed the controller after loading from 
    flash.  Fixed.

    In regard to flushing, SDS has put in a separate menu item, "flush", 
    which lets the user flush only the weights or the transfer function, as
    well as everything.

3.  SDS couldn't get the save to disk to fail.  There is no transfer 
    function storage or weight storage on disk.  20,000 words was too 
    much with the current software.  TF models and weights are only 
    stored with flash.

4.  There was a bug in the transfer function numbers.  To make it
    work properly you had to load a new number of tf weights and then
    manually flush.  This has been fixed.


*****************************************************************

.\OBSOLETE\BUILD149     Version 1.01, (Build 149) 14 Feb 2000
                        SHARC build 13

*****************************************************************