thankx ..but with NLS method we won't be able to give the exact gains.. say 2.567. Because we are translating a analog controller to digital the significance of these gains might be imp.. as we will already face problems due to rounding. Can u please elaborate on vscale thankx again waitng for ur reply iqbal On Wed, 11 Oct 2000 12:34:30 Sreejith Pattuvakkaran wrote: >Hi Iqbal, > > For preserving the precesion, you need to vscale(vector scaling) >the input data.i.e. You have to convert the input data(input buffer) >to block floating point format with an nls(no. of left shifts) value. >Basicaly you need to scale all input data with that nls value. >After the computations, you have to compensate the scaling by >right shifting the output data with same scale. > >Does it answer your question?? > >regards, >sreejith > > 10% cash back on all your calls through 2000 at Lycos Communications at http://comm.lycos.com |