home *** CD-ROM | disk | FTP | other *** search
- rev.3/2/87,V1.01
- MXFILTER.COM HAS BEEN REMOVED FROM THIS DISK AND PUT ON DISK 4. NOTE
- THAT MXFILTER.PAS IS & HAS ALWAYS BEEN ON DISK 4.
- THIS WAS DONE TO MAKE ROOM ON THIS DISK FOR CORRECTED VERSIONS OF FFT AND
- LUDEC. The com files on this disk,(except CHEBFIL)require a math coprocessor.
- Pascal programs on this disk may be compiled with Borland's
- Turbo Pascal compiler. CHEBFIL also requires Borland's Turbo Graphix Toolbox
- to compile it. The .FON files, ERROR.MSG & PLOTGR.CHN must be resident for
- CHEBFIL.COM to run. It requires a graphic terminal.
- A compiled version of FFT.PAS has been added to this revision. The files
- FDFT.IN and JEC have been added. The Ludec program has been revised.
- Tip for you non-pascal users, data must be seperated by one or
- more spaces when entered into pascal programs. G.H.
-
- Mr. Jeff Crawford submitted the following programs and the following
- description of them:
-
- MXFILTER, see disk 4
- BAIRSTOW
- ROMBERG
- SPIRLIND
- CHEBFIL
- LUDEC
- FFT
- CIRANL
-
- All programs ( or subroutines) have documentation internal to the
- software: i.e. all references. Enclosed is a brief description of
- each.
-
- MXFILTER: This program allows one to estimate the maximum
- possible passband width while satisfying specific out of band
- requirements. At the present time it is set up for a maximum of 4
- out-of-band frequencies. The limitation lies in how much
- information may be displayed on the screen at one time. If this
- were changed, no other modification would be necessary.
-
- BAIRSTOW: A complex root finder based on the Barstow method.
- This method is adequate for general polynomials to possibly 15th.
- degree, depending largely upon the location of the respective
- roots. For polynomials exhibiting widely spaced roots, the
- accuracy is limited for the higher order polynomials.
-
- ROMBERG: An integration program allowing the user to place
- any integrand into the function "fnc" block and doing the
- integration. This is far more than a general integration in that
- it encompasses Richardson's Extrapolation which GREATLY increases
- convergence time for a minimal number of steps taken. With
- Richardson's convergence, only 16 steps may be necessary compared
- to 128 or 256 without it, dependent upon the behavior of the
- function naturally.
-
- SPIRLIND: A program which uses Neumann's form to calculate
- the dc value of inductance of round spiral inductors. The
- negative mutual inductance is also calculated due to the
- proximity effects of the ground plane.
-
- CHEBFIL: A program which calculates the theoretical response
- of Chebyshev filters when given real-world Q values. The poles of
- the respective Chebyshev polynomial are calculated in each case
- and the group delay calculated. This information is calculated at
- each frequency of interest and used with the real-world Q to very
- accurately determine the in-band insertion loss and roll off near
- the ripple frequencies. If accurate values for the Q being used
- are used with this program, the results agree in-band very
- closely. Graphics is included and uses the graphics package from
- Boreland International.
-
-
- LUDEC: LU Decomposition for the solution of systems of
- linear equations. This form is optimal in that the entire
- calculation is performed "in place" and no additional memory is
- required. Errors are less than with the typical Gaussian
- reduction. Ex: order=3, A(1,1)=2,A(1,2)=-1,A(1,3)=0,A(2,1)=-1,
- A(2,2)=2,A(2,3)=-1,A(3,1)=0,A(3,2)=-1,A(3,3)=2, B(1)=6,B(2)=0
- B(3)=-6. THEN X1=3,X2=0,X3=-3. WHERE A*X=B.
-
- FFT: Performs forward and inverse FFT's for any number of
- increments which is a factor of 2, 16, 32, 128 etc. The "fft.in"
- file has a routine in it which fills the arrays used in the
- computation. The fft.in file on the disk fills the array with a
- damped sinusoid. The data created by this file is called "jec".
-
-
- CIRANL: Performs frequency analysis of any circuit entered
- into the circuit definition file contained within. The routine is
- based upon the floating admittance matrix. One starts out with an
- n x n matrix and the routine "collapses" this matrix to a 2x2
- with the input and output ports (which are variable) as the user
- defines. All four admittance parameters are calculated and may
- be used in any of the conventional two port representations for
- further analysis. At present only subroutines for R, L, and C are
- included but may be easily expanded to include active devices,
- transformers, transmission lines, and other useful elements.
-
-