




                                


              HF WEFAX FOR LAPTOPS AND PC CLONES



     With  an  8088 laptop (Sharp PC-5000), and more recently  an  80286 
laptop (Bondwell B310), I have developed some new versions of the simple 
interface Wefax2 that may be of interest to others. The programs for the 
PC-5000 use it's cassette input for FSK demodulation so no interface  is 
required. With it's built-in 24 pin printer, it gives identical MAP  re-
sults to that covered in reference 2.

     Using the B310 computer with it's 12 mhz. speed, 1 megabyte of RAM, 
and high capacity hard drive memory storage, considerably more capabili-
ty is possible. One feature I had added to the PC-5000  version which  I 
found  very valuable was the ability to store and review each map before 
deciding whether to print it or not. With the B310's hard drive, you can 
literally  store  a whole day's worth of maps and review  for  selective 
printing later.

     The  new simple interface schematic is slightly different so as  to 
fit into the most commonly available port on laptop (and other)  comput-
ers,  the  serial communications port. Figure 1 shows the  details.  You 
will note that I used the same isolation transformer as in the 2  previ-
ous  versions1,2.  To avoid power supplies, I used two of the  COM  port 
lines  as  sources  of +12 volts and -12 volts to power the  op  amp  IC 
clipper. These lines are turned ON by software. By using a low quiescent 
current IC (the 741), and limiting resistors and diodes in  series  with 
these  lines, the current load on the COM port is kept well  below  what 
the  widely used RS232 LED test boxes draw. The diodes are essential  to 
block  reverse  voltages on the software controlled lines. Even  a  full 
short  failure of the 741 should be tolerable to the computer  interface 
circuits.  An alternate is to power the 741 with two 9  volt  batteries. 
The  IC gives a well-clipped symmetrical square wave into the  DSR  port 
line. The software task is to demodulate and process this for display.

     Since  we wanted the highest possible resolution which uses all  of 
the data coming in and also wanted to maintain proper  vertical/horizon-
tal  aspect  with a standard NLQ printer, we settled on 960  pixels  for 
each line and a store and optional print of every line of data received. 
The incoming frequency is measured on every 1/2 cycle with a loop  coun-
ter  which  is started and stopped by successive zero crossings  of  the 
audio  square wave. At any time, we have stored the most  recently  com-
pleted 1/2 cycle count. The pixel period (960 periods per 1/2 sec.)   is 
created by setting the computer's tone generator to 960 cps. This  gives 
a  zero crossing once every desired pixel period. We inserted a step  in 
the  audio loop counter, such that once each count cycle (6  microsec.), 
the  pixel tone is checked for completion  of it's 1/2 cycle (ie;a  zero 
crossing).  When a pixel tone crossing is detected, the loop counter  is 
momentarily halted, and the most recent completed audio count is  passed 
into memory. Then the audio count is resumed. By interleaving the  moni-






toring  of  audio frequency and the pixel timer zero  crossing  (updated 
every  6  microsec.), one maintains precise pixel timing,  a  high  data 
rate, and information registration that is a fraction of a pixel.

     For a map, the audio count measured for  each pixel is  thresholded 
against a single value and stored as a single 1 or 0. Code at the end of 
each  line checks for keyboard HALT and writes the line on the  display. 
The 80286 speed allows this code execution to squeeze into a fraction of 
a  pixel. The digital resolution of the tone generator is too coarse  to 
set it precisely on 960 cps, so it is set a fraction of a pixel high and 
the extra time is used for the overhead processing described above  plus 
a  fine tune timer allowing line timing within about 1  microsec.  Since 
the  tone  is turned off and back on (at precisely 1/2 sec), you  get  a 
nice "tick" on the tone as an audio cue for machine operation. With  640 
KB  of RAM, you get about 30 minutes of HIREZ map storage.

     The gray shaded satellite pictures require more complex processing. 
An impact printer can give limited "shading" in a single pixel by strik-
ing a pixel spot more than once, so we used a double printer pass to get 
more  shades.  We chose a 2 across by 3 down matrix of pixel dots  as  a 
compromise  between  resolution and grey shades. A double  printer  pass 
gives 2x6+1 shades. I added one more shade at the "white" end by averag-
ing  across two matrixes to decide whether to register a  singly  struck 
dot in a 4x3 space. The number of dots to be "sprayed" in the 2x3 matrix 
is  determined  by  averaging these 6 pixels and  comparing  against  12 
threshold  values. Dot density is made a linear function of  WEFAX  fre-
quency shift.

     This broadening of the resolution cell unfortunately tends to  wipe 
out  the  fine geographical overlay lines on the  pictures.  We  recover 
these  lines from the individual pixel data and overlay it on  the  gray 
shades  picture. Gray shade processing --Measure, Average, Spray,  Over-
lay, and Screen write--needs 3 lines of data before it can start on  the 
first 2x3 matrix. Even with high speed, the time required for processing 
3  lines exceeds a pixel dwell time by considerable. To limit  the  data 
thrown away to a minimum, the processing was squeezed into the tail  end 
of  the  third line. Missing pixels are filled with counts  from  pixels 
directly above, giving a small loss of fidelity in a limited region. 

     For those wanting to cut printer time (and wear) and memory storage  
in  half, there is an 8 gray shades program which uses only one  printer 
pass.  This works better with the Diconix ink jet portable printer.  You 
can store the pix in 14 shades and print in 8 or 14 shades to match  the 
printer.

     Having used the condensed mode2 every-other-line maps operationally 
for several years, I was skeptical that I would add much real difference 
with  every-  line recording. And that is partly true,  particularly  in 
strong  signal, low QRM conditions. In poor conditions the  double  info 
rate  really  helps. Even multi-path which is  correlated  over  several 
seconds  causes less obscuration. We still include a condensed  mode  in 
the collection of programs since it uses 1/3 as much memory stores,  and 
about  1/4  as much printer paper. You can print down one  side  of  the 
paper, rotate paper 180o, print down other side;then turn paper over and 
repeat! Some maps have detail that requires the full resolution. 







     I've tried to configure the programs for maximum compatibility with 
IBM clones. The speed required dictated using a machine language program 
for each configuration, but it is interfaced to the user through a BASIC 
program  so  you can POKE in timing values to customize the  program  to 
your computer's speed. Scheduling capability to fit my needs is  includ-
ed,  but anyone with BASIC experience can reconfigure to his own  needs. 
An  Epson compatible printer that accepts graphic command  ESC,"L",n1,n2 
and  carriage advance command ESC,"3",n is required. If you have a  slow 
speed 8088 computer, you can probably get the MAP programs to work,  but 
the  GRAY programs are probably too complicated to adapt.

     Copies  of the programs are available by sending me a freshly  for-
matted  disk  with an SASE disk mailer. I can record 360 KB.  on  5-1/4"   
and 720 KB. and 1.44 MB. on 3-1/2" disks.  


References:

1   B. Vester,  "HF WEFAX for the IBM PC, PC Jr., and  C-64",  Technical 
Correspondence, May 1987 QST pp. 40-43

2    B.  Vester, "C-64 WEFAX  Improvements",  Technical  Correspondence, 
January 1988 QST pp.47-49


                                                  Ben Vester
                                             4921 Bonnie Branch Rd.
                                             Ellicott City, Md.,21043


























