9/13/97

Msg# 34608  Type: B  Stat: $  Date: 19 Feb 1620

To: AMSAT @ WWW

Subject: FO-20 Modem Modification

From : G3RUH @ GB7DDX._223.GBR.EU



             FO-20 PSK Modem Improvement for Microsat Use

             --------------------------------------------

                       by James Miller G3RUH

                            1991 Feb 19

Introduction

------------

  Some users of my 1200 bps PSK modem have commented that data recovery

with the microsats is poor, difficult or in a few cases impossible.



  Personally I have no problem whatever with my standard modem and

standard FT736R, but persistent reports prompted further investigation.



  Fuji-Oscar-20 and AO-13 (400 bps) generate "pure" PSK.  That is, the

carrier phase is either 0 degrees or 180 degrees, and nothing else.



  Extensive investigation shows that all three microsats (AO-16, WO-18 and

LO-19) generate random phase noise.  I (G3RUH) found peaks of +/-27

degrees on occasions with +/-10 degrees RMS scatter typical.



  My modem, being designed "cold" and in wide circulation (1986) before

FO-12 had even transmitted a packet, expected clean PSK.  So its post

phase detector (i.e. data) filter is an RC network R30/C3.  This is

entirely  satisfactory for FO-20, but it hasn't really enough smoothing to

deal comfortably with the microsat phase jitter.



  The cure is to upgrade the data filter.  There is a spare op-amp on the

board, so by the adding just 4 resistors and 2 capacitors, a 3-pole filter

can quickly be implemented (15 minutes).  The filter offers some 30 db

attenuation to the effects of excessive phase noise, and you'll get

effortless 100% data recovery from PACSAT, WEBER and LUSAT.



Application

-----------

  This modification is for PCBs  marked "JAS-1/FO-12 MODEM (C)1986 G3RUH",

issues 1 & 2.



Circuit

-------

  Note that the existing R30 is removed, and the input of this circuit is

from its left-hand hole, close to the legend "VR2".  The existing C3

remains and becomes part of the new filter.  The circuit uses a spare

section of op-amp U5 so only 6 components   Ra,Rb,Rc,Rd,Ca,Cb are new.



                     Cb | |

             +----------| |-------------------+

From         |       10n| |                   |

U7           |                 | \            |

pin 3        |                 |   \          |

       Ra    |   Rb          12|     \        |

>---/\/\/\/-++-/\/\/\/-+-------| +     \  14  |     Rd             To TP3

      27k   |   27k    |       |    U5   \--+-+---/\/\/\/---+----->

            \          |     13|         /  |       15k     |

            /          |    +--| -     /    |               |

         Rc \      Ca  |    |  |     /      |      [  C3  -----  ]

        27k /        -----  |  |   /        |      [  10n -----  ]

            \        -----  |  | /          |      [        |    ]

            |      4n7 |    |               |               |

            |          |    +---------------+              ===

            |          |       

        ----+----------+----> 6v (U5 pin 3)       3rd Order Bessel Filter

                                                  =======================



Note: 10n = 0.01uf   = 10000 pf )  Use 10% or

      4n7 = 0.0047uf =  4700 pf )  better.



Construction

------------

  Ra,Rb,Rc,Rd,Ca,Cb are fixed to the underside of the PCB.  Pin 14 of U5

is the pin nearest the legend "U5", and don't forget to link pins 13 & 14. 

Make sure that components and leads cannot accidentally touch PCB tracks.



Test - the "EYE"

---------------

  Tune in a satellite as normal.  Set the oscilloscope to 2v/div vertical

and timebase 0.2 ms/div.  Trigger from TP4 (1200 Hz), and examine the

waveform at TP3.  You should see the characteristic "eye",  with amplitude

6v pk-pk, centred on approximately 6v.



             |<---  1 bit  --->|

    9v    . . . . . ... . . . . . . . ... . . . . . .  1 s

                  .      .         .        .     

          .     .          .     .            .     .         Typical

            . .              . .                . .           data eye

    6v       .                .                  .            pattern

            . .              . .                . .

          .     .          .     .            .     .

                  .      .         .        .     

    3v    . . . . . ... . . . . . . . ... . . . . . .  0 s

                     ^                 ^

                   sample            sample



  The TNC samples this waveform at the widest part of the eye.   If it is

above the mid-line, it detects a "1", if below a "0".  So the quality and

openness of this eye is a measure of detection robustness.



Don't Forget

------------

 Two points that are as important now as they ever were.



1.  AUDIO CARRIER FREQUENCY.  Make sure that you have the optimum audio

carrier frequency.   It'll be in the region of 1400-1700 Hz (at TP1). 

Lock the modem on to a satellite at the start of a pass when doppler

changes slowly.  Look at the eye and VERY, VERY gently vary VR1.  This

will remotely auto-tune the radio up and down. Leave VR1 at the point

where you judge the eye to be at its best; wide open and symmetric.



2. DETECTOR THRESHOLD.  Input SSB radio noise to the decoder.  Adjust VR2

for average mid-scale reading on the centre-zero tuning meter. 

Alternatively, set it for exactly 50% of the 12v supply.  It should

ideally be the same voltage as at the mid level of the eye.



Thanks

------

  Special thanks to VK5AGR, VK5HI, VK5ZK and G4WFQ for acting as testers

to validate these findings.



Distribution

------------

  You are invited to circulate these notes on all appropriate BBSs.



   73 de James G3RUH @ GB7DDX 1991 Feb 19   (Please note my new BBS)