QRZ! Ham Radio 11/QRZ Ham Radio Callsign Database - Volume 11.iso/files/packet/tapr89.txt

Text File | 1998-02-02 | 40KB | 755 lines

Blow-by-blow Report of the 1989 TAPR Annual Meeting

by Paul Williamson, KB5MU
from the SANDPAC Newsletter


On Saturday and Sunday February 25 and 26, 1989, the Tucson
Amateur Packet Radio Corporation (TAPR) held their annual
meeting.  Approximately 75 people were in attendance from around
the US, plus one each from Canada and Brazil.  TAPR President
Andy Freeborn, N0CCZ, opened the meeting promptly at 9:00 AM on
Saturday.  He announced the results of the Board of Directors
election.  The following were elected to serve a three-year term
on the Board: Franklin Antonio, N6NKF; Bdale Garbee, N3EUA; Steve
Goode, K9NG; Eric Gustafson, N7CL; and Lyle Johnson, WA7GXD.

A discussion of the possibility of a new Amateur license class
not requiring a Morse Code examination was on the agenda for
Sunday morning.  In preparation for that discussion, N0CCZ took a
poll of those present.  35 were in favor of a codeless license,
13 opposed, 7 uncommitted.

Harold Price, NK6K, served as program chairman for the meeting.
He introduced Phil Karn, KA9Q, who spoke about "Recent
Developments in TCP/IP".

        Phil Karn, KA9Q: TCP/IP

Karn showed a diagram of the Internet protocol hierarchy to
remind the audience of the general structure of the protocols.
Since he has given the introductory talk at previous TAPR
meetings, he gave only a quick overview of the protocols.  He
then showed a very long list of companies that sell or support
TCP/IP-based products, to show that TCP/IP is a widespread de
facto standard.  He then listed the software modules that are
currently a part of the KA9Q TCP/IP implementation:

IP Applications: FTP, Telnet, SMTP, Finger

AX.25 Applications: Mailbox, BBS/Internet Gateway

Protocols: TCP, UDP, IP, ICMP, RIP, NET/ROM, Ethernet, AX.25,
SLIP, AppleTalk

Drivers: drivers for many Ethernet cards

Karn went on to point out that NET's capability to use a NET/ROM
network to carry IP packets is perfectly in keeping with the
Internet concept of tying existing facilities together into a
working network.  However, link layer protocols cannot patch over
physical layer problems.  That is, the radios and modems must be
able to communicate bits from one station to the next reliably
before the protocols, no matter how well designed, can do a good
job.

An industry standard for Ethernet device drivers has been agreed
upon, and Wollongong (a major TCP/IP vendor) has agreed to
release their device drivers to the public domain as they are
converted to the new standard.  The NET software has been
converted to work with the Packet Driver standard, so it can take
advantage of this working base of driver software.  In addition,
drivers already exist for the Eagle board, HAPN (Hamilton Area
Packet Network) board, the DRSI PC-Packet Adapter, and the
Pac-Comm PC-100 board.  KISS TNC firmware exists for almost all
TNC's, including the TAPR TNC-1 and TNC-2, the VADCG board, the
AEA PK-232 and PK-87, and various Kantronics models.

The major work in progress on the KA9Q NET software involves an
extensive rewrite of the internal structure of the software.
Karn has implemented a simple multitasking kernel (referred to as
"NOS") to replace the awkward up-call interface of the existing
software.  It uses lightweight processes and the Berkeley sockets
interface.  The main idea is to make it easier to integrate
applications software with the basic networking software.  Other
work in progress includes automatic routing, possibly including
the experimental SPF protocols currently being tested on NSFNet,
and intelligent host interfaces for use with computers without
plug-in adapters.

Karn has switched compilers from Aztec C to Turbo C version 2.0.
This is good news for many who are interested in making
modifications to the NET software, since Turbo is much less
expensive, more available, and faster than Aztec.

Karn has also been working on various performance enhancements in
the NET package.  The amateur packet environment has unique
problems due to the very low speed of the radio channel, so it
presents an opportunity for original research in protocol design.
Some of the ideas Karn is working on include round trip time
variance estimation, slow start, congestion window thresholding,
and retry timeout clamping.  He is also adding the capability to
fragment IP packets over multiple AX.25 frames, allowing the use
of shorter frames for better performance over poor channels.

        Greg Jones, WD5IVD: TexNet

The next speaker was Greg Jones, WD5IVD, of the Texas Packet
Radio Society, which designed, installed, and operates TexNet.
TexNet is a 9600 bps statewide network running on custom
hardware.  It uses RCA 700 transceivers in the 440 MHz band.
Jones showed a diagram of the network, which currently includes
three links via leased telephone lines.  Linked lines are used as
a stopgap to help build network connectivity, which helps to
build interest in the network in the newly-connected areas.  The
leased lines can then be replaced with amateur radio links.

TexNet includes a Network Management System, which performs
several functions for the network.  The network keepalive
function of the NMS polls each network node every eight minutes.
If any node fails to respond to several consecutive polls, the
NMS issues a reset command to that node.  The poll responses
contain various information about the operational status of the
node, which is used by the Report Collection and Analysis
function of the NMS.  The information is collected into reports,
which are made available on the network PMS message system for
users to download.  The data collection is also used for network
engineering, permitting the network managers to detect and
correct nodes that have failed or have become isolated from the
network.

Some sample network statistics were presented.  In one 14-day
period, the network recorded 674 connections, of which 137 were
to the network Packet Message Systems (there are three), 403 were
connections through the network to another station, and 5 were to
a node's Conference Bridge facility.  (They consider the
Conference Bridge to be underutilized, and are considering making
the conference network-wide to increase interest.)  An average
weekday sees 2 megabytes of data pass through the network; on a
weekend day the figure is 9 megabytes.

Other networks based on the TexNet hardware and software exist in
Oklahoma and Michigan.  Maps of these networks were shown.  The
Texas group currently has 300 members, and supports three active
working groups: the TexNet Support Group, the Texas Networks
Group, and the Mailbox/BBS Group.  They are currently working on
several projects: a protocol converter/gateway to other networks,
such as TCP/IP or NET/ROM; a new version of their weather
information server, required to interface with the National
Weather Service's new distribution system; software enhancements;
and backbone radio improvements.

        Tom Clark, W3IWI: MicroSat

Tom Clark, W3IWI, took the podium to speak about the MicroSat
project.  The first Ariane 4 launch, currently scheduled for July
15, will carry four MicroSats and two UoSATs of similar design.
The MicroSats will be Pacsat/NA (North American packet
satellite), Pacsat/LU (Argentine packet satellite), DOVE (Digital
Orbiting Voice Encoder), and a satellite from Weber State College
containing a CCD video camera.  All of these satellites will
carry a BBS-in-the-sky system for worldwide store and forward
mail and bulletin transmission.  They will ride on a platform
installed atop the top stage of the Ariane 4 rocket, in the
otherwise-unused corners around the base of the primary payload.
This space provides an inexpensive opportunity to launch several
small satellites.

More opportunities to launch small satellites may be afforded by
Pegasus, a commercial rocket plane being developed by American
Technologies.  Pegasus is a solid-fuel rocket with a small wing,
which can be dropped from a B-52 like the X-15 experimental
rocket plane.  Since the solid-fuel rocket has no throttle, the
main way to control the final orbit is by adding ballast to the
payload.  The Pegasus has been configured so that small
satellites can be used for this ballast.  This may provide a
stream of cheap launch opportunities for MicroSats

The satellites are quite small, only 7 inches on a side.  They
are constructed from five modules, which are stacked up and
bolted together to make the satellite's structure.  Each of the
modules has a different function.  The basic four modules contain
the transmitter, receiver, CPU, and power systems.  The fifth
module contains whatever special application is desired, such as
the CCD camera on the Weber satellite.  All six sides of the cube
are covered with solar panels, which together provide an average
of 6 to 8 watts.

This slim power budget forces the design to be very frugal with
power.  YT3MV has designed a transmitter for the MicroSats that
is 80% efficient from DC to RF.  The CPU is a low-power design by
WA7GXD inspired by the PS-186 and based on an NEC V40 processor.
It contains a 2K boot ROM, 256 kilobytes
error-detecting-and-correcting (EDAC) program RAM, and 8
megabytes of static RAM, and consumes approximately 1.5 watts.
The receiver is a high-performance design by W3IWI.  It makes
remarkable use of low-cost commercial components, such as a
Motorola receiver chip designed for cordless telephones and
inexpensive Toko coils from Digikey.  Clark displayed a prototype
receiver in the flight configuration.  Power budget calculations
indicate that the MicroSats will be capable of continuous
operation.

Another innovative cost-reduction feature of the MicroSat design
is the interconnection scheme.  Instead of an elaborate handmade
wiring harness connecting signals every which way (construction
time: 2 months), the MicroSat is connected largely by a single
ribbon cable between the modules (construction time: 20 minutes).
Each modules contains a Motorola 14469 AART chip, which enables
the modules to communicate serially over a miniature local area
network.

Clark then discussed the uplink bandplan of the two Pacsats.
Their uplink passbands are divided into four 16 kHz channels
each, spaced at 20 kHz.  Since the satellite band on 2 meters is
only 140 kHz wide, the highest channel of one satellite coincides
with the lowest channel of the other.  Two different modulation
schemes are planned.  1200 baud Manchester FSK, compatible with
existing Fuji-OSCAR 12 modems, will be used by less sophisticated
stations.  These stations need not track their uplink frequency
according to the satellite's Doppler shift, since the passband of
each channel is wide enough to accommodate Doppler plus the
narrow 1200 baud waveform.  4800 baud FSK, however, requires more
bandwidth, so users of this waveform must precompensate their
uplink frequency to remove the Doppler shift.  It is expected
that this waveform will be used primarily by gateway and mail
forwarding stations.

Clark explained the choice of Manchester FSK at 1200 baud and
straight FSK at 4800 baud.  The use of HDLC on the baseband data
ensures (through bit-stuffing) that the fundamental frequencies
in the baseband data are 75 Hz to 600 Hz for a 1200 baud data
stream.  These frequencies do not pass through a normal SSB
filter.  Manchester encoding amounts to a DSB modulation around a
1200 Hz carrier, shifting the spectrum right into the center of
the SSB filter.  With four times the data rate at 4800 baud, the
baseband frequencies are four times greater: 300 to 2400 Hz.
This is exactly the passband of an SSB filter, so no Manchester
encoding in necessary or desired.

The choice of waveform is made on a per-channel basis under
software control.  The Harris chip used for demodulation provides
an appropriate matched filter for either waveform on command.
This will permit some channels to be operated at 4800 baud for
the use of gateways while others are operated at 1200 baud for
user stations.

Clark responded to several questions from the audience.  Q: Where
are the bar magnets?  A: The satellites are stabilized by four
bar magnets, which are mounted in the corner edges of the
satellite.  The magnets also provide some additional structural
support to the satellite frame.  Q: How does TAPR fit in?  A:
TAPR and AMSAT have a heavy overlap of personnel.  In addition,
TAPR has funded much of the hardware for Pacsat/NA, for a total
contribution of $21,300.  Q: What about thermal considerations?
A: A detailed thermal model shows that everything runs cold.  The
temperature is predicted to be 0 C plus or minus 5 degrees C.
This is a good temperature for the NiCad batteries.

        Jon Bloom, KE3Z: MicroSat Power Module

Jon Bloom, KE3Z, of the ARRL Laboratory, described ARRL Labs
contribution to the MicroSat project.  He has designed the power
module, including the batteries and battery charge regulator.
The solar cells are arranged in clips of 20 cells in series,
providing about 22 volts when illuminated.  The power control
system provides 10V unregulated and 5V and 8.5V regulated to the
rest of the satellite.  The power control system is heavily
instrumented with telemetry sensors.  A clever circuit based on a
saturated toroidal core is used to measure current without
dissipating much power.

        Harold Price, NK6K: MicroSat Software

Harold Price, NK6K, described the MicroSat software design.
Since the processor in MicroSat is a V40-based system, software
development can rely heavily on standard IBM PC compilers and
tools.  MicroSat software is written in assembler and Microsoft
C, linked with Microsoft LINK.  It is then loaded into the
satellite's processor by a cooperative process between the
satellite's bootstrap ROM loader and the ground station PC.

A primary goal of the software design is ease of implementation
for new satellite applications.  A small multitasking operating
system has been implemented, along with an operating system
services library that emulates the standard "stdio" C library
interface.  It provides the application with a disk-like
interface to the 8 MB bulk memory.  This operating system is a
port of QCF, a product developed by NK6K's company and donated to
AMSAT for satellite use.  WB6YMH is implementing all the low
level I/O routines that communicate directly with the satellite's
hardware.  With these facilities, the applications developer need
only program in portable C and link with the provided libraries
to develop satellite software.  This design enables the software
team to be divided between the operating system and applications
developers, which has never been possible in previous amateur
satellite software development efforts.

The software in the satellite is constantly collecting data from
many sensors throughout all the modules of the satellite.  This
telemetry data is transmitted (at low power) on the downlink
whenever the downlink is not otherwise occupied.  It is also
accumulated into files, which are made available to users as
downloadable files.  In the case of the Weber satellite, some of
the CCD pictures will also be made available for downloading.

The satellite is protected against software bugs.  The only
software which is burned into PROM is a small simple bootstrap
loader.  The actual operating software is uploaded (after launch
if necessary) into EDAC RAM, where it is protected against most
radiation upsets by the EDAC circuitry.  A watchdog timer resets
the processor if the CPU fails to kick it periodically.  A
hardware "Fire code" circuit allows a ground station to reset the
satellite's processor even if it is totally crashed.

The satellite is expected to be easy to access.  The downlink
signal should be quite loud, comparable to the signals from the
Space Shuttle SAREX transmissions.  Omnidirectional antennas
should be adequate, except when the satellite is in view of
heavily populated areas.  An FO-12 modem is the only special
hardware needed to communicate with the satellite.

        Jan King, W3GEY: MicroSat

MicroSat program manager Jan King, W3GEY, displayed the MicroSat
chassis that was used for vibration tests.  It was subjected to
15 G's RMS.  He said that all the solar panels  (which were once
believed to be the critical-path items)had been received from the
vendor.  Each satellite has two transmitters; it is possible (but
not currently intended) to operate both at once.  It is yet to be
determined whether the two transmitters will be placed on the
same frequency or two different ones.  The satellite is expected
to have a long service life.  Its lifetime will probably be
limited by radiation damage to the large RAM array.  The
batteries are 6 Ah GE aviation NiCads, which have functioned
flawlessly on UoSAT for over five years. The total materials cost
for a MicroSat is approximately $43,500.

        Dave Toth, VE3GYQ: HF BBS Network

Dave Toth, VE3GYQ, discussed the HF network of coordinated BBS
systems.  This system is designed exclusively for mail
forwarding.  No users are permitted to use the system directly,
and users are discouraged from transmitting on the network
frequencies.  Several frequencies on various bands are used,
depending on propagation and the required range.  40 meters is
used for intrastate forwarding, 20 meters for interstate
forwarding, and 15 meters for intercontinental forwarding.  30
meters is also used, sometimes in night/day alternation with 10
meters.  Each frequency has an assigned manager, and Toth
provides central coordination for all frequencies.

The network has several deficiencies.  It has low throughput, due
to low data rate and difficult conditions on HF.  It is subject
to the vagaries of HF propagation.  And, it is subject to the
hazard of new or inconsiderate packet operators.  They sometimes
operate on the network frequencies, causing drastic reductions in
network throughput.

In December 1988, the HF network experienced a congestion
collapse.  Throughput remained near zero for several days.  It is
believed that the problem was caused by network operators raising
the retry limit in response to poor conditions.  As more and more
stations retried more and more times, the network approached a
state in which the channel was saturated with retransmissions.
When all stations backed off the retry limit to the original
value and held back mail transmissions for a few days, the
network returned to normal operation.

The HF network plans to test a variation on the AX.25 protocol
called "prioritized ACK".  The idea of this scheme is to give
acknowledgments effective priority on the channel over other
transmissions.  This is thought to reduce the performance
degradation due to the hidden terminal problem.  Suppose station
A is transmitting to station B, and station C has a packet for
station D that it is ready to transmit.  Station C hears station
A, so it is waiting for the end of A's transmission.  In AX.25
Version 2.0, station C would transmit its data as soon as A's
transmission ended (allowing for DWAIT and possibly random
backoff time).  In the prioritized ACK version, station C is
required to wait long enough for station B to send a RR frame to
station A before transmitting.  This permits the RR from B to
arrive at A without collision, even if C cannot hear B.  The HF
network will attempt to determine if this modification actually
leads to increased throughput.

        Mike Lamb, N7ML: No-Code License

After lunch, Mike Lamb, N7ML, of AEA spoke about the code-free license
proposal.  Lamb is on the ARRL committee to study the possibility
of proposing a no-code license to the FCC.  He pointed out that
the early days of packet radio were marked by not only technical
accomplishments, but also impressive salesmanship by the likes of
Pete Eaton.  He asked if TAPR members would be willing to act as
salesmen for a no-code proposal; 15 to 20 hands were raised.
Lamb pointed out that six MHz of spectrum at 800 MHz was
estimated by the FCC to be worth $1,200,000,000, and that the
amateur UHF-and-up spectrum is extremely valuable to commercial
interests.  He stated that bands like 902-928 MHz will be as
important to the Amateur Radio Service in a few years as 144-148
MHz is now, and that we should seek to populate these bands now
to avoid losing them to commercial interests.

        Bdale Garbee, N3EUA: N6GN 1200 MHz Transverter

Bdale Garbee, N3EUA, had several projects to discuss.  The WA4DSY
56 kbps modems have a 28 MHz IF interface, and thus require a
transverter to 220 MHz or higher.  These transverters are quite
expensive and not very available.  Also, spectrum space on 220
and 450 MHz is scarce.  To solve this problem, N6GN is working on
a transverter from 28 MHz to 1200 MHz for use with the DSY
modems.  Spectrum is still available on 1200 MHz, omnidirectional
antennas are practical, and Japanese RF building blocks are
available inexpensively.  The design can also be used at 900 MHz.
The N6GN design, which is currently in prototype, provides 1 watt
or 18 watts if a final Mitsubishi brick is installed.  It has
either a 28 MHz or a 144 MHz IF, permitting use with the DSY
modems or with a 2m satellite station.  It provides fast T/R
switching using PIN diodes, and can be run full duplex by
splitting the LO.  The materials cost is estimated at $150 to
$200.  Garbee asks that potential users of this design contact
him with comments.

        Bdale Garbee, N3EUA: Microwave Ethernet

Garbee is also working on a project to use 10 GHz or 24 GHz Gunn
diode transceivers designed for police radar applications to
transmit very high speed data.  He uses standard Ethernet
transceiver chips with the clock oscillator modified to produce
2.4 Mbps instead of the usual 10 Mbps.  The reduced rate is
necessary to permit use of a Motorola FSK demodulator chip on the
receive side.  He has a prototype working on the bench at 2.4
Mbps.  Operational concerns include stability over temperature
and the validity of current path performance predictions.  Garbee
estimates that a 2 Mbps half-duplex system can be built for
around $200, plus the cost of the computer and Ethernet interface
card.

        Bdale Garbee, N3EUA: KA9Q NET Release

Garbee is also handling the release of KA9Q's NET software.  One
final release of the current (122587.33 derived) software is
planned before the initial release of KA9Q's new NOS NET.  This
last release will contain many changes submitted by many
individuals.  Garbee has been receiving about 100 electronic mail
messages per day, about one third of which contain code
fragments.  Several HP and Apollo technical writers have
volunteered to update and enhance the user documentation for the
package.  Garbee was unwilling to commit to a schedule for the
release, but admitted that late March was a possibility.

        Mike Chepponis, K3MC: Awesome I/O Board

Mike Chepponis, K3MC, described his development of the "Awesome
I/O Board".  The board is a smart I/O controller designed to plug
into a slot on an IBM PC or AT computer.  It contains an 85C30
serial communications controller chip, connected by four DMA
channels to an onboard NEC V40 processor running at 8 MHz.  The
85C30 provides two full-duplex channels.  The V40 shares 8K or
32K of dual-ported RAM with the host PC via the 8-bit XT bus.
Chepponis displayed a wirewrap "not quite working" prototype
board.

Chepponis described two other configurations for similar designs.
In one, the V40 stood alone with its own ROM and RAM, and was
connected to multiple 85C30 SCC's.  One SCC was connected for DMA
and used for a high-speed channel, and the others were connected
for interrupt-driven I/O and used for low and medium speed
channels.  In a second configuration, a stand-alone V40 was
connected to a single 85C30 for DMA.  One channel of the V40 is
used for high-speed communications, and the other channel is
connected to a Macintosh host computer's Appletalk network port.

Chepponis listed a few of the applications he envisions for high
data rate packet radio.  At 56 kbps, "instant" mail delivery, six
TDM digital voice signals, FAX at less than 10 seconds per page,
digital SSTV, and multiperson realtime interactive services such
as remote schematic drawing become possible.  At 2.0 Mbps, file
servers, compressed fast scan TV, 200 TDM digital voice signals,
and various picture formats (even animated) are possible.

Chepponis compared the cost of a conventional 1200 bps packet
station ($1000) to that of a 56 kbps packet station using his I/O
board ($1700).  In terms of bits/second/dollar, the 56 kbps
station is about 25 times more cost effective.

        Phil Karn, KA9Q: WA4DSY 56 kbps Modem

Phil Karn, KA9Q, took the podium again to discuss the design of
the WA4DSY 56 kbps modem.  He stated that this modem is not
well-known in the amateur community.  He has a network of three
DSY modems working on 220 MHz in northern New Jersey.  The DSY
modem uses a modified form of MSK (minimum shift keying).  In
terms of phasor diagrams, MSK can be understood as a vector that
rotates 90 degrees every bit time, at a constant amplitude and
phase velocity.  Thus, it traces out the unit circle on the
phasor diagram.  This modulation scheme has a constant envelope,
but a wide spectrum.  The DSY modem achieves a narrower spectrum
by having the phasor cut the corners of the unit circle, tracing
out a diamond on the phasor diagram.  This results in sinusoidal
phase velocity, and a non-constant envelope.  The DSY modem
generates a signal about 74 kHz wide at 56 kbps, or plus and
minus 14 kHz deviation.

The DSY modem consists of three circuit boards, available from
GRAPES.  The Transmit Encoder board contains a clock generator,
data scrambler (to remove any low-frequency or DC components from
the data), and a state machine waveform lookup table driving two
DACs.  The DAC outputs form the I and Q channel signals, which
are lowpass filtered and passed to the Modem board.  The
modulator consists of a standard I/Q modulator with a 0 dBm 28
MHz output and a receive chain based on a Motorola 3357 FM IF
chip with a crystal controlled LO.  The analog baseband data is
sent to the Receive Decoder board, where a slicer and descrambler
recovers the data and a PLL recovers the clock.

Karn stated that he is a big fan of the DSY modem.  He beleives
that the user stations should be operating at least 56 kbps.  The
channel access problem is critical.  He recommends crossband full
duplex operation, so that collisions may be detected and aborted.

        Dan Morrison, KV7B: Modem Comparison

Dan Morrison, KV7B, described his experiments in collaboration
with Eric Gustafson, N7CL, comparing a phase detection
demodulator for FSK with a simple filter demodulator.  Their goal
is to better understand how HF packet works and how to improve
performance of HF modems.  They used a Kenwood TS-440S and a
Dalanco Spry DSP board together with a beta-test DSP signal
analysis software package running on a PC.  They used the DSP
software to simulate the two demodulators.

Morrison showed several graphs of power versus frequency that
show a discrepancy between the theoretical power spectrum for
random data and the spectrum of actual packet data.  The actual
spectrum shows an additional peak in the center of the main lobe,
which Morrison attributes to the HDLC bit stuffing.

Morrison modeled a naive discriminator, having a linear response
over the frequency range of interest.  This demod performed very
poorly.  Then he modeled a filter demod, whose graph resembles
that of the discriminator to some extent.  Integrating the
frequency response of the filter demod results in a curve that
matches the spectrum of the signal fairly well; a perfect match
would imply an optimal demodulator.  This demod generates good
eye patterns.  Next, he added limiters to the filter demod.
Filters have a reputation as the bane of HF demodulator
performance.  It still generated good eye patterns.  He then
added noise, and still got good eye patterns.  He concludes that
the filter demod works just about as well as the phase detection
demod as long as the signal is bandlimited before the
demodulator.

Morrison presented a graph of a signal before and during a severe
multipath fade.  The attenuation of one tone was clearly visible.
These signals were demodulated, and both demods did well as long
as the envelope of the signal was above the noise level.  A
member of the audience asked "why not demodulate just one tone?",
and Morrison replied "That won't work very well.  What if the
multipath fade took out the tone you were demodulating?".  Tom
Clark, W3IWI, stated that the conceptual model of "tones" was a
poor one for modulations with bit rates about as big as the
shift.  Perhaps the incidental amplitude modulation caused by
selective fading could be used by a sufficiently smart
demodulator.

Morrison then presented graphs of eye patterns for a very noisy
signal received with a 2.4 kHz SSB filter.  This generated a
hopelessly bad eye pattern on the phase demod, and a pretty good
eye pattern on the filter demod.  This is a result of the
narrower frequency response of the filter demod.  Narrowing the
input filtering to the phase demod to 500 Hz produced similar
performance to that of the filter demod.  Morrison concluded that
a well-designed filter demod works very much like an ideal
noncoherent demodulator.  Either modem needs a narrow bandpass
filter before its first nonlinear stage for good performance.

A member of the audience asked if increasing the shift from 200
Hz to 600 Hz would help performance.  His intuition was that it
would.  Morrison stated that it would not, and Clark agreed.
Both had run experiments verifying their conclusions.

        Eric Gustafson, N7CL: TAPR Hardware Projects

Eric Gustafson, N7CL, described three new developments from
TAPR's hardware designers.  Two of them are tiny circuit boards
designed to improve the DCD performance of a TNC's demodulator.
These circuits were described in the last Networking Conference
proceedings.  One is for a TNC-2, and the other is for AM7910
based modems.  The other new development is a hardware upgrade
path for TNC-1 owners.  This board piggybacks on the TNC-1 board
and permits the TNC-1 to run TNC-2 firmware.  This allows TNC-1
owners to keep up with the evolution of TNC firmware, even though
TNC-1 firmware updates have not appeared.  None of these items
are priced or available yet.

Gustafson described the prioritized ACK proposed modification to
the AX.25 Level 2 protocol.  He stated that code implementing
this version of the protocol for TNC-2 will be available for
testing on TOMCAT and CompuServe soon.  The code is based on TAPR
version 1.1.6.

        Jim Vogler, WA7CJO: 10 GHz EME

Jim Vogler, WA7CJO, was introduced by coworker Tom Clark, W3IWI.
Vogler was the first to succeed at 10 GHz EME.  He showed us
pictures of his EME station.  He has a 15' dish with a moving
mass of one ton and a half-power beamwidth of 0.5 degrees.  He
points the antenna at the moon by listening for the noise
reflected from it at 10 GHz, and tracks by means of a video
camera mounted on the edge of the dish.  He can control antenna
pointing to 30 seconds of arc, and frequency to a few hundred
Hertz.  He uses a surplus 100W TWT amplifier with a scalar ring
feed for about 20 MW EIRP.  The TWT beam was defocused, so he
used permanent magnets to refocus the beam.  He was able to hear
his echoes from the moon with 15-20 dB S/N.

Vogler said a few words about the problems of creating a power
amplifier for 1.2 MHz and up.  Transistors for 20W at 1.2 GHz are
$50 and falling.  Amplifiers can be built, but commercial ones
are very costly.  Kits can be a problem, since considerable test
equipment is required to align the amplifier, and mistakes that
destroy the device are costly.

        Jon Bloom, KE3Z: ARRL and Packet

The STA authorizing automatic control of certain stations on HF
has been renewed for another year.  The League hopes to have a
permanent rules change in place by the time it expires.  The
Digital Committee and Membership Services Committee will examine
the results of the testing and propose new rules.

ARRL has established a Technology Grants program, whereby
developers of amateur radio technology can apply for funding from
the League.  This is partially a result of Paul Rinaldo, W4RI,
pushing for improved HF modems and protocols.

Expect a MicroSat article in an upcoming issue of QST.

TNC's are being integrated with the SAREX transceivers in
anticipation of the next ham in space, schedule for STS-35 in
spring of 1990.  The request for prioritization has been
submitted to NASA HQ for the experiment.

        TAPR Business Meeting

President Andy Freeborn, N0CCZ, resumed the chair and conducted
the annual business meeting.  He gave a summary of the year's
activities.  The office complex was closed, and Chris the office
manager now operates the TAPR office from her home.  TAPR had no
booth at Dayton '88, but has requested a double booth in the new
wing for Dayton '89.  A rewrite of the TAPR Bylaws is in
progress.  The DSP Project is now an official joint project of
TAPR and AMSAT.  The two organizations will share the risks and
benefits fully.  Two AT-type computers outfitted for CAD work
were purchased to support hardware design efforts.  They are
presently in use in Tucson and St. Louis.  TAPR has contributed
$21,300 for Pacsat/NA hardware, and is serving as the procuring
agency for all Pacsat hardware purchases.

PSR Editor Scott Loftesness, W3VS, has announced his intention to
retire at the end of the calendar year.  A new volunteer is
sought.  TNC-2 licensing fee income will be ending soon.  A new
project is necessary to provide a source of income.  The Board
has been in continuous session via CompuServe HamNet during the
year; nine votes were taken during that time.  An income
statement and balance sheet were presented.  With an income of
about $61000, the total assets and liabilities of the
organization are $122,500.

A question and answer period followed.  One member stated that
the Board should keep the membership better informed about its
day to day activities, and there was general agreement.  A member
asked about the status of the DSP project.  W3IWI answered that
Bob McGwier, N4HY, has been writing software for the Dalanco-Spry
development system, including several new demodulators.  Since
there is no prototype hardware in the final configuration yet, he
has not begun porting the software to the target system.  W3IWI
guessed that late summer would be earliest possible availability.
The V40 hardware has been designed, but layout has not begun.
The other boards are being fabricated in a prototype version.
They have had problems finding satisfactory CAD software for PWB
layout.  There may be a problem with the V40 chip and interrupts.
It appears that TAPR is standardizing on the V40; this is
important so that we can build up a base of knowledge about its
quirks.  The DSP Project is personnel-limited; more qualified
volunteers would be welcome.

Another member asked if the DSP Project was the only new
development under way.  Freeborn answered that there was another
project in the early design stages: a 9600 baud VHF radio/modem.
This project was mentioned briefly in PSR.  TAPR is wary of
announcing products before they become real, so they try not to
make a lot of noise about a project this early in the design
phase.

Pete Eaton, WB9FLW, led the membership in thanking Andy Freeborn,
N0CCZ, for his outstanding efforts in the past year in
reorganizing the organization.  Applause.

The meeting was adjourned for the day at 5:15 PM.

An additional session was held on Sunday morning, with a single
agenda item:  the no-code license proposal.  Harold Price, NK6K,
led the discussion.  He has been charged, along with other
committee members, with creating a concrete proposal of some sort
to provide to the ARRL committee.  The discussion proceeded in
stages: opening remarks, purposes, and specifics.  An appropriate
background was provided by sounds of hymn-singing from the church
service being held next door.

Your reporter will not attempt to reproduce all the arguments
presented in this 3-hour discussion.  Though the discussion
remained intelligent and civil for the most part, there were
diverging opinions.    The following represents an attempt to
summarize the semi-consensus that evolved during the meeting.

TAPR's purpose in getting involved in the issue in the first
place is to help the ARRL to lead the amateur community to accept
some proposal.  Last time no-code came up, the proposal was
violently opposed by a large fraction of the community.  The
ARRL, being a political organization, will be reluctant to
reverse this position unless it believes there is significant
support for the proposal in the community.  By presenting a
workable proposal and standing behind it, TAPR may be able to
encourage ARRL leadership to try again.

Our purpose in seeking a license that does not require Morse Code
proficiency is threefold:
1. To ensure that crucial spectrum allocations in VHF and up are
utilized and remain available to the Amateur Radio Service,

2. To lure more technically competent people into Amateur Radio, and

3. To lure more young people into Amateur Radio to ensure the future
of the Service.

The specific proposal should have the following characteristics:
o    The existing licensing structure remains as it is.  No privileges
will be given to or taken from any existing licensee.  To do so
would be to invite controversy.

o    An addition licensing structure consisting of two license classes
would be added.  These classes would grant operating privileges
above 30 MHz only.  One license would be fairly easy to obtain,
with a written exam similar in difficulty to the Technician exam.
This license would have very limited privileges.  The licensee
might be prohibited from controlling repeaters, homebrewing
transmitters, or operating high power; and/or the license could
be made nonrenewable.  The license would be limited to selected
bands or subbands.  The other license would be difficult to
obtain, with a written exam more difficult than any now in use.
This license would have full amateur privileges above 30 MHz.

The group was unable to agree on the specifics of the
restrictions to be placed on the beginning license class, or even
on an appropriate name for it, in the time available.  The
Board's committee is charged with fleshing out the proposal based
on the opinions stated during the meeting.  At the end of the
discussion, Freeborn asked how many opinions had been changed by
the arguments.  Several people in the anti-no-code and
uncommitted camps admitted to being swayed toward no-code by the
arguments they had heard.

The TAPR Annual Meeting was adjourned at 12:28 PM.