73, de N2HTT

///////////////////////////////////////////
T-R Time Machine

Posted: 24 Jun 2016 07:06 PM PDT
https://n2htt.net/2016/06/24/t-r-time-machine/


This past January, I became aware of a new event celebrating the vintage
rigs popular in the 50s, 60s and 70s, the Novice Rig Roundup. I love the
tube rigs from that era, and absolutely had to put together a station and
join in.

I already had on hand a few transmitters capable of the job: two QRP 2-tube
rigs based on a classic MOPA circuit (you can find links to several on this
page), and a Knight T-60 kit built transmitter. The T-60 was particularly
appropriate, as it was designed to operate at less that 75 watts input,
making it the perfect Novice transmitter.

I operated all of these transmitters under crystal control, and after a
week of trying to make contacts, gained real respect for those ops that
learned the ropes that way. For someone who was licensed well after solid
state transceivers were the norm, operating a separate transmitter and
receiver, constrained to a single frequency, was a real challenge.

Back in the day, ops would call on their crystal frequency, and then listen
up and down the band for a reply. In the modern world of automatic
zero-beating and narrow passband filters, being stuck on one frequency just
doesnt work all that well. After a week of not too much success, I was
determined to add a VFO to my vintage station.Â*Operating with a separate
VFO, in addition to a transmitter and receiver, does complicate things a
bit.

Separate transmitter/receiver stations always involve some kind of
switching arrangement if a single antenna is used for both components.
TheseÂ*can range from a simple toggle switch connecting the antenna to
either, to automatic RF-sensed switching which allows for full break-in
operation. (Full break-in, the ability to hear the receiver between the
elements of the Morse code characters, is what we are used to in modern
transceivers, and there are a number of commercial, kit, and homebrew
solutions to provide this kind of switching for separates.)

Typical switching devices, called T-R (transmit-receive) switches, usually
provide the following functions, in the correct sequence

On key down:

mute receiver
switch antenna to transmitter
key the transmitter
provide sidetone (optional, but handy)


On key up:

un-key the transmitter
switch antenna to receiver
un-mute receiver


Adding the VFO to this mix increases the complexity a bit, because of some
of the characteristics of VFO operation. Unlike a crystal oscillator, which
does not provide a signal until the transmitter is keyed, a separate VFO
can run and be keyed independently of the transmitter. The VFO can run
continuously, or be keyed on and off in synchronization with the
transmitter. Both methods offer advantages and disadvantages.

Allowing the VFO run continuously, and just keying the transmitter, will
undoubtedly produce the most stable oscillation. The VFO will drift less,
and show no start-up instability on each keyed Morse element. However,
there is an odd side-effect to this mode of operation: something that was
known as backwave back in the day. Suppose you hear a signal in your
receiver and want to reply, by tuning your transmit frequency match exactly
(this is called zero-beating the signal.) With the VFO running, you will
hear your oscillator in your receiver (if it is not muted) and the tone,
the backwave, will blot out his signal. You wont hear him.

Also, if you are relying on listening for your transmitter in the receiver
while keying as way of providing a sidetone, the backwave will either
obliterate your keying, or make it sound weird. So all in all, the let it
run all the time approach is not too desirable.

The alternate approach, that of keying the VFO in synch with the
transmitter, is the one that was most often used. In this scenario, the VFO
running only during the times you are actually transmitting a Morse
element sounds ideal, right? Except there is an issue that arises in this
case: chirp. The VFO, when transitioning from off to stable running can
exhibit slewing of the frequency. This bird-whistle effect is called chirp,
and is the hallmark of a poorly run station.

Operators would attach a C to your RST signal report to indicate there was
chirp on your signal, and if it was really bad and you were noticed by an
Official Observer (OO) station, you could be the recipient of a dreaded
pink slip notification to clean up your act. Chirp on your signal is to be
avoided at all costs.Â*Ive had some direct experience with this phenomenon.
I had a Heathkit HW-16 transceiver coupled with a HG-10 VFO. The thing
chirped like a cage of finches. It was one of the reasons I finally sold
the rig.

There are a few things that can be done to minimize or eliminate chirp with
synchronized keying. One is to run the VFO on its own power supply. Another
is to allow the VFO to stabilize before keying the transmitter, and to
un-key the transmitter before the VFO. Some tricky timing is required. This
approach was first explored in tube circuits in the 1950s, and paved the
way for modern transceivers with internal T-R switching.

Living in the 21st century and having access to cheap and efficient
microcomputers, it is easy to shift the keying into the future, by
introducing delays between keying the VFO and the transmitter. This is the
idea behind my VFO-friendly switching system, the T-R Time Machine.
T-R Time Machine, front panel

The external event of closing the code key drives a chain of events that
put the signal on the air lagging the keying by a few milliseconds, but
always keying the transmitter only when the VFO is in stable oscillation.

The sequencing works slightly differently depending on whether you want to
be able to hear signals between the elements of your sending. This mode,
called QSK (one of those Morse code signals meaning I can hear you if you
interrupt), will wind up transiting between transmitting and receiving
several times a second as you send. Non-QSK mode (for want of a better
term,) keys flips the antenna and keys up the VFO, and leaves you in that
state while you send. The station remains in transmit mode for a fixed hang
time after the last Morse element sent, before flipping back to receiving.
The T-R Time Machine can manage either mode.

In non-QSK mode, the T-R Time Machine starts the VFO, delays a couple of
milliseconds, and then starts keying the transmitter in time with the
external keying. The VFO runs continuously, but the receiver is muted so
you do not hear the backwave. After the hang time has elapsed since the
last keyed Morse element, the station is sequenced back to receiving. The
hang time is set to be just a bit longer than the typical pause between
phrases, to minimize the amount of switching.

The first morse element is robbed of a millisecond or so of duration, but
at 25Â*WPM that amounts to about 3% shorter, and only the first transmitted
element is affected. It is completely unnoticeable on the air.

QSK mode time-shifts your keying by a couple of milliseconds, so the keying
the transmitter lags the keying input. On key up, the VFO runs for a
millisecond or so past the unkeying of the transmitter. The effect is to
slow the output keying very slightly, but again it is unnoticeable on the
air.

My Arduino sketch re-uses some code I had written for the Digital Fist
Recorder project, but adds a new concept to control the keying: a coding
device called a Finite State Machine. Using this approach, you model the
process as a series of states. Each state performs some logic on entry, and
based on current conditions, decides how to transition to the next state.
Organizing the code this way makes it very easy to maintain. The switching
tasks, things like key down, VFOÂ*start, transmitter start, etc. lend
themselves nicely to representation in the code as discrete states, and the
resulting code is quite clear to follow.

I have not yet posted the code to GitHub, but intend to do so, and will
post an update here when it is availble. As always I am releasing this code
as open source under a GPL license.

Building the TRTMÂ*naturally divided into to two phases: the easy part and
the hard part.

The easy part was building the boards containing the Arduino and switching
components. The TRTM consists of several pre-assembled components:

An Arduino: I used a standard Uno R3 board, but also had implementations
running on a Pro Micro clone

(the sketch for TRTM is small, and will run on just about any Arduino board)
A dual relay board
Two Key-All HV boards

T-R Time Machine, interior view

In addition, there are two voltage regulators which I built using ICs and
Manhattan style assembly. One regulator drops the input voltage to a
regulated 9 VDC to run the Arduino. The second provides a regulated 5 VDC
to run the relay board and the Key-All modules. In early prototypes of the
TRTM I ran into difficulties sourcing enough current from the Arduino to
run the outboard switching components; this version sidesteps the issue.

The hard part for this project was fitting all the components, and the
large number of external connections into an enclosure. I went through
three iterations before settling on the version shown here. The project
required:

three antenna jacks,
four phono jacks,
external power pole mounting,
four indicator LEDs,
a speakerÂ*(the speaker is mounted in a hole on the bottom of the enclosure,
under the main board and fires down),
a panel mounted pot for sidetone volume,
two switches, one for power and one to select QSK mode are also on the
front panel,
as well as two input jacks for a key. The two jacks are wired in parallel,
and provide either 1/4 mono or 1/8 stereo plug inputs, since I have keys
wired both ways,


Drilling all those holes in the right place proved challenging (you can see
the extra holes in the boards, and on the bottom of the enclosure where I
had to move things around.)

An additional feature inÂ*the latest enclosure is the complete isolation of
the switching circuit from the Arduino, to avoid RF feeding back into the
computer and causing issues. I accomplished this by mounting all of the
switching connections (the antenna and phono jacks) on a panel made of
polycarbonate plastic. This material is inexpensive, and easily machined
with ordinary hand tools.

The plastic plate is mounted on the back panel of the aluminum enclosure
through oversized holes, so the switched grounds are completely isolated
from the enclosure ground. Actual switching isolation is provided by the
use of relays, and the Key-All units which are opto-isolated from the
control circuits.
T-R Time Machine back panel

This project was difficult to complete. In additionÂ*to all of this drilling
and screwing was the difficulty I had getting the LEDs to work. Yes, I was
not able to light an LED using an Arduino

It plagued me for weeks, until I finally sat down and carefully stepped
through the code to discover that I had never initialized the digital pins
used for the LEDs as output pins in my sketch. Silly code bug, but it
really drove me nuts.
T-R Time Machine with working LED indicators

So I finally have a very nice, working, VFO-enabled T-R sequencer ready to
go. Unfortunately, I still havent gotten a VFO to work properly with my
Knight T-60, but that will be a story forÂ*another day. Until then,

73

de N2HTT