Gene
Thanks for taking the time to write it up. I'm going to study it and see if
I want and can mod my TR-4CW with RIT..
73
hank wd5jfr
"Gene Gardner" wrote in message
...

Date: Sat, 10 Jan 2004 13:43:36 -0600
Subject: drake-mod

LOW BUDGET - HIGH PERFORMANCE
Upgrade for Older Bargain Transceivers.

One of the main reasons that older transceivers were abandoned was the
need for strict frequency stability. Most amateurs communicating on SSB
today have come to expect transceivers to maintain frequency stability to
within 50 Hz of the desired frequency for natural voice intelligibility
and uniformity
within a group.
Note: This project was apparently of no interest to one of the Ham
magazines so I will submit it for this newsgroup. With such widespread
exposure, perhaps a small percentage will find it of interest. It is
rather lengthy because it provides some theory of the design so that it
can be adapted for other similar equpment.

Several of the older vacuum tube models otherwise provided very good
basic performance.....150 to 300 watts PEP with very good band-pass
filters providing steep-skirted selectivity. Vacuum tubes normally last
many years. At such bargain prices, and with modification, these
transceivers can provide a good interim station for younger amateurs who
have limited money, but plenty of experimental enthusiasm. Many older
amateurs who already have modern transceivers sometimes enjoy working on
side projects as a creative extension of the hobby.

The Drake-TR3 transceiver is described in this article. It is frequently
available on E-bay, and at Hamfests for surprisingly low cost. It (and
probably
Drake-TR4) seem ideal for this project: The same VFO range (4.9 to 5.5
MHz) is used on all bands, and is quite stable even before modification.
It has two high quality sideband filters (USB&LSB), so that no
frequencies have to be changed when switching sidebands. The second
requirement is an old CB transceiver. It does not have to be working,
except for features which are probably still useful: the internal 10.240
MHz crystal oscillator, the progammable divider-phase-locking synthesizer
IC (such as a PLL02A, and several others) and the 5.4 volt regulator.
Either a 23 or 40 channel, and AM or SSB models can be used. The 40
channel provides more channels if you choose the simpler limited option.
Many channels provide suitable divisor numbers, while others will not
without
modification.
The more interesting and more useful option however, is to
use nine-positions of miniature switches to fully control the PLL
counter-lines
for the maximum divider numbers from 1 to 512.
Since several types of transceivers may be of interest, and a variety of
junk box crystals from World War 2 are often available, more detail of
the design
theory will be described.
The fundamental theory of the original CB is this: The internal 10.240
MHz crystal oscillator (10,240 kHz) is internally divided by 1024 in the
PLL IC, which results in a 10 kHz reference. Then the VFO is mixed with a
crystal oscillator to yield a difference frequency, in the 2 to 3 MHz
range. This 2000 to 3000 kHz is divided by 200 to 300 (depending on which
of the CB channels the selector switch is on) to yield 10 kHz. This is
compared to the reference 10 kHz in the phase-detector and produces a DC
control voltage out which changes direction according to whether the VFO
is slightly high, or
slightly low in frequency. This is connected to a vari-cap diode in the
VFO to force it to lock to the correct frequency. A note of interest at
this point: To avoid confusion later when using the Truth Charts
associated with the channel selector and the programming logic, there are
five gaps in the CB frequency range where it advances 20 kHz instead of
10 kHz. Also channels 24 and 25 are inserted ahead (frequency-wise)
between channel 22 and 23 per the EIA numbering plan. Knowing these
exceptions will help avoid confusion when writing down a TRUTH-TABLE for
your model, i.e. following the trend of binary numbers advancing by one
for each channel number, except for those noted above.
Power for the CB and additional components can be conveniently supplied
by half-wave rectifying the 12.6 AC filament supply in the Drake
TR-3. This yields about +17 vdc unloaded with a large filter capacitor. A
series dropping resistor of about 22 ohms (2W) was used to drop the
voltage down to about 12 volts. It is probably wise to get the +12 volts
for the the additional transistors from the load side of the CB power
switch to avoid the +17.5 surge stored in the capacitor.

Getting to the design theory of this Drake TR-3 upgrade: The original CB
10.240 MHz crystal oscillator and its circuits to the PLL-02A synthesizer
are left intact to provide the 10kHz reference. The requirement now is to
take the output of the DrakeVFO (4.9 to 5.5 MHz) and divide it down to
10kHz. These frequencies are too high to divide down directly and are
mixed with a separate crystal oscillator to translate them down to the 2
or 3 MHz range. The CB channel selector switch provides 40 divider values
by switching binary lines HI or LOW. Fundamentally, the PLL02A is capable
of division N=1 to 512, but only 40 divisions are customized for the
original CB requirements and will be modified for use on the transceiver
as perceived by utilizing the Truth Charts of binary logic. Usually the
64 bit (pin 9) is isolated from the channel switch and tied HI for more
options on this model (Kraco KCB-4020).
Pins 7,8,9,10,11,12,13,14,15 (when +5.4v HI), respectively represent
256,128,64,32,16,8,4,2,1 divisors. The effective division is the sum of
all the HI
pins. The 256 and 128 are often wired HI or LOW and the remainder are
switched by the channel selector.

The theory is best explained by taking the three cases used in the Drake
TR-3. It uses an I.F. frequency of 9 MHz. On 80 M, an incoming frequency
of 3875 kHz gets mixed with a VFO frequency required to provide an I.F.
frequency of 9 MHz. Thus 9000 minus 3875 requires the VFO to be at 5125
kHz. This is too high to divide directly, so a rather arbitrary crystal
frequency oscillator of 7425 kHz (a good choice because it allows
coverage of the five older phone bands and has a last digit of 5). It is
mixed with the Drake VFO which has a small coax appearing near an
optional connector at the bottom cover. This yields a difference of 2300
kHz which is divisible by 10 and is sent to the PLL02A. It is programmed
to divide by 230 yielding 10 kHz internally to be phase-compared to the
internal reference 10 kHz. The phase detector provides a positive (but
directional) DC out which connects to a vari-cap or varactor which
changes its capacity according to the DC reverse-bias applied. In this
project, an ordinary Power MosFet was used in a special configuration to
provide better capacitance control. The TR-3 offers an exposed bare wire
going from a tap on the VFO inductor, to pin 7 on V-2. A 47 Pf coupling
capacitor is added from there to the MosFet device as seen in the
schematic.

A second case on 20 Meters is an incoming frequency of 14.320 MHz which
is higher than the I.F. and is mixed with the VFO to provide 9 Mhz.14,320
minus 9000 requires the VFO to be to be 5320 kHz. An alternate crystal in
the vicinity of the first is used, but this time having a last digit of
zero to assure that the mixer output is divisible by ten. 7370 kHz was
used and 7370 minus 5320 yields 2050 kHz. The program pins are set to
divide by 205 to yield the required 10 kHz. Note that alternating between
these two crystals provides lock-in every 5 kHz. Any other favorite
frequency could be accommodated with a different crystal with the
appropriate last digit to provide division by 10. Substituting a good VXO
circuit might be a substitute for changing crystals. It should be noted
that a few PLL-02A's already include the option of grounding pin #4 which
causes it to divide the 10.240 MHz xtal by 2048 and provides an internal
reference of 5 kHz. In this case, either of the added crystal options
above would provide lock every 5 kHz. If your PLL-02A does not have the 5
kHz option (the NTE-1167 does not), it can be done with one section of a
SN74LS74 flip-flop by tying pins 1,4,14 to +5, tie pin 2 to 6, trigger IN
on pin 3, 5120 kHz out from pin 5, and of course pin 7 is ground. A +5
voltage regulator supplied from +12 volts will provide power. The trigger
input to CK, pin 3 may require an NPN 5 volt emitter follower with a 200
ohm emitter resistor to trigger the flip-flop.

The third cases of 40, 15, and 10 M all use the Drake mixer crystal
appropriate for that band. It is mixed with the VFO and the difference
frequency can be considered the "virtual" VFO. On 15 Meters Drake uses a
35.5 MHz crystal. For an incoming frequency of 21.300 MHz, 21,300 plus
9,000 requires a "virtual" VFO of 30,300. The 35,500 crystal minus actual
VFO = "virtual" 30,300, or actual VFO=5200 kHz.
Select the 7370 kHz mixer crystal again, and 7370 minus 5200 yields 2170.
Dividing by 217 yields 10 kHz in the PLL.

It may be preferable for some to have more direct control of the
programming pins on the PLL:
Physically separate pins 9 thru 15 of the PLL from their circuits....by
cuttng the trace to the selector switch with a razor blade. Pins 7 and 8
are more difficult. Then mount a 9-section miniature switch(often found
on old computer boards) within 3 or 4 inches and wire the nine pins 7
through 15 directly to the switches, left to right respectively. One side
of the switches go to +5.4 volts, and the others to the individual pins.
They are SPST that only supply +5.4 for logic HI's. The PLL pins pull
themselves LOW internally.
It is easy to program a divisor number by starting at the most
significant bit (256 on the left-hand switch) and progressively take the
biggest bite permitted toward the desired number...noting the running
tally and the remaining counts required. It's almost fun, and you have
full easy control of counts N= 1 to 512. The final divisor count is
simply the sum of all the HI pins and their values. If you choose this
option, a 23 channel CB has the same benefits as the 40 Channel.
The PLL also has the nice feature that pin 6 goes HI when the
phase-control is locked on (or prevents CB transmitting if it was not
HI). This is wired to an additional small transistor to make an LED glow
when the frequency is in locked on.
The output of the additional mixer is tuned to resonance (2 to 3 MHz
range) and will probably require switching an additional switched-in
capacitor to provide suitable resonance for the entire range (or
re-tuning of a slug). This project used a SPST (center open) to provide
three options: adding no additional capacitance, or two other values for
low, mid, and upper range of the 2 to 3 MHz i.f. range. A test point on
the DC bias of the MosFet variable capacitor gives a useful indication of
the changing DC required to prevent VCO drift, and provides a voltage
reading that allows you to set it to the center of its dynamic range
using a low-cost LCD voltmeter. These voltmeters are usually available at
Hamfests for $10.
There are some CB models that have PLL's not suitable for this project
approach (TC9105P in a Motorola 500 series for example), except that its
10.240 MHz crystal oscillator could be used if a substitute PLL is used.
NTE-1167 seems to be about the only substitute available along with its
data sheet (www.nteinc.com) but at a rather substantial price of
$21.(www.mcmelectronics.com).
This modifcation was tested with the bottom of the TR-3 off and the CB
circuit board and oscillator/mixer circuitry fully exposed. It drove a kW
amplifier with open-wire antenna feeders in the same room, and performs
with the same frequency stablity as modern equipment.
Of course the external crystal oscillator (s) must be exactly on
frequency. Usually small parallel capacitance, or small series
capacitance or inductance, can tweak these on frequency. On 80M and 20M
the only other thing than affects the result is the Drake 9 MHz
oscillator (C-130), but it can only be moved very slightly, because it
must remain centered midway between the two USB-LSB bandpass filters.
(Also
tweaking the trim capacitor in the CB 10.240 MHz crystal has a small
effect). But
40M,15M, and the 3 sections of 10M offer a better option by tweaking
those crystal oscillators L1, L5, and L2 respectively.

Note that when the VFO is not frequency- locked, or if the CB is turned
off, the
TR-3 continues to perform as it did originally except that the dial
accuracy will not track as well. The calibrator and the sliding cursor
pretty much compensate for this problem. The plastic dial scale can be
rotated (to aid calibration) without loosening any set screw as described
in the manual. It is secured only by compression, and can be slipped like
a clutch.

This project could also be an interesting approach to the five new SSB
channels on 60 Meters if the VFO is shunted with capacitance to tune
lower (e.g. 3669.5 MHz vfo added to 5330.5 60M channel yields 9 MHz i.f
frequency). But 3669.5 vfo is too high for the divider to count. A
readily available 6.000MHz (5999.5) crystal can be mixed with the VFO to
yield 2330 MHz which is divided by 233 to yield the 10 KHz reference
frequency required at the phase detector.
Hopefully, the power amplifier tuned circuits are broad enough, without
modification to provide the 50 watts output allowed on 60 meters.

__________________________________________________ ______________