surplus 1-watt TBQ-3018 VSAT amplifier module for 10 GHz
amateur band operation
Figure 1: Top: Typical installation of an outdoor
unit. A feedhorn is attached and the unit is mounted in
a carrier that permits rotational adjustment of
polarity. (Yes, those are 10-meter dishes in the
background...) Bottom: Examples of two of the outdoor units
from which the amplifier modules were obtained. The
feedhorn and "OMT" (Ortho-Mode Transducer) have been removed
on these units as they may have been re-used. The upper
unit had been deployed near the ocean and was mechanically
damaged: Its amplifier module was the source of the one
in the pictures below. Click on a picture for a larger version.
the very amplifier shown on this page, the K7RJ 10 GHz
beacon was put online on 11 September, 2011 from grid
square DN31it operating on a frequency of 10368.250 +/-
500Hz. Located approximately 110 miles northwest
of Salt Lake City, it has been heard by a number of
stations in the Salt Lake area and has been shown to be
a reliable "weak signal" source.
This beacon uses Frequency-Shift CW (FSCW) with the
beacon message consisting of an ID and grid square
locator followed by a 30-second carrier at the
"key-down" frequency. This beacon operates at a
power level of 1.0 watts feeding a
horizontally-polarized omnidirectional slot antenna and
the frequency stability is such that it is possible to
use "waterfall" display programs (such as Spectran or
Spectrum Lab) to detect it at sub-audible levels.
This page is a work in progress and it will be updated
as time permits.
Added diagram for amplifier bias control circuit and 8 volt
switching power supply (8/19/2011) - See below.
Working on the Satcom industry, I noticed a growing pile dead of
Hughes "Tigris" ODU (OutDoor Unit) modules in the bone yard.
Made from the late 1990's into the early 2000's, these were
various returns from the field and were, for various reasons,
declared dead or not economically unrepairable. Since it
will often cost about as much to repair a unit as to replace it,
"dead" units are often put in bone yard and used as a source for
mechanical parts to repair other units.
The typical failure modes of these units may be grouped into two
Environmental. Being outside, they are expected
to weather the elements over time, but inevitably mother nature
takes its toll. In this category there are two types of
Moisture ingress. Typically, moisture gets in
through one of the unit's RF connectors (typically F-type) due
to either improper/inadequate weatherproofing at the time of
installation, or the gradual breakdown of the weatherproofing
material and as you might expect, if water gets into a
connector and combines with voltage, corrosion ensues.
Sometimes it is possible to replace just the connector, but
occasionally, it gets farther than that and damage to the
Mechanical damage due to corrosion. Being cast
aluminum, these units are more subject to degradation when
installed in an environment that contains salt - such as near
the coast. Inevitably, this attacks the aluminum and
weather seals get compromised and/or parts of the casting
corrode and/or are broken when one attempts to remove
screws. At that point, the unit is mechanically
considered beyond repair although the electronics inside may
still be perfectly serviceable.
Lightning/surge. Although they are
well-protected against transients, etc. Lightning or other
sources of transients (due to ground loops, open shields,
etc.) will occasionally blow up the circuitry associated with
the input terminal (IF amps, regulators, etc.) of these units.
Over the years it was noted that it's fairly rare that failure the
component of interest in this case - the output power module - is
actually the reason that the unit was pulled from service.
Being buried deep inside the unit, it seems to be fairly
well-protected against electrical impulses and it is also unlikely
that even some moisture ingress and related corrosion would
be severe enough that the amplifier module and the circuitry
surrounding it would be damaged to the point of being unusable.
As can be seen from the top picture in Figure 1, these units
are typically mounted in a bracket that allows the polarity to be
rotated to match that of the satellite at the earth station's
location. A circular feedhorn is typically used to illuminate
an offset-fed antenna that is typically around 1 meter in diameter
and this is coupled to an "OMT" (Ortho-Mode Transducer) that
provides the diplexing between receive and transmit as well as
separating signals from the two polarities.
The bottom picture of Figure 1 shows two such units having
been removed from their mounting brackets along with the removal of
the OMT. The upper unit - having been deployed near an ocean
coast - shows severe corrosion and some physical damage: As
the aluminum oxidized, it expanded and popped the waveguide flange
apart and further damage was done when the screws holding waveguide
feedhorn had to be chiseled off! The lower unit fared much
better, not having been exposed to such a corrosive environment and
was likely pulled from service to do an electrical problem.
It is quite likely that a unit in condition similar to the upper (and
more badly-corroded) will will yield usable parts for our
application as it was probably not damaged by an
impulse or other electrical mishap, but failed due to moisture
ingress and/or mechanical failure - and the the board inside was
likely (mostly) in good shape as the unit's ultimate failure
resulted in immediate replacement. It turned out that both
units contained good amplifier modules and the upper unit showed
only minor evidence of moisture on the board.
There are 2 watt - and maybe some 0.5 watt - units, but the
majority of the units of this style are likely to be the 1-watt
The prototype amplifier that was tested and described below
used components from the badly-corroded unit shown in Figure 1.
A brief description of the circuit:
In a nutshell, here is how these ODU modules work:
The signals from the indoor unit are conveyed via either 50 or
75 ohm coax (it was designed to accept either). Typically,
the older versions had "N"-type connectors while the newer units
used "F" connectors.
The unit is powered by 19 volts DC (nominal) sent up the
coax. A portion of this is sent to the LNB while other
voltages are produced inside the unit to operate the various
For receive, an external LNB converts the 11.7-12.2 GHz
signals from the satellite to the standard 950-1450 MHz
L-band using a 10.75 GHz local oscillator. These signals
are diplexed inside the unit onto the coaxial cable to be sent
back to the indoor unit.
A VCO running at 1/2 the transmit frequency (7.0-7.25 GHz) -
locked to a 109.375-113.28125 MHz signal at 1/64th of
that frequency coming up the coax from the indoor unit.
The output of the VCO is frequency-doubled and amplified to
the 14.0-14.5 GHz transmit frequency.
This transmitter is modulated using OQPSK - a version of QPSK
that does not require any sort of amplitude component to be
present and can be modulated solely via a special sort of
frequency modulation (e.g. FSK). This modulation is
imposed on the "reference" frequency (at 1/128th
of the ultimate transmit frequency) and the PLL,
tracking this modulation, conveys it to the output
frequency. (Yes, the index of the modulation in the
109-113 MHz range is very low!)
A 1 MHz signal is modulated with a 1 kHz square wave that has
a varying duty cycle that is used to control the transmit
The duty cycle of the 1 kHz component varies from 0% (e.g.
complete absence of this signal) representing full power to
approximately 66%, representing "minimal" (but not "zero")
A 10 MHz on/off signal is used to enable/disable (e.g. "key")
the transmitter. (10 MHz = transmit.)
As a bare minimum, the 10 MHz and 109-113 MHz signal must be
present to enable transmit.
For our use we'll cut away a section of the circuit board and the
portion that we need to extract contains two important pieces:
The TBQ-3018 GaAsFET MMIC, described below.
A low-level MMIC driver stage. This was part of the
original circuitry, but since it is located very close to the
power module itself, it is easy to retain and make use of
it. Marked simply with "C8", this device is an RF2048 made
by RF Micro Devices. While the '2048 is characterized only
to 8 GHz, like many other MMICs it has useful power and gain
well above its nominal operating frequency range. This
device is typically biased to 40mA resulting in 3.6 volts or so
across the device. (The lead with the diagonal cut is the
input. A data sheet for this MMIC may be found at
To complete the parts count we need only slice the trace on the
input lead of the input MMIC and install a DC blocking
capacitor: This capacitor may be found on the remainder
("discarded portion") of the circuit board having been used to bock
the DC for this very MMIC!
Preparing the amplifier module for use:
For the units that I've investigated the output power module is a
Teledyne TBQ-3018 and it is for these units that this
description is targeted.
Th TBQ-3018 is a 1-watt (nominal) GaAsFET MMIC module that is
intended for operation in the Ku uplink frequency band of 14.0-14.5
GHz. In digging around on the web, I was able to find a data
sheet for this module and careful inspection revealed something
interesting: Although most of the specifications showed its
use in the 13-15 GHz range, one chart in particular showed the 1dB
compression output over a range of 11 to 18 GHz, and the power
output at 11 GHz was still a respectable +27.5dBm, or about 560
Having several of these units to mess with, I decided to try my hand
at getting one to work in the 10 GHz amateur band, so here is a
general description of the procedure:
Figure 2: Top: The entire circuit board from the outdoor
unit. The section containing the TBQ-3018 power
amplifier is in the lower-left corner of this picture, having
already been cut from the main board. Bottom: The small foil shim that was under the
power module. This piece should be carefully saved as it
is used both for thermal conductivity and RF grounding.
The "tab" on the side opposite the mounting holes goes under
the portion of the board with the output RF connector. Click on a picture for a larger version.
Set up your work in a static-controlled area. At
the very least, use a large cookie-sheet or large, shallow
baking pan to contain the parts connected to ground.
Before picking up any part or setting a part back down, touch
the sheet or other grounded object to discharge any accumulated
static - particularly if you have walked across a room or
haven't touched the sheet for a while.
Disassemble the unit.
There are some hex-head screws that hold the cover in place
and these should be removed. It is not too uncommon for
these to snap off, particularly if the unit had been used in a
coastal area and the threads have galled. After undoing
the screws, the cover will lift off, although a bit of prying
may be necessary.
Using a T-10 "Torx" (tm) bit, remove the die-cast aluminum
shield that covers a portion of the circuit board and save
it. Carefully note if any RF absorbing foam sticks to
the circuit board or its components and put it back in the
Remove the two screws holding the RF power module into
place. These are usually two small "Philips" head
Remove the rest of the screws holding the circuit board in
place. These are also T-10 screws. Note that there
are also screws on the voltage regulators as well.
Unsolder the RF connector from the end of the board opposite
the waveguide flange. This could be either a type-N or
an "F" connector. Often, the connector is "stuck" and
cannot be removed, so the long center pin will have to be
either clipped off or bent out of the way.
Note that there are two "F" connectors on case on the
backside of the circuit board. The center pins of these
connectors are soldered to the circuit board.
Lift the board out while applying heat to to the center pins
of the two "F" connectors, one-at-a-time. It is usually
easier to lift up the edge of the board opposite the "F"
connectors and then heat one of the pins, and then the other.
Once the board is out, remove the waveguide probe pin near
the power module. Clear solder from the hole so that the
center pin of the SMA connector may be inserted.
Note that there is a small piece of metal foil that was
mounted underneath the power module: Save this as it is
useful for heat transfer as well as assuring a solid RF ground
between the flange of the RF power module and the bottom of
the circuit board itself.
Cut away all but the portion of the circuit board
containing the output power module.
The first step was to extricate the module and this was done
by cutting away the section of circuit board containing the
power amplifier away from the rest of the board using a
hacksaw. When doing this, take care to minimize the
likelihood of static buildup. Make sure that you leave
about 1/8"-1/4" (3-6mm) from the gold-plated portion.
Keep in mind the amplifier module itself is mounted on
rather fragile leads and that care should be taken to avoid
As noted above, I'd first removed the output waveguide probe
and cleaned the hole of solder.
Clean up the circuit board section: The edges of
the board were then "cleaned up" using a fine-tooth file to
remove the "raggedness" from the sawing.
Cut away a portion of the shield: The next step
was to saw off a portion of the cast aluminum shield covering
the unit - see the top picture in Figure 3.
Fortunately, this is fairly easy as the needed portion of the
shield can be easily separated at a narrow section.
The shield is very important: In
testing on the bench, the unit was operated without a shield,
but it was extremely unstable and tended to break into
self-oscillation. The use of the shield and its
contained RF absorbing material allowed the unit to be
A suitable enclosure was chosen. I chose a small
die-cast aluminum box as it would also serve as a heat
sink. With its backside being flat instead of finned, this
facilitated the installation of flush-mount SMA
connectors. If you have access to a metal-working tools,
it may be possible to cut and mill portions of the original case
in which the amplifier was mounted or to cut out areas among the
fins of a standard heat sink to clear the screws and RF
Mark the holes in your amplifier box. Very
carefully, I laid the extricated piece of circuit board in the
box and marked where the mounting holes needed to be
drilled. The former position of the waveguide mounting
hole was noted as was marked the position where the input
connector would be placed.
Drill holes in the box. Holes were drilled in
the die-cast box in the marked positions to accommodate -
including the holes where the input connector would be.
Select and mount the RF connectors. For RF input
and output connections, I used "3-hole" SMA connectors:
These are the types with a 2-hole mounting flange and the 3rd
hole being the center conductor itself. The case was
marked and drilled to accommodate mounting of these connectors
as well. These connectors may be seen in Figure 3.
"Single-hole" chassis-mount connectors may not be suitable
for this project and "5-hole" units - if usable - would have
to be carefully mounted as to have their screws avoid
interference with the shield cover and the circuit board
traces - or just two of the four mounting screws used. If
larger holes are used in the mounting box it may be possible
to solder the connector directly to the bottom (ground plane)
of the circuit board, but note that the foil on these boards
is quite fragile and connectors mounted in this way could
easily be torn off when tightening the SMA fitting!
Check and drill holes in the board for the antenna RF
connectors. The board was then "dry-fitted" and the
holes for the input SMA connectors were marked on the bottom of
the circuit board. Corresponding holes were then drilled
in the bottom of the board to accommodate the connectors'
Re-check the fit and drill more holes. Without
the fragile circuit board in place, the die-cast shield was
installed and held in place with the already-drilled mounting
screws and holes.
At this point additional holes were drilled where the two
SMA connectors would be mounted, through the cast shield
As necessary, filing was done with a set of "needle files"
to make sure that the screws fit well and were aligned
properly - the idea being that it should be possible to insert
and tighten the hardware without needing to shift the board
Note that one of the mounting screws does not go
through the shield cover and is used to hold the board down
before attaching that cover.
Near the output connector, it may be necessary to file off a
ridge on the top of the die-cast shield to provide a flat
space for the nut for the screw that holds the output SMA
connector in place.
Check the mounting of the RF connectors. Make
sure that the SMA connectors will mount to the die-cast box
without the center pin shorting to the bottom foil of the
circuit board or putting mechanical stress on it.
Add holes in the shield for power/bias connections. Small
the shield cover to permit the connection of the MMIC's bias and
Add a groove to allow connection for the MMIC's
power. A groove was filed into the shield cover
near the driver MMIC to allow a piece of wire to be passed
between it and the circuit board for powering the MMIC.
Disconnect the power supply traces from the bottom side of
the circuit board. On the bottom side of the
circuit board the traces connecting to the drain and bias leads
for the power amplifier module were identified. Also
identified was the power supply trace to the input MMIC
amplifier. (Pictures to be added.)
Using a small drill bit from the bottom side of the board,
the circuit board "via" leading to the top side was drilled
out. It's not necessary to drill completely through the
board to do this.
The idea here is to disconnect this via from the
trace on the bottom side of the board as the board itself will
be bolted to the die-cast box which would otherwise short out
those connections. Usually, it is also easy to pull up
the trace and remove it as well although this won't disconnect
the via, either. (It may be possible to pull up the
trace to the via and then drill a countersink into the
die-cast box to prevent shorting of the via.)
Figure 3: Top: The lower casting is the unmodified RF
shield cover while the upper shows which portion was
removed. Be sure to save any pieces of RF absorbing
material in the shield castings.. Center: A modified, extracted amplifier board
section with the "input" being at the bottom-left. Extra
holes have been drilled to accommodate the 3-hole SMA
connectors. Also note the cut in the trace on the left
side in preparation for a DC blocking capacitor to be added
for the input predriver MMIC. The above picture shows an
early test unit on which I'd broken some of the leads of the
TBQ-3018 power module and resoldered - a reminder that the
leads are VERY fragile! Bottom: Annotated picture showing the various
connection points. Click on a picture for a larger version.
Clean up the board. Clean up and de-burr the
holes drilled into the circuit board as necessary.
Install the input blocking capacitor. Somewhere
along the trace between the cut-off portion of the board where
the input connector will be mounted and the input predriver
MMIC, cut a slot in the trace about 1/8" (or 3-4mm) wide.
Remove the chip capacitor that is still on the remaining main
board that had been on the input of this same MMIC and place it
over the slot in the trace. This is done to provide DC
blocking for the predriver MMIC. This cut may be seen
along the left side of the bottom images of Figure 3.
Connect the power/bias wires. On the top of the
board, small pieces (about 4" or 10cm) of wire are soldered to
the supply points of the output module and the input MMIC as
noted in the bottom image of Figure 3. I used #30
wire-wrap wire for this. These make the bias and drain
connections for the output amplifier and the bias supply for the
predriver MMIC.The drain supply for the power
module connects to a small bypass capacitor while the bias
supply connects at a 350 ohms chip resistor. The supply
for the predriver MMIC connects at the junction of the chip
resistor and its bypass capacitor.
Clean the inside of the shield cover. Inside the
cast shield cover are small pieces of RF-absorbing foam that are
very important to the stability of the amplifier. When
drilling and filing, these pieces of foam and the adhesive
holding them into place tends to cause small flakes of metal to
accumulate: Very carefully remove these small pieces of
metal without removing/losing the pieces of RF absorbing
Bolt the power module and foil shim into place. Locate
foil that had been placed under the RF power module (see the
bottom picture of Figure 2) and put it into place
where the power module will go in the die-cast box. The
"tab" on this piece of foil sits under output terminal of the
amplifier module where the RF output probe used to be.
Using #2 or #3 hardware (very small screws) loosely attach the
output power module and board to the die-cast box with the piece
of foil underneath it.
Re-check hardware/hole alignment! Check the
alignment of the mounting holes and adjust as necessary.
Be prepared to remove the board again and file the die-cast box
(and drilled holes) as necessary to assure a somewhat loose fit
of the hardware.
It is important that one prevents the board from shifting
once the power module is bolted down and putting stress on the
leads. As can be seen from Figure 3, I managed
to break the leads of the device by moving it around too much
while trying to mount the board!
Start installing screws and re-check alignment. With
alignment, install the one screw that does not go
through the shield casting and tighten it: This is the
screw that goes into the hole closest (and just to the left) of
the input driver MMIC as seen in the bottom picture of Figure
3. Once the alignment holes have been re-checked and
it is verified that stress won't be placed on its leads, tighten
the two screws holding the output module down.
Install/solder the RF connectors. Temporarily
install the screws holding the two SMA connectors and solder the
center pins into place on the circuit board. Once this is
done, remove the screws holding the SMA connectors.
Install the shield cover. Install the die-cast
shield cover, carefully threading the wires connected to the
bias and drain of the power module through the holes drilled
into the shield. For the pre-driver MMIC, make sure that
the wire is routed in the groove that was filed into the casting
and not pinched.
Fit the shield cover into place. Loosely install
screws into the original mounting holes in the casting - but do
not install the screws for the SMA connectors just yet.
After verifying that everything is in alignment and that no
wires are pinched, tighten the screws that are used to hold the
shield in place.
Tighten down the screws for the RF connector. Now,
the screws that hold the SMA connectors in place.
It is important that this be the last step as these
"non-original" holes may be slightly out of alignment with
those originally in the shield casting. Because these
holes need to follow the contour of the original casting, the
drill bit may have followed something other than the intended
course and knocked them out of alignment. It may be
necessary to drill larger holes and/or do some filing to get
these screws to fit nicely.
Note: Tightening the screws holding the RF connectors
into place while the other screws are still loose could shift
the board and break the wires connecting the output power
Install bypass capacitors on the power/bias leads. On
electrolytic capacitors were installed to provide low-frequency
stability. For the drain lead, a 100uF unit was used while
a 10uF capacitors was used for the predriver MMIC's supply and a
0.1uF was used for bypassing of the bias lead. The values
aren't critical, but I wouldn't recommend less than 4.7uF or
more than 100uF for the power leads. It is recommended
that low-ESR capacitors rated for 105C (high temperature) be
used. A 100 ohm, 1/4 watt resistor is used to drop the 8
volt drain supply to the proper voltage/current for the MMIC
input stage (e.g. approx. 40mA with a MMIC terminal voltage
of approx. 3.6 volts.)
Do an ohmmeter check.
It should be noted that when referenced to ground, an
ohmmeter test on the gate bias lead will show the presence of
the 350 ohm chip resistor.
The drain lead will show a resistance of an ohm or so with
zero gate bias, so it is difficult to rely on this reading to
indicate a shorted device or connecting wire unless you have
an ohmmeter that will resolve very low resistance and
you take into account the resistance of the connecting wire
Before we apply power, there are a few precautions to be taken:
The output power amplifier MUST have a source
of negative bias voltage - around 1 volt - in order to function
properly, and the data sheet warns against applying drain
voltage without the proper amount of negative bias! For
initial power-up, this bias voltage needs to be set to about 1.5
volts to guarantee complete cut-off the drain current. In
testing, this unit showed around -0.5 volts for a
drain current of 600 mA, but note that this will vary with
temperature and from unit-to-unit.
Also, the drain voltage - nominally 8 volts - MUST
be current-limited for initial testing. For this, an LM317
current source can be constructed to limit the maximum current
to 0.6-0.8 amps, or if you have one, a current-limited power
supply may be used. Once proper operation has been
verified, the drain current may be increased as necessary.
As noted above, the input MMIC is operated from the 8 volt
drain supply: A 100 ohm 1/4 watt resistor is used to drop
Do NOT use a power supply that will deliver more than
600-800 mA for initial testing! YOU HAVE BEEN WARNED!
Allowing the drain current to exceed 1 amp -
particularly if the bias voltage is too low - can instantly
destroy the amplifier module!
(These points are important enough that I thought that I'd say
them more than once...)
Figure 3: Top: The repackaged TBQ-3018 VSAT amplifier
under test on the workbench. Center: The amplifier installed in the die-cast
box with the shield covers in place. Notice the wires
for the drain and gate bias emerging from holes in the shield
casting. Bottom: The "other" side of the amplifier,
showing the "3-hole" SMA connectors and the screws holding the
board, connectors and shield covers in place. Since
these pictures were taken, feedthrough capacitors have been
added and holes drilled and tapped to allow mounting of the
amplifier to a heat-spreading plate. Click on a picture for a larger version.
The recommended procedure is this:
Connect a load suitable for 10 GHz to the output.
Typically, one would use an attenuator between the output
amplifier and the microwave power meter, but remember that there
may be more than 1 watt of RF present, so pick a load/attenuator
that can handle the power.
Apply the negative bias voltage first!
this will pinch off the drain current. Take care not to
exceed -1.5 volts. Note that on the circuit board is a
350 ohm resistor that loads the DC bias line.
Next, apply 8 volts of current-limited drain voltage.
monitors both the drain voltage and drain current
If you are using a voltage regulator, it is best to apply
voltage to the INPUT side of that supply (or regulator)
first. The reason for this is that such a supply will
likely have a filter capacitor on itsoutput
and if, for some reason, the bias is gate bias is too low,
that output capacitor may discharge across the power module,
exceed its ratings and destroy it instantly! Applying
the voltage to the input of the drain supply/regulator will
help assure a "soft start" and avoid such momentary
high-current pulses. The 100uF capacitor on the drain
line noted above will help protect the device, but be careful,
Slowly decrease the bias supply while noting the drain
current. Rather suddenly, the drain current will
start to rise: This typically occurs at approximately -1.0
Adjust the bias voltage for about 0.6 amps of drain current
for initial testing. Note that with a
current-limited power supply, further decrease in bias voltage
or even additional increases in drive will cause the amplifier
to attempt to draw more current, pulling down the drain voltage.
Now, connect the predriver MMIC's power lead to a voltage
source. For testing, one may use a 300 ohm
resistor in series and a +12 volt supply for the MMIC to limit
the current to the 30-40mA range, or one may put a 100 ohm
resistor in series with the 8-volt supply. The actual
voltage on the MMIC is unimportant - just select it (and a
series resistor) for approx. 40mA of current. If you do
use the current-limited 8-volt supply to power the MMIC stage,
keep in mind that if current-limiting occurs, the MMIC's supply
current will also decrease along with the RF drive to the module
- but this is probably not a bad thing!
You may now run RF tests!
When powering down, remember to remove the RF drive and drain
voltage before you remove the bias voltage!
Observations and comments:
The data sheet is a bit vague and contradictory on the
device's maximum ratings. For example:
It shows the maximum input power as being +1dBm, yet one of
the graphs implies test data to +10dBm.
The maximum drain current isn't specified, but "typical"
values of 650mA under test conditions are given while the
various curves show currents as high as just over 1 amp.
Remember that when the board is "hacked" as described, a
"pre-driver" MMIC is present on the chunk of circuit board, next
the the output module. Again, this MMIC has yet to be
positively identified so it is unknown how much gain this device
has, but at 6-8dB minimum gain would be a reasonable
The gain of the TBQ-3018 amplifier module alone is stated as
being in the area of 30dB at the nominal 14 GHz frequency.
Down at 10.368 GHz, however, its gain is likely to be much
lower. The "guesstimate" that it is around 20dB in the 3cm
The measured gain of the combination of the input predriver
MMIC and the TBQ-3018 was measured as being
approximately 34dB at the 1dB compression point at 10.368 GHz.
When run at or below the 1dB compression point, about -5 dBm
is required to drive the combination of the predriver MMIC and
output amplifier. At 8 volts and 600mA, this power level
correlates to a bit over +27dBm (500mW) at 10.368 GHz.
The amplifier was also tested at 1 amp of drain current at a
maximum of 8 volts. At full saturation and with careful
adjustment of input bias, over +32dBm (about 1.6 watts) was
achieved at 10.368 GHz. This power level was maintained
for only about 10 minutes as the small die-cast box and the 20dB
attenuator being used between the amplifier and the power meter
were getting very warm! There was no apparent harm
observed. (This test was done just prior to the
18-hour "burn-in" at +30dBm.)\
The unit was run with the drain current limited to 800 mA at
and output power of +30dBm (1 watt) for about 18
hours with the die-cast box partially covered to increase
temperature and further-stress the amplifier. The power
remained stable to within a few 10ths of a dB during
the entire test period. The required input power level was
approximately -2dBm at 10.368 GHz.
As noted below, the unit picture in Figure 3 has been installed in the K7RJ 10 GHz
beacon. It has operated for several weeks, continuous
duty, will running at an output power of 1 watts which, when
final installation and integration with the bias/drain current
controller circuit, correlated with a drain current of roughly
750 mA at 8 volts.
When driven into saturation, the bias voltage may be "tweaked"
to achieve maximum power output, but it is unknown how stable
this sort of adjustment might be over wide temperature
excursions. Again, with a current-limited supply, maximum
saturated power may occur with the drain voltage being "folded
back" to less than 8 volts.
While I would not recommend trying it, the unit
has been proven to be quite forgiving of poor output termination
(e.g. no load) and has shown no signs of instability or damage.
Using the amplifier - the K7RJ 10 GHz beacon:
One ongoing project is nearing completion - the K7RJ 10 GHz
beacon. This beacon will ultimately be placed at some property
that Ron, K7RJ, owns that is located in the (remote!) northwest
corner of Utah (grid square DN31) at a distance of 120-ish miles
from Salt Lake. Because it is line-of-sight to the mountains
near Salt Lake City it should provide a weak signal source that
should also demonstrate various weather-related propagation
effects. (It should be
noted that contacts between Salt Lake and this location have
already been made on 10 GHz.)
This beacon is fairly simple: An OCXO (Oven-Controlled Crystal
Oscillator) is located - along with the FSK-CW keying circuit -
indoors, with a length of RG-6 coax to convey the 96-ish MHz signal
and multi-conductor cable for both power and monitoring to a
weathertight box located outdoors, at the antenna - an
omnidirectional waveguide slot. In this box is a "Brick"
oscillator locked to the OCXO's output is then fed to the amplifier
shown in Figure 3 which is set
for an output power level of +30dBm (1 watt) and a frequency of
approximately 10368.249 MHz.
It was for this project that I constructed two circuits: An
8.2 volt switching regulator, and a bias control and
8.2 Volt switching regulator:
Because the amplifier is located remotely, a 24 volt AC supply is
used to minimize I/R drop and since an 8.2 volt DC supply at just
under an amp was needed, a simple switching voltage converter was
constructed to minimize both power consumption and heat generation
within the confines of this box. This circuit is based on the
National Semiconductor LM2575-5 step-down converter which requires
only a few extra components and because of its efficiency (in the
85-90% area) it produces relatively little heat and requires no
heat-sinking other than that afforded by its being attached to a
mounting plate inside the box. The circuit depicted in Figure 4 requires that only low-ESR
electrolytic capacitors are used throughout and it also includes
additional L/C filtering on both the input and output lines to quash
conducted 52 kHz energy from the switching supply to prevent its
modulating the 10 GHz signal as well as being conducted outside the
box and interfering with HF operations that one might conduct
The only "critical" component - aside from the electrolytic
capacitors - are the inductors: L302 should be rated for
several amps and can be anything from 100uH to 470uH, but something
in the general area of the 330uH specified is recommended.
Because they are used for filtering, the precise values of L301 and
L303 are unimportant and I simply used some toroidal chokes plucked
from some scrapped switching power supplies: Anything from
47uH and up may be used as long as it can handle at least an amp
with minimal voltage drop - particularly L303.
YouTube video showing the K7RJ 10
GHz beacon using the amplifier described on this page.
The LM2575-5 regulator was used rather than the "LM2575-ADJ" since
the former was on hand and the latter was not, but the "ADJ" version
could have been used if the values of R301 and R302 were
appropriately selected for the desired output voltage or, in either
case, if R301/R302 were replaced with a potentiometer in the 1k to
5k range connected between the output at C303/L302 and the wiper
connected to the "FB" pin. Because of the slight voltage drop
in the amplifier bias control and current limit circuit described
below and interconnecting wiring, the voltage from this regulator
was set slightly high - to 8.2 volts - so that under load, close to
8.0 volts was obtained at the amplifier itself.
In the K7RJ beacon, the main power supply was from a 24 volt AC
transformer which was half-wave rectified twice: Once to
provide the negative
voltage for the "brick" oscillator, and again to provide the positive
supply for this switching regulator and the controller board,
below. The 8 volt switching regulator an the 7812 regulator on
the controller board (plus another switching regulator on the
negative supply bus used to power the "brick) are capable of
operating from the 28-32 volt DC bus, this 8 volt switching
regulator is perfectly capable of running from as low as 11 volts.
When initially testing the beacon as a whole it was noted that heat
buildup in the outdoor box was the main technical problem to be
solved! Enclosed in the weatherproof, fiberglass enclosure, a
temperature rise of over 70 degrees F could be expected even with
the use of the switching regulator, and if a linear regulator had
been used, the heat problem would have been much worse! To
keep the electronics inside at a reasonable temperature, passive
thermal transfer technique were employed to maintain safe levels -
especially on sunny, 100+ F days! Another tactic was to
construct a sun shield for the box using a small solar panel which
is then used to power a fan that moves air through the box.
Amplifier bias control and
current limiting circuit:
The TBQ-3018 amplifier chip requires a bit of external bias control
and monitoring circuit to keep it "happy" - namely, to maintain the
drain current at a safe and consistent level under varying
conditions. To do this, the circuit needs to monitor the drain
current and dynamically adjust the bias voltage to maintain that
current. An additional feature is a current limiter that
offers a bit of extra protection to the amplifier in the event that
the bias line be accidentally disconnected or shorted out during
testing. Two supply voltage are required: An 8.2 volt
supply for the amplifier module itself, and a higher voltage for the
op-amp and voltage converter - nominally 12 volts. In this
particular case - since the main V+ supply is in the 28-32 volt
area, a 12 volt 3-terminal regulator (a 7812) is used to provide the
lower voltage while a 5 volt regulator serves as an onboard voltage
reference. If a regulated 12-15 volt supply were available,
the 7812 regulator could be omitted entirely and the op amp and 555
negative voltage charge pump could operated directly from that.
Figure 4: Top: The schematic of the 8 volt switching
regulator. Bottom: The
schematic of the bias control and current limiter. Click on a picture for a larger version.
Circuit description - see Figure 4:
U401A, Q401 and R401-R403 form a high-side current sense circuit
with the voltage on the emitter of Q401 representing 5 volts for one
amp of current flowing through R401. The output from the
emitter of Q401 is applied to U401D which is compared with the
voltage from the wiper of R406 and if it exceeds the threshold, the
output of U401D goes up and the P-Channel MOSFET Q402 is turned off,
limiting the current to the amplifier to the pre-set level.
D408 is necessary as without it, this circuit can "latch up" if the
output the output goes to "full scale" (e.g. the output is
short-circuited) - a condition caused by the base-collector junction
of Q401 conducting back to U401A's non-inverting input if its output
goes above about 8 volts. An LM324 is used for this circuit
because it is capable of operating down to the negative rail on both
the input and
output: If another type of op-amp is used, make certain that
it has this capability!
For bias regulation, output voltage from the emitter of Q401 is
first buffered by U401B and then applied to U401C which compares the
voltage at the wiper of R409 with the voltage from the current sense
circuit: If the current is too low, the output of U401C goes
up and pulls the "Amp Gate Bias" closer to zero (e.g. less negative)
which increases the drain current, but if the drain current is too
high, the output from U401C drops which permits the bias line to
become more negative to decrease drain current and in this way, the
drain current is dynamically regulated. Zener D401 limits the
positive swing from U401C to 5.6 volts, which, after passing through
the voltage divider consisting of R412 and R413 (plus a 350 ohm
resistor on the amplifier board itself) that can vary from about 0
volts (e.g. 5.6 volts at the top of D401) to about -1.2 volts when
U401C's output is at zero volts.
Because the amplifier needs a negative bias voltage, U402, a 555
timer, and associated components are used to form a charge-pump
voltage converter using D404, a red LED, as both a power-on
indicator and as a regulator to clamp the voltage to the 1.9-2.1
volt region. As noted above, a voltage divider consisting of
R412 and R413 (plus the aforementioned resistor on board the
amplifier itself between the bias line and ground) serves to provide
an adjustable negative bias that will "servo" to maintain a constant
drain current under all drive and temperature conditions.
Diode D104 and transistors Q402 and Q403 are used to detect a
failure of the negative supply which would cause the bias
voltage to go positive and saturate the amplifier, putting it into
current limiting. If the voltage from the bias generator
disappears, Q402 turns off which, in turns, turns on Q403 which
applies a positive voltage to the gate of Q402, the P-Channel MOSFET
(which should be attached to a small heat sink) and turns it off,
disconnecting the amplifier's drain supply.
There is one potentially dangerous condition that is not detected
by this circuit: The loss of the 12 volt supply if the 8 volt
supply is still present. If this occurs, the amplifier may be damaged as Q402, the
P-Channel MOSFET, will be fully turned-on, no (negative) gate bias
will be present - turning the amplifier module completely "on" - and
no current limiting will occur! Admittedly, this particular
condition was not anticipated when the circuit was first designed,
but adding a simple PNP transistor/resistor/diode circuit around
Q402 could easily be used to detect if the "+12" volt supply bus
drops below 8 volts, turning on the PNP and forcing Q402 off.
This is not shown in the
schematic, but I may add it later. It is worth
mentioning that the aforementioned condition did accidentally occur
during testing in the beacon and the 8 volt switching regulator
current-limited at some unknown value (probably over 2 amps) but the
amplifier was apparently unharmed!
Adjustment and setup:
The current limiting should be set to be about 20% higher than the
desired drain current and this can be approximately calculated by
measuring the voltage across the wiper of R406 and dividing by 5.8
to yield the current in amps. This may also be verified simply
by using an ammeter between the source of Q402 and ground and
measuring the short-circuit current. An alternative method is
to simply set the wiper of R406 to the high side (+5 volts), verify
that the current-limiting works, and then set the bias point for
operation. One would then decrease the setting of R406 until
current limiting just
occurs, measure the voltage on its wiper, and then adjust it upwards
after it has been verified that the current limiting circuit is
operational, the amplifier may be connected after first pre-setting
the wiper of R409 to ground to set the current to minimum.
While watching the current drawn by the amplifier - which may be
done with either an inline ammeter or by measuring the voltage on
pin 7 of U401 and dividing by 5 - slowly increase the current using
R409 until it reaches the desired level. If the drain voltage
suddenly drops with increased current, it may be necessary to (at
least temporarily) increase the setting of R406 in the event that
the current limiting is activating.
As with many FET amplifiers, the gain is related to the bias and the
saturated output power level will increase with higher
current. Clearly, there is a limit of safe power dissipation
for this (or any) device - partially based on how well heat-sinked
the device is and also its maximum ratings. The data sheet
that I found for the TBQ-3018 is a bit unclear as to what the
absolute maximum operating drain current of the amplifier is, but
850 milliamps and 1 amp are figures noted in the data sheet and
testing indicates that this amplifier can operate at 850 mA for long
periods and at 1 amp for at least a short while if well heat-sinked.
It was noted on the amplifier shown in Figure 3 that with about +2dBm of drive, a drain
current of about 750 milliamps yielded a saturated power output of
about +30dBm (1 watt) at 10.368 GHz which was the desired output
power for the beacon.
Still to do:
More testing! So far it looks as though this
amplifier should be easily capable of just under 1 watt PEP
output on 10.360 GHz with reasonable linearity or just over a
watt on CW with the drain current set to about 1 amp. For
full-time key-down beacon operation (e.g. FSCW) a power level of
1 watt would appear to be "safe" given adequate heat-sinking -
but that shouldn't be too difficult since only about 6-8 watts
of heat need to be dissipated!
Update: This amplifier is currently
undergoing testing for long-term use in the K7RJ 10 GHz
propagation beacon at an output power level of 1 watt. Thus
far, the results look promising!
Build a permanent power supply.For
actual in-field use, a bias voltage generator and drain
current/voltage regulator is required and when this circuit is
completed, details will be posted here. . This circuit
should also properly sequence the drain voltage, applying it
only after the bias voltage has come up to its proper value,
reversing the process when the amplifier is "un-keyed."
The design that I have in mind will regulate the bias voltage
by "looking" at the current-limited drain supply voltage, but
Update: I've built and tested a
regulating power supply for this amplifier in preparation for its
use in the K7RJ 10 GHz propagation beacon. This supply uses
a switching regulator to provide the 8 volt drain supply, it has
both overcurrent and bias-fail protection and it uses servo
techniques to dynamically set the bias voltage - see above!
The small die-cast box seen in the pictures is adequate for
intermittent (e.g. SSB or CW) use where the duty-cycle is
limited and the amplifier's power supply is keyed with the
radio's PTT, but its heat dissipation capability should be
increased if continuous duty/beacon use is anticipated.
Update:When this amplifier was
installed for use on the K7RJ 10 GHz propagation beacon, its
die-cast box was bolted to an aluminum plate to provide additional
heat-sinking. Testing has shown that this degree of heat
dissipation is entirely adequate.
Improve this web page! I plan to put another one
or two of these units together and when I do, I'll take better
pictures and improve the documentation. As the opportunity
and time permits, I'll add to this page - possibly adding
information about other types of similar units.
At the moment I have information only about the
older-style Hughes "Tigris" outdoor units - specifically the
1-watt varieties that may include part numbers in the
"1025901-xxx" series. There is a 2-watt version of
this unit (possibly, with a
part number of "1028350-xxxx") and there are a number
of newer modules out there as well, but I have not had
the opportunity to investigate the usability of parts in
these other units! Unfortunately, it is not
always easy to tell the 1-watt units from the 2-watt units from
the outside unless one compares part numbers with a magic
list... which, apart from the information above, I don't have at
There are a number of other manufacturers of similar units -
often referred to as "BUCs" (e.g. Block UpConverters) that take
L-band and convert it to the transmit frequency. I
know nothing about these other brands!
Before you ask me any questions about whether or not the
unit(s) that you have onhand can be used at 10 GHz,
please re-read the above two statements!