VHF COMMUNICATIONS 4/2006

John Fielding, ZS5JF

Transistor Frequency
Multipliers

1. signal requiring linear amplification then

it is limited to the output power it can
Introduction produce because of the maximum voltage
swing on the base or collector. If the
input voltage swing exceeds the base-
emitter forward voltage the junction
In a previous article we looked at the use starts to behave like a rectifier diode and
of diode multipliers. These all suffer distortion occurs. We can simplistically
from one serious disadvantage, the con- think of a multiplier transistor as being
version loss when multiplying to high two separate items. The first is the diode
orders can be quite high. This makes junction contained within the base-emit-
them only useful for low orders of multi- ter and the second is a very linear voltage
plication of perhaps 3 as a practical amplifier between the base and collector.
maximum. By overdriving the base-emitter junction
Using an active device such as a transis- we generate harmonics of the input fre-
tor can overcome the inherent loss of the quency and these are then amplified in a
simple diode multiplier, because the tran- highly linear manner by the collector-
sistor can have appreciable gain when emitter portion of the transistor.
used as a multiplier, which eases the Unfortunately it isn’t as simple as this,
design of multiplier chains where high but the simplistic idea forms the basis of
multiplication is required. If the transis- all transistor frequency multipliers. We
tor is operated in grounded emitter con- could place a multiplier diode across the
figuration with the input RF drive applied base-emitter and couple into the transis-
to the base the impedance is medium and tor base via a capacitor. By generating
a greater RF voltage swing can be ob- harmonics in the external diode these are
tained from relatively low drive power. then amplified by the linear biased tran-
sistor. Some commercial two-way radio
multiplier chains used this technique.
Unfortunately the efficiency this achieves
2. is worse than the diode on its own if we
include the DC power being drawn by
Principle of Operation the transistors before and after the diode
multiplier.
By utilising the inherent transistor base-
If a transistor is used as an amplifier for a emitter diode junction we have a better

224
 VHF COMMUNICATIONS 4/2006

Fig 1: Basic
multiplier circuit.

method of generating high levels of rent will flow. The transistor is being
harmonic power. Of course we still have operated as a Class C amplifier and the
to filter out all the unwanted harmonics waveform at the collector is similar to a
at the collector and select the desired one square wave, which is rich in harmonic
with as little loss as possible. However, content. All that is required is a suitable
by altering the conduction angle of the band pass filter to select the wanted
transistor we have a method of enhancing harmonic and reject the remainder.
certain harmonics and hence reducing the
unwanted ones. On the face of it this seems to be perfect.
But one thing to be aware of is that the
The most basic circuit is shown in Fig 1. multiplier is very critical to the RF drive
Although this will work it does have a level, too little and it will snap off and
number of problems, which we will too much and it will saturate and the gain
consider later. will fall dramatically. Also the threshold
is very dependent on the base-emitter
The drive signal to be multiplied is forward voltage. If the ambient tempera-
applied to the base of TR1 via C1. The ture varies over a wide range the base-
inductor L1 across the base to ground emitter voltage does the same. At very
holds the base at DC ground and looks low temperatures the voltage required to
like a high impedance to the input fre- make the base-emitter junction conduct
quency. The emitter is also directly will be greater than when the ambient
grounded, so for the transistor to conduct temperature is high. So although it
the RF voltage across L1 needs to exceed works fine in the shack it may not work
the base-emitter junction voltage of about at all from a contest site on top of a
0.6V. This means that when the RF mountain in winter!
voltage is below the threshold nothing
happens, but as soon as the base voltage The danger with this simple circuit is that
exceeds about +0.6V on the positive the base-emitter junction can be driven
peaks of the RF waveform the transistor very negative on the opposite half cycle
will start to turn on. If the RF drive is and small signal bipolar junction transis-
raised even more the transistor will be tors have a limited reverse base-emitter
conducting for a greater portion of the break down voltage, as little as 2V can
RF envelope and so more collector cur- destroy the junction.

225
 VHF COMMUNICATIONS 4/2006

Fig 2: Improved
frequency
multiplier.

Another failing with this simple circuit is is just below the conduction point the RF
that the impedance presented to the driv- drive voltage swing is greatly reduced.
ing stage varies wildly as the drive level This translates into less drive power
varies. Below the threshold voltage the being required and higher input imped-
multiplier appears close to an open cir- ance. Fig 2 shows the modifications
cuit, at high drive levels the multiplier needed.
appears as low impedance, just like the
diode multipliers. The value of the emitter resistor is deter-
mined by the order of multiplication
required. For high multiplication factors
the resistor value is higher than for low
3. order factors. This is to reduce the
conduction angle to enhance the higher
Improving the Basic Circuit harmonics. In simple terms we need to
move the conduction more towards Class
C as the harmonic order increases.
The value of the base potential divider
The first problem we will address is the inter-reacts with the emitter resistor. Es-
temperature sensitivity. In a linear am- sentially what we need to achieve is the
plifier using the common emitter con- DC voltage on the base with no RF drive
figuration we normally include a resistor to be just below the conduction point or a
in series with the emitter to ground to very small collector current flowing, a
provide a small bias voltage. This helps few hundred μA typically. With the
to hold the collector current more stable values shown in Fig 2 the base voltage
with variations in temperature. with respect to ground is approx. 0.62V.
The second problem is the sensitivity to But the collector current flowing in the
the RF drive. We can to some extent emitter resistor raises the emitter voltage
reduce this tendency by applying a small above ground. For a collector current of
amount of forward bias to the base- 20mA flowing when the base is driven
emitter junction with a potential divider the emitter will be approx. 0.2V above
of two resistors. By selecting the resistor ground, making the effective base-emit-
values such that the base-emitter voltage ter voltage about 0.42V. Hence the input

226
 VHF COMMUNICATIONS 4/2006

Fig 3: Incorrect
collector coupling.

RF drive needs to be greater than about ranging the circuit a little can make a
0.3V peak to turn the transistor on. distinct improvement.
The disadvantage of this circuit is the In Fig 1 and 2 the collector is connected
emitter decoupling capacitor; this needs to a radio frequency choke with a wide
to be a low reactance to not only the bandwidth. The coupling capacitor
input frequency but also all harmonics. serves to block the DC supply and also
This requires quite a low inductance transfers the AC signal to the following
capacitor and a SMD chip type is nor- band pass filter. In Fig 3 is shown the
mally the best choice. If leaded capaci- wrong way of coupling out the harmonic
tors are used, for example ceramic disks, signal.
the emitter decoupling may not be good
enough for the higher harmonics and The collector of TR1 is tapped down the
some loss of gain will occur at the higher inductor L1. L1 is resonated to the
harmonic products. required frequency by the variable ca-
pacitor C5 and this tuned circuit could
the first portion of a multi-pole band pass
filter.
4. This tapped inductor method is com-
monly used for amplifiers such as inter-
Coupling Out Of the Collector mediate frequency amplifiers where lin-
the Required Harmonic earity is critical. The tapping of the
inductor allows a good impedance match
to the collector, the conjugate match.
However, in a frequency multiplier this is
This is where most published designs get exactly the opposite of what we require.
it wrong. Although the multiplier stage To understand this logic we need to look
appears to work satisfactorily often rear- more closely at how the bipolar junction

227
 VHF COMMUNICATIONS 4/2006

Fig 4: Transistor parasitic capacity. Fig 5: Inherent varactor diodes in a

BJT.
transistor works and its internal structure. when configured as a tripler. The writer
used this technique when building the
Fig 4 shows the two major parasitic 1152MHz LO chain for a 2C39A power-
capacitances inherent in the bipolar junc- mixer SSB transverter for 23cm and
tion transistor. These are the collector to obtained nearly 1W output at 384MHz.
base (CCB) and collector to emitter This used 150mW input at 96MHz with
(CCE). There is one other which exists the circuit configured as a 4x multiplier.
between the base and emitter but this The use of idler circuits at the input and
does not affect the operation, only the output is the secret to efficient multipli-
input matching. cation.
Normally in a small signal linear ampli- This technique is not just applicable to
fier application the capacitance is rela- the MRF629 (which today is obsolete)
tively constant and can be tuned out. but even small signal BJT RF transistor
Because the RF voltages appearing exhibit this effect when a sufficiently
across the junctions are small the capaci- large RF voltage is developed at the
tances are constant. But in a frequency collector. Hence, we need to match the
multiplier with correct collector matching collector to a much higher impedance
the RF voltages are not small, in fact we than suggested by the formula for conju-
want to make them as large as possible, gate matching. This also explains why
without exceeding the collector-emitter junction field effect transistors (JFETs)
secondary breakdown voltage of the de- are poor frequency multipliers, as they do
vice. The reason is that with a large not exhibit the same varactor effect as
voltage swing the parasitic capacitance is BJTs. Fig 5 shows the inherent varactor
not constant but behaves like a varactor diodes in BJT
diode. Motorola in an Engineering Bul-
letin [1] gave details of how the varactor In Fig 6 is shown an example of the
diode characteristic can be exploited to correct collector matching for a fre-
make more efficient frequency multipli- quency multiplier. This uses a two-
ers. This used the MRF629 a 12V/2.5W section top-coupled band pass filter to
UHF power transistor as a parametric select the required harmonic. The Q of
multiplier. This exhibited a power gain the tuning components needs to be suffi-
of over 6dB when driven by a 100mW ciently high to develop the required RF
signal at 150MHz and developed as voltage swing.
much as 700mW output at 450MHz

228
 VHF COMMUNICATIONS 4/2006

Gig 6: Typical
frequency
multiplier stage.

The input to the multiplier is via a TR1. The coupling into the next section
capacitor with a medium impedance at of the filter is made via C5, a very low
the operating frequency. The drive value capacitor to prevent loading of the
source is a low impedance stage such as collector. This capacitor is also the top
an emitter follower and with an output coupling for the band pass filter so it
voltage swing of at least 1V p-p. The needs to be a very low value, which suits
tuned circuit of L1 and C3 selects the the circuit configuration well. The sec-
required harmonic. This develops the ond inductor L2 is resonated by C6, the
necessary high impedance, and hence the coupling to the next stage is performed
higher voltage swing, at the collector of by C7 which is also a low value capacitor
because we need to prevent loading on
the band pass filter, which would alter
the selectivity response if excessive load-
ing occurs. The values of the inductors
and capacitors need to be chosen to suit
the frequency. If more selectivity is
required it is often better to incorporate
this into the following buffer amplifier
before the next multiplier stage.
Fig 7 shows the typical spectrum at the
output of the band pass filter. This
circuit is the one shown in Fig 6 and uses
an MPSH-10 transistor as a tripler. The
input frequency is 50MHz at a level of
5mW in 50Ω and the wanted output at
150MHz measures approx. +13dBm
Fig 7: Typical spectrum for a BJT (20mW). The worst spurious products
multiplier.
229
 VHF COMMUNICATIONS 4/2006

5.
Multiplier Stability

In general, frequency multipliers using

BJTs are inherently very stable because
the input and output are tuned to different
frequencies. Unlike a tuned amplifier
where the input and output are the same,
the only thing that can make a frequency
multiplier unstable is either poor layout
or the internal parametric components of
the transistor. If the drive level is
Fig 8: Unstable multiplier spectrum. marginal the transistor can exhibit erratic
operation. If the input drive varies over
too great a range the multiplier can snap
on or off as the level varies. Hence, it is
are about 50dB down on the 150MHz necessary to have some reserve drive in
signal. This will require further filtering hand. A good minimum is about 3dB
in another band pass filter before it is above that required to ensure reliable
applied to the next multiplier or used by multiplication.
the equipment. When tuning up a multiplier stage the
One thing not obvious is why a bipolar power output should vary smoothly as
junction transistor is the preferred device. the input and output are aligned. If the
Some schematics show junction Fets as multiplier snaps on and off as the tuning
multipliers. This is a poor choice be- is altered it is usually a sign that the input
cause the Fet does not contain the same drive level is marginal. A good check of
type of varactor junction as the bipolar. this is to measure the base voltage of the
Hence, the drive power required is much BJT with full input drive. If the drive is
greater to get any significant harmonic adequate the base voltage should be a
generation. Fets make good mixers voltage slightly negative with respect to
because they are used in a “additive ground. A value of -0.25V to -0.75V is
mode” with a second carrier, the local normally about right. When the input
oscillator. The local oscillator energy drive level is more than required the top
causes the drain-source current to swing potential divider resistor in the base bias
up into the saturation region and the non- chain can sometimes be omitted and the
linearity is greatest under this condition. drive develops a negative base voltage
The Fet is operated as a saturated switch across the base to ground resistor. In
to “chop” the input waveform and cause cases such as this it is prudent to connect
additive and subtractive products, one of a reverse biased diode across the base to
which is the wanted mixing product. If a ground to clamp the base reverse voltage
junction Fet and a similar die size bipolar to less than 0.7V to protect the base-
junction transistor are compared in a emitter junction.
multiplier circuit the bipolar will deliver Fig 8 shows the spectrum for a multiplier
more harmonic output power and require stage where the input drive level has
less drive power. Hence, junction Fets been deliberately reduced to show the
are not very suitable for straight multi- effect that instability causes.
plier duty.

230
 VHF COMMUNICATIONS 4/2006

Fig 9: Test point

added to multiplier
stage.

6. creasing the collector feed resistor value

to limit the collector voltage.
Optimum Supply Voltage

7.
In all the years of designing frequency
multipliers using BJTs it has been found Choice of Transistor Type
that there is a certain range of supply
voltage where the optimum performance
occurs. It is not quite clear why this is so
but is probably to do with the inherent As strange as it may seem we do not
parametric varactor diodes in the BJT. need to use a “super-transistor” for an
However, from many experiments the efficient multiplier. Often the choice of a
optimum supply voltage on the collector- transistor is dictated by the frequency of
emitter junction is about 7 to 9V, the so operation and hence the ft as an amplifier
called “sweet-spot”. With voltages far needs to be quite a bit higher than the
from these limits the efficiency suffers, operating frequency. This however is for
with too low a voltage the power output a transistor operated as a linear amplifier,
falls away and with too high a voltage the when used as a highly non-linear multi-
efficiency also falls off. Hence, when plier we can often get by with less exotic
designing a multiplier chain using BJTs it devices.
is often best to use a supply voltage of 8 Form experiments made on a wide selec-
to 9V. This is convenient as a 3-terminal tion of transistor types it has become
regulator can be supplied by the nominal apparent that even lowly devices such as
12V-vehicle supply. the 2N2222 with an ft of only 100MHz
If the drive level into the next stage is too work perfectly well up to a few hundred
high a simple cure is to starve the inter- MHz when operated as a doubler or
stage buffer amplifier transistor or tripler stage. For the frequencies higher
MMIC following the multiplier by in- than this a move to a popular type such
as the MPSH-10 or the BFR92A or

231
 VHF COMMUNICATIONS 4/2006

Fig 10: Push-Pull

doubler.

BFR93A in the SMD SOT-23 package nate in 50Ω so that normal test equip-
work well even as high as 2GHz, which ment can be connected. This not only
the manufacturers data does not suggest. introduces extra loss but also adds unnec-
Of far more importance is the drive level essary components to the circuit. We can
and the filtering components used to safely assume that the base input imped-
select the wanted harmonic. If the filter ance of a typical BJT when driven at the
has excessive loss then the advantage of optimum level is about 200Ω to 500Ω.
using an exotic transistor is wasted. Therefore the drive power is not going to
be too high; perhaps +10 to +13dBm in
50Ω when transformed up will give
sufficient base-emitter voltage swing.
8. Probing a high impedance portion of the
Test Points circuit with a RF AC voltmeter often
causes detuning because of the probe
capacity. A simpler method is to incor-
porate some test points where a 50Ω
When aligning a multiplier chain it is instrument can be attached without
helpful to be able to break into the chain breaking the chain. Often two resistors
at critical places to attach test equipment can achieve this and the 50Ω equipment
to see what is going on. Some designers can be attached as and when required
make all the multiplier stages to termi- without upsetting the system. This con-

232
 VHF COMMUNICATIONS 4/2006

Fig 11: Push-Push

tripler.

cept is illustrated in Fig 9. The two monolithic integrated circuit amplifiers

resistors are a high value one to tap into (MMICs) as frequency multipliers. The
the high impedance point and a second choice of devices is somewhat limited
termination resistor to suit the test equip- but several writers have reported good
ment. A coaxial cable can be connected results. However, because of the internal
and taken to a spectrum analyser to see biasing components of typical MMICs
the spectrum. The amplitude will not be these devices do not really lend them-
correct but the spectrum should still be selves to this application. Whereas it is
the same. true that any transistor can be over driven
to force it into a non-linear operating
portion this seems not so easy with the
typical MMIC. As the typical MMIC is
9. designed to interface with 50Ω termina-
tions intuitively we can see that we
Using MMIC Amplifiers as cannot run the collector into the high
Multipliers impedance necessary to induce the varac-
tor diode effect.
Where MMICs can be made to work is
There have been several application with an external shunt diode across the
notes and other articles published con- input, with DC blocking capacitors to
cerning the use of wide band microwave prevent the bias being upset, to generate

233
 VHF COMMUNICATIONS 4/2006

need to be wound for the best balance

and this limits the higher frequency per-
formance.
This circuit can also be modified to form
a push-push tripler where the predomi-
nant harmonics are the odd order ones.
This is shown in Fig 11.

11.
Multiplication Order

Fig 12: Wide band sweep of LO In many cases we need to multiply by a

signal. fixed number and this is clearly defined
by the starting and finishing frequencies.
the harmonic energy and then to use the For example a receive-converter for
MMIC as a conventional amplifier to 23cm may require a 24x-multiplier chain
raise the level. However, when you run to arrive at the final LO injection fre-
the calculations of the total power re- quency. A 24x-multiplier can be ar-
quired the efficiency still ends up being ranged in several different ways. We
quite low and a cheaper BJT will prob- need to multiply by 24, which can be
ably out perform it at far less cost. broken down into doublers, triplers or
MMICs are good as inter-stage buffer quadruplers (2x, 3x or 4x). Choosing
amplifiers but not so good as multipliers. which multiplier factor to use and where
in the chain is often not simple. As a
general rule it is often better to make the
first multiplier stage with the greatest
10. multiplication factor, so in this case we
might choose the first multiplier stage to
Special Multipliers be a 4x. Against this must be weighed
the difficulty of filtering out the harmon-
ics close to the wanted multiplier prod-
Where we need additional suppression of uct.
certain harmonics we can utilise some of As an example, consider the 23cm
the ideas found in diode multipliers. Fig receive-converter for an IF of 144MHz.
10 shows a push-pull doubler using two The starting frequency is 24x less than
BJTs. The circuit suppresses the odd the LO injection, which in this case is
order harmonics by about 30dB and 1152MHz. This determines the crystal
leaves just the even order harmonics. frequency as being 48MHz. This is a
The variable resistor VR1 allows the good choice as the 3x product is 144MHz
circuit to be balanced by adjusting the and if a test port is provided on the first
emitter currents of TR1 and TR2. This multiplier then a small amount of the
provides the best null to the unwanted 144MHz signal can be used to set the
products. For optimum performance a band edge on the 144MHz tunable IF. If
dual-transistor in a single package offers the 144MHz receiver is accurately cali-
the best results. brated in frequency readout the receiver
The wideband transformers T1 and T2 can be used to zero beat the crystal

234
 VHF COMMUNICATIONS 4/2006

oscillator exactly onto 144MHz. Hence, band pass filter is likely to be easier.
in this case the first multiplier needs to be This also simplifies the filter design and
a tripler. it is likely to have less insertion loss than
the narrower filter previously required.
However, consider the case where the In some cases commercial helical filters
filtering of the first multiplier is insuffi- can be obtained aligned for the required
cient to reject all the 48MHz spaced frequency which allows a more compact
products. If these are allowed to pass design. Companies such as Toko and
through the subsequent multipliers then Temwell list suitable filters in their
we will end up with 48MHz products standard product range that occupy very
spaced either side of the final LO fre- little board area and these have adequate
quency. The 1152MHz product is not of attenuation to the unwanted products.
concern but the image at 1008MHz
(288MHz away) will cause degradation Unfortunately not all amateurs have ac-
in the receiver performance. cess to a spectrum analyser to ascertain if
the multiplier chain filtering is adequate.
Hence, it is better to first double the Many constructors have little more than a
48MHz to 96MHz, band pass filter the RF diode probe and perhaps an absorp-
result as well as possible to reduce the tion wave-meter or milli-watt meter to
48MHz products to negligible levels, and align the multiplier chain. Although it
then proceed from there. No integer may appear to work satisfactorily it is
multiplication factor using 96MHz can only when it is examined on a spectrum
produce a signal at 1008MHz so this is a analyser that the truth emerges, some-
better plan. times causing a big surprise! Unwanted
Assuming we choose the first multiplier products at the final LO of a receive-
as a doubler, this leaves a total of 12x converter only cause the operator a prob-
further multiplication to arrive at the final lem, but if the same dirty LO is also used
LO injection. Filtering at 1152MHz will to mix up in a transmit converter then we
require a fairly good filter and to ease the are potentially causing spectral pollution
design of the filter we need to place the to other users. This is not only irrespon-
unwanted harmonic products as far away sible behaviour but probably in contra-
as possible. We could choose the final vention of our license conditions. These
multiplier to be a doubler and hence the spurious signals may interfere with sensi-
injection frequency into the final multi- tive receivers used by other services,
plier would then be 576MHz. If this perhaps radio astronomy, civil or military
choice is made then we need to assess if communication services and should be
it is possible to arrive at 576MHz from avoided for obvious reasons!
96MHz. The answer is we can by One factor which is misunderstood by
multiplying by 6, this would be a 2x and most amateurs is how clean the LO
a 3x multiplier in either order. signal needs to be. If the signal is being
If we chose the more traditional route of used to drive a mixer to down-convert or
multiplying by 3 from 384MHz to arrive up-convert then the harmonic content of
at 1152MHz then we need a total multi- the final signal is not important, in fact
plication of 4x from 96MHz; this could given the correct phase relationship a bit
be 2x & 2x in two separate stages or a 4x of 2nd and 3rd harmonic content can be
in one stage. Again we need to assess the beneficial to the following mixer. The
difficulty of making a band pass filter at following mixer will regenerate the har-
384MHz that can reject the 96MHz monics of the input signal so there is no
spaced products with sufficient attenua- need to suppress the 2nd and higher
tion. The safest method would be the harmonics of the LO signal. What is of
dual 2x approach as the filtering of a far greater importance is to eliminate as
192MHz spaced product with a 384MHz far as possible the lower multiplier prod-

235
 VHF COMMUNICATIONS 4/2006

ucts especially the ones closest to the LO frequency can have a dramatic effect
wanted LO frequency. on the image and other spurious response
characteristics. If the 23cm band is used
To illustrate the problem take a look at as an example then the choice of 28MHz
Fig 12. This is the tripler circuit shown as the 1st IF is not a good choice. For
in Fig 7 but with the spectrum analyser this we need a LO at 1268MHz and the
sweep increased to cover up to 1GHz. image will then be at 1268-28MHz =
Although the products close to the 1240MHz which is right in the middle of
wanted 150MHz are acceptable the far the aeronautical DME and radio altimeter
out products around 700MHz are very band worldwide! Typical radio altim-
bad. They are barely 30dB down and are eters are downward looking radar sys-
due to a poor layout where the harmonic tems fitted to most aircraft and these
energy was able to hop over the filter and radiate up to 10kW of pulsed power.
get picked up in the output amplifier DME systems are used for distance
stage. A contributory factor is the topol- measuring and the aircraft radar interro-
ogy of the band pass filter; it is a top- gates a land-based transponder, this then
coupled design and hence tends to act sends a reply pulse delayed by a known
like a high pass filter to the higher amount. By measuring the time delay the
harmonics. If the spectrum analyser aircraft can calculate the exact distance to
sweep had not been opened up we would the ground transponder. A typical air-
be blissfully unaware of how bad the craft DME transmitter is about 300W and
spectrum was. With the writer’s spec- the ground station anything up to 10kW.
trum analyser covering from 10kHz to
12.4GHz it is possible to see significant The writer recently had to do a signifi-
levels of harmonic energy all the way up cant system revision for a radio telescope
to nearly 3GHz. This is with a transistor receiver for the Hydrogen Line at
“officially spec’ed” only to an ft of 1420MHz that is under development. It
650MHz. was originally proposed to use a
1280MHz 1st LO and the 1st IF was at
If significant levels of spurious signals at 140MHz. Only recently it became appar-
the LO exist then these are potential ent that a significant amount of L-band
interference generators for frequencies microwave signals on or near the image
far removed from the wanted channel frequency of 1140MHz from aircraft
frequency. Most microwave receiver DME transponders existed with high sig-
front ends do not have very high selectiv- nal levels caused severe interference and
ity because of the nature of LNA design; this forced a change in 1st LO and 1st IF.
few LNAs have anything like the re- The 1420MHz receiver has only 40dB of
quired rejection characteristics because image rejection because of the nature of
this would destroy the low noise figure the LNA and front-end filter design to
required. If the unwanted signal is of obtain the required sensitivity. The mini-
sufficient amplitude to get past the front mum discernible signal of this receiver
end filtering and into the mixer then with the LNA noise figure of 0.18dB is
these will appear as “birdies” or “sprogs” less than –146dBm, which is about 20dB
when the receiver is tuned across the better than a good 23cm receive con-
band. If you are really unlucky then you verter. In this case it was found that
may find that your weak signal DX changing from a “Low-Side” injection
channel is completely obliterated by a scheme to a “High-Side” injection placed
strong signal far from the wanted fre- the image frequency at about 1700MHz
quency. Running some simple mixing which is also a protected portion of
spurious calculations can highlight poten- spectrum reserved for SETI research and
tial problem areas. the 1st IF then occurred at 152MHz
The choice of the 1st IF and hence the which is another protected piece of spec-

236
 VHF COMMUNICATIONS 4/2006

trum for Radio Astronomy. The 1st LO RF fields when transmitting 250W on
then occurred at approx. 1572MHz. 23cm. This caused FM when using SSB
or a chirp when CW was used. With
adequate shielding and supply line by-
passing these problems disappeared with
12. the new oscillator module.
Special Local Oscillator It must be appreciated that the filtering
inductors need to be well shielded, the
Techniques best option is pre-wound canned coils or
helical filters such as those made by
Toko. If air wound inductors are used
During the upgrade to the writer’s 23cm the coils radiate spurious energy and this
transverter an unusual approach was can hop over shielding plates and effec-
used. The original design used a 5th tively bypass the filter. The RF voltage
overtone crystal at 96MHz, but this suf- at the collector of a multiplier stage can
fered from excessive warm-up drift and be many volts, and these make good
other problems. Because the desire was “mini-transmitters” unless the RF field is
to make as little change as possible it was contained in shielded compartments.
decided to make a new 96MHz oscillator Grounding, layout and shielding are all
module that then fed into the existing vital parts of any oscillator chain design
modules. This new module used a and construction.
12MHz TCXO, which was multiplied up
to 96MHz, a factor of 8x. The first
multiplier used a 2x and the second a 4x. 13.
After filtering and amplification the
96MHz signal was fed to the original References
oscillator multiplier chain. A test port on
the second multiplier provided a small
amount of 144MHz harmonic energy to
calibrate the 144MHz transceiver. [1] EB-70A “Frequency multiplication
Because the first multiplier output spec- simplified by internal faraday shields in
trum contained products spaced 12MHz MRF629”, Motorola Semiconductors
and 24MHz either side of the wanted
frequency the filtering required was quite
severe. If these products are not removed
at the source then it practically impossi-
ble to eliminate them further down the
multiplier chain. The aim was to attenu-
ate these close-in products by at least
70dB, in the end a figure of over 80dB
was achieved by careful attention to the
early band pass filters. The 96MHz final
output was clean; the worst spurious
product was -86dBc, which was consid-
ered sufficient. Filtering, shielding and
supply line bypassing achieved a com-
pact and clean module suitable to drive
the existing multiplier chain. One of the
other problems with the 96MHz crystal
oscillator was its susceptibility to strong

237