BASIC LINEAR AMPLIFIER

by Harry Lythall - SM0VPO

Table of Contents
Introduction
Modulation Modes
Classes of Bias
Class-A Amplifier
Class-AB Amplifier
Component Values
Transformers
Prototype Design
Conclusion

- Object of this article.

- Basic info about modulation.
- Basic info about classes of bias.
- One transistor amplifier project.
- Two transistor amplifier project.
- Rough/easy guide to component selection.
- And how they are constructed.
- A (belated) prototype for this project.
- A few parting words and tips.

Introduction
This is both a project AND a bit of basic theory regarding the design of wideband HF power amplifiers. As usual, it is prompted as a result of e-mail in
which I have to repeat myself over and over again. The object is to show a
couple of basic configurations and practical examples of QRP linear
amplifiers, although I have somewhat oversimplified a few things. But as an
author I have the right to do this! So I will NOT delve deep into mathematics
or give loads of formulas, but some degree of calculation is needed by the
reader. I shall try to make it as simple as I possibly can, but you may need to
take your socks off.

Modulation Modes
A "Linear Amplifier" is just an amplifier that does not create excessive
distortion, and it can be used to amplify a ready-modulated signal without
changing the characteristics of that modulated signal. It can be used for
amplifying most transmission modes, for example:
CW Continuous Wave

- Unchanging frequency or amplitude, usually interpreted as

Morse Code.

Amplitude
- A CW signal, the amplitude of which conveys information.
Modulation
DSB Double Side-Band - Amplitude Modulation with unused component removed.
SSB Single Side-Band - Amplitude Modulation with unused and duplicated
AM

components removed.
FM

Frequency
Modulation

- A signal with constant amplitude but varying frequency.

There many other modes, but I am only concerned with the above, which
compromise the major modes I am interested in. CW and FM have an
unchanging amplitude to convey information so non-linear amplifers are not
really required, but it does no harm to use a linear amplifier to amplify CW
and FM signals. Linear Amplifiers may be quite frequency conscious, making
them unsuitable for such techniques as "frequency Hopping" or even video. In
this article I will be concerned with wideband (untuned) amplifiers for use in
the HF band (1MHz to 45MHz).
Another use of the "power Amplifier", is for use in receiving antenna
distribution systems. You may wish, for example, to use one antenna for a
whole housefull of receivers. The linear power amplifier is an excellent unit
for this purpose, between the antenna and the passive signal splitter. Yet
another, further, different use is that of amplifying antenna signals where there
is the likelyhood of strong "in-band" or close signals could "block" a normal
low-power receiver pre-amplifier.

Classes of Bias
Before we continue, perhaps a word about "Classes of Bias". The info I give
here is perhaps a little over simplified, so the more experienced reader may
wish to jump over this section. The next section (Class-A project) is coloured
dark green.
Class A
When a signal is applied to a transistor, that signal comprises both positive
and negative swings of current or voltage. Unfortunately, transistors cannot
handle a negative current, so we must add a standing bias current. To this
current will be added and subtracted signal currents, so this bias must exceed
the maximum signal in the negative 1/2 of the signal waveform. This is ClassA bias. It is used in single-transistor amplifiers but has a poor efficiency,
practically not greater than 45%.
Class B
Class-B dictates the device has little or no standing DC current and therefore
amplifies positive 1/2 waves of the applied signal. Due to the lack of a
standing and heavy DC current, it is a lot more efficient than class A. When
applied to a single device it creates a LOT of distortion, but when used in a

complementary or push-pull amplifying configuration, little distortion will

result. A practical efficiency would be in the order of 80%.
Class AB
This class of amplification is where the device is biassed somewhere between
class-A and Class-B and when used in a complementary or push-pull
amplifying configuration, little or no (comparatively) distortion will result.
Class C
There is another class of amplification: Class-C. I will mention this in passing
but I will not be concerned with this class in this article. Class-C is where the
amplifying device is heavily cut-off (non-conducting) so that small signals are
not even amplified at all. This generates tremendous distortion and harmonics,
both odd and even. This class of amplification can only be used for ultraefficient amplifiers without any amplitude information, such as CW and FM.
It is useless for AM and SSB. Because of all the harmonics, Class-C is used in
frequency multipliers.

Class-A Amplifier

This is a typical single-transistor Class-A amplifier. R1 and R2 divide the

supply voltage to feed the Base of the transistor, via the secondary of T1. This
normally sets the Base to about 25% of the supply voltage. R3 emitter resistor
sets the standing DC current through the transistor. T2 matches the collector to
the load.
The single transistor circuit is quite capable of delivering several watts of RF
power, with the correct choice of components. The big disadvantage of this
circuit is that TR1 must have a high standing DC current, which usually
means the transistor must be fitted with a heat-sink. Another downfall is that
the "hfe" of transistors fall with increased current. This means that as the RF
signal changes it's level throughout the cycle, so, too, will the gain of TR1. As

the current rises, the gain falls. The following diagram shows quite clearly the
effect on the RF output signal.

Waveform #1 is the desired waveform but if the transistor is not grossly

under-rated then waveform #2 is what you will get. As you can see, this
introduces a LOT of 3rd harmonic. It may not be as bad as shown, yours may
20dB or 30dB down so it will look perfect to the naked eye. Your neighbours
may have other views, especially if they are trying to watch "The Bold & The
Beautiful" on their one-eyed god. If you over-rate the transistor (use a 10 watt
transistor to drive only 1-watt) and use a decent output low-pass filter, then all
should be well.

Class-AB Amplifier

Here is a better solution. TR1 and TR2 each amplify opposite halves of the RF
wave-form and each half is combined in T2, which matches TR1 and TR2 to
the load. Again, R1 and R2 divide the supply voltage to feed the Base of the
transistors, via the secondary of T1. This is normally set to about 1-volt so
TR1 and TR2 are conducting just a small amount of current. If a trough in the
waveform causes a transistor to stop conducting then there is no problem, the
other transistor will be doing the work. The emitter resistor limits the current
and improves linearity.
Since this amplifier is symetrical, the third harmonic will be greatly reduced,
as long as at least one of the transistors is always conducting something. So in
effect, one transistor amplifies the +ve 1/2 cycles and the other transistor
amplifies the -ve half cycle.

Waveform #3 represents a decent sinewave for comparison, but waveform #4

shows you what happens if the class of bias changes from Class-AB to ClassB and beyond. This particular circuit is efficient, typically 80%, and only
dissipates any significant heat during RF activity. This circuit can be used to
deliver RF power from 100mW to well over 100-Watts, with suitable choice
of components.

Component Values
Components for Class-A Amplifier
The Class-A amplifier must be so biased as to have a standing current
exceeding the maximum signal current variations. So let us begin at the
output, since it is the output power that dictates the bias current. Let us assume
we want a 100mW output power and the supply voltage is 12-volts (13.8v).
The circuit we shall be using is:

With a 12 volt (13.8v) supply, the complete sine-wave peak-to-peak value can
only vary from about 2-volts to about 13 volts, or a total swing of just over 10
volts. The RMS value of voltage is therefore 10 x 0.3535 = 3.5 volts. This is
the maximum amplitude of our output voltage at the collector of TR1. To
deliver 100mW of power, with only 3.5 volts, we need a current of W / E = I
or 0.1 / 3.5 = 0.028 Amperes = 28mA. Since this is the RMS value, then the
peak current will be 28 x 1.414 = 40mA. So the output signal at the collector
of TR1 needs to vary from -40mA to +40mA at peaks. We must therefore add
a standing bias that is greater than 40mA, say, 50mA. So now the current will
vary from 10mA to 90mA with an average value of 50mA.

I will select a transistor, such as the BC547, with an "hfe" of about 200. With
50mA standing Collector current, then the Base current is Ic / hfe = 50mA /
200 = 0.25mA. This means that R1 + R2 must burn at least 2.5mA (10 x Base
current), so R1 + R2 must be about 5520 Ohms. R1 = 4K7 and R2 = 820R
would be a good division using preferred resistor values, and this would give
about 2.05 volts at the Base of TR1, resulting in 1.3 volts at the Emitter of
TR1. R3 = Ve / Ie = 1.3v / 50mA = 26 Ohms. 27 Ohms would be close enough
since I have rounded up the odd number here and there.
C1 and C2 should just be big-ish. 22nf would be fine and the electrolytic is
normal practice for RF. 47uf is fine. C3 will increase the gain, but also
increase distortion quite a lot.
Components for Class-AB Amplifier
The Class-AB amplifier does not need be so biased as to have a standing
current exceeding the maximum signal current variations. Just a small
standing current of 5mA is adequate for up to about 1 Watt, 20mA for up to 10
watts and 80mA up to 100 Watts. But we will make an amplifier for 5 Watts
using 2SC2078, a device used for CB RF output transistors. You can pick
other devices, such as 2SC3781 - a line-output transistor for portable TV sets.
Let us assume we want a 5 Watts output power, and the supply voltage is 12volts (13.8v). The circuit we shall be using is:

Since two devices are each handling 1/2 the input waveform then the peak
collector voltage of one device will be about 12v volts and the RMS collectorto-collector voltage will be about 8 volts. To deliver 5 watts, the RMS current
will be W / E = I or 5 / 8 = 0.625 Amperes = 625mA. Since this is the RMS
value, then the peak current will be 625 x 1.414 = 885mA. Each transistor will
therefore be conducting almost 1 Ampere at signal peaks. Using a power
transistor with Hfe = 80 then Base current is going to be about 11mA. R1 +
R2 = 13.8v / 11mA = 1254 Ohms. This time I will increase the current a little
and use resistors of R1 = 1K0 and R2 = 100R. This means the emitters of the
TR1 and TR2 will be about 0.5v DC. R3 should drop no more than 1 volt at
full output (900mA) so 1 Ohm will be about right here, BUT the 100-Ohm
resistor must be reduced in value to set the standing bias current to 10mA or
so. R2 should therefore be a preset variable resistor.

Another variation that would be more stable would be to replace R2 with an

ordinary silicon diode. R1 could therefore be selected to give TR1+TR2
standing current of 10mA and no adjustments need be made, even with
varying supply voltages.

Transformers
As a general "rule of thumb" you can assume the input impedance of a stage is
equal to the emitter resistor. It may not be 100% accurate, but it will put you
in the right area and is certainly close enough for winding coils. In all these
examples I used torroidal transformers. These comprise a ferrite slab with two
holes through them. You can place two ferrite beads together and treat them
exactly the same - this is what I do, and it is cheaper. This is the method I used
in this photograph:

T1 100mW
T1 is designed to match the input to the base of the transistor. In our 100mW
Class-A single transistor amplifier, the base impedance is a little uner 25
Ohms, and the input is 75 Ohms. With an impedance ratio of (making the
numbers easy) 80:20 = 4:1. The turns ratio is the square-root of the impedance
ratio = 2:1. 8-turns primary and 4-turns secondary. I have always found it
advisable to feed transistors a little lower impedance. The circuit then

becomes more predictable, even with different transistors. I picked 8-turns for
the 50/75-Ohm winding because experience has taught me that this is a good
general figure that works from 1MHz to 40MHz with most ferrites.
T2 100mW
T2 in the 100mW circuit matches the load to the collector of the transistor.
When working out the collector impedance FORGET THE DC BIAS. It is
only the signal that is of interest. In out 100mW amplifier we had 3.5vRMS at
28mA, so the collector impedance is 3.5/.028 = 125 Ohms. To match 125
Ohms to 75 we need an impedance ratio of 125:75 = 1.7:1 or a step-down
turns ratio of 1.3:1. T2 could have two configurations - 8 turns with a tapping
at 2 turns from the collector - OR - 8-turns primary with 6-turns secondary.
T1 5 Watts
The 5-watt version has a much lower Base impedance. In this case it is around
1 Ohm for each transistor. If the input is 50/75 Ohms then the impedance ratio
is about 64:1+1 which is a nice round figure of 8:1+1 turns ratio. 8-turns
primary winding and 1+1 turns secondary. This means 2-turns with a centretap. Use a little thicker wire for the lower impedance windings.
T2 5 Watts
The 5-Watt version output transformer must match 8 volts (RMS) at 625mA =
12.5 Ohms to the output of 50/75 Ohms. Interesting that for 50-Ohms the
impedance ratio is 1:4 so the turns ratio is 1:2 (2:4). The primary is tapped, so
1+1:4 is the correct turns ratio. The primary consists of 4-turns with a centretap (2-turns + 2-turns) and the secondary of 8-turns.
T1/T2 Multistage
You could make a multi-stage transformer in exactly the same way - assume
somewhere around 8-turns to 12-turns is 50/75-Ohms and wind accordingly. If
you had, for example, the 100mW version feeding the 5-Watt version to make
a 3-transistor Power Amplifier, then the coupling transformer will be "T2
100mW Primary" and "T1 5-Watt Secondary". This would be 4 + 4 primary
and 1 + 1 turns secondary, just like this:

T1/T2 Ferrites
When I look in my junk-box I find many ferrites. I pick the GREY ones for
RF use. The miniature ferrites I use are about 3mm Diameter with a 1mm
hole. The length is about 4mm. These ferrites work well up to about 1 watt of
power. One of the most common mistakes is to use ferrites that are too big.
The larger they are then the worse they are at higher frequencies. The ferrites I
use here seem to work well up to 150MHz without a problem. You do,
however, need bigger ferrites above about 1 watt. Today they are expensive or
hard to get but I have had some success modifying "junk-box-ferrite". If you
want to drill ferrites then freeze them in a large block of ice and use nice new
tungsten drill bits in a high-speed drill.

Prototype Design
This is the difficult bit. I have already used these circuits as building-blocks
for many years, so it is a bit late to start talking about prototypes. So I propose
to just "throw together" a prototype 250mW QRP linear amplifier, using
Class-AB and include the mandatory low-pass filter on the PCB. The board
has sufficient space to add larger ferrites and the transistor pin-spacing allows
you to insert TO220 power devices to make that larger 5 Watt amplifier. The
TO220's may be mounted below the board, bent over and bolted to a heatsink.
This will also support the PCB, but it is a bit fiddly when it comes to
positioning the legs and bending them over. You even could modify the
artwork a little to use two pairs of BC547 transistors and get 600mW out (T2
secondary = 5 turns).
But I will continue with my 2x BC547 and be content with 250mW. I don't
normally give PCB component layouts here due to limited space on the
projects server, but in this case I will make an exception since it is only 12K.

Here is the simple amplifier. It only needs 5mW of drive at 1.8MHz to give
300mW out so it makes an ideal addition to an AM oscillator to generate AM
or QRP SSB signals.

The circuit to accompany this PCB is:

As you can see, the circuit is exactly as discussed above: yet another case of
simple is good. Component values would be:
R1 = 4K7, R2 = 1K0, R3 = 47R, C1 = 10n + 22n + 10uf, C2 = 22nf,
T1 = 8-turns primary plus 4-turns secondary with centre-tap,
T2 = 8-turns primary with centre-tap plus 4-turns secondary.

C4
C5
C6
L1
L2

1.8MHz
1800pf
2700pf
1800pf
4.80uH
4.80uH

Frequency Dependent Components

3.5MHz
7MHz
10MHz
14MHz
21MHz
28MHz
1000pf
470pf
350pf
150pf
100pf
91pf
1800pf
820pf
610pf
330pf
220pf
180pf
1000pf
470pf
350pf
150pf
100pf
91pf
2.75uH 1.40uH
1.05uH
0.70uH
0.46uH
0.35uH
2.75uH 1.40uH
1.05uH
0.70uH
0.46uH
0.35uH

When it is all built it looks something like the photograph below. Here you
can see the coil construction quite clearly and that there is loads of space for
much larger ferrites than the little tiddley ones I have chosen. The rest of the
board is quite well spaced out so there is plenty of room for using larger
components than those I chose. This particular photograph, however, is is the
same board foil I have provided but the push-pull output also drives another
push-pull output stage to give 5 watts out.

Conclusion
I have avoided mathematics as much as possible and tried to use simple Ohm's
Law and "finger maths" with a description. For example the output collector
impedance of an amplifying stage is given by the expression "Vcc squared
over output power in Watts". This, however, is a MAXIMUM value. The
impedance may be a little lower, but never higher. In this respect I believe that
a little explanation and finger-maths is more than adequate for the amateur
constructor. If you wish to go deeper then do as I did - buy books and read.
If you now want to see an older project based upon the information given on
this page then look at my 10 Watt Power Amplifier project Since this project
will deliver over 15 Watts a little more attention has been paid to neutralising
(330R + 10nf) and bias control. The bias control is even temperature
compensated.
So now all you have to do is to wait until I can add the PCB foil pattern file to
my download section and you can start building your own QRP transmitters
based on my design. Small tip for you, a 5-turn loop around the coil of
my Grid Dip Oscillator (GDO) will provide several milliwatts of RF drive - an
excellent testing tool.

Note that I have tried to split the project up using a different text colour for each paragraph. If it worked
then please let me know (via my messageboard) since any feedback on presentation is appreciated,
however small, positive OR negative.

Return to INFO page

TEN WATT LINEAR AMPLIFIER

by Harry Lythall - SM0VPO
Introduction
It is quite easy to get a watt or more with very simple equipment, but to get
more than 5 watts becomes a little more difficult. This article describes a 10
watt linear amplifier that is capable of delivering over 15 watts into 50 ohms
and uses cheap plastic transistors that are used in CB equipment. If you have
difficulty in finding 2SC2078 then lift the lid of your CB set to find a suitable
alternative. The bias generator transistor, TR4, is marked TIP31 in the circuit
diagram, but here you can use just about anything that will fit. You could even
use another 2SC2078, if you had money to burn, but more practical
components would be TIP41, TIP3055, MJE3055. All that matters is that it
will pass up to 1 Ampere and have the correct base details in a TO220 case.

Circuit

The amplifier has a wide bandwidth, from 1.8 MHz through to over 30 MHz.
The drive level required is only about 2 - 5 mW under 14 MHz, rising to 10
mW at 30 MHz. You can therefore make a good QRP CW rig with nothing

more than this PA and a simple crystal oscillator. I can achieve 12 watts out of
mine using a 10-turn loop around my Grid Dip Oscillator! I can get over 15
watts from my Marconi signal generator, but above about 12 watts it is being
over-driven an may not be very nice to look at on the spectrum analyser. The
circuit was designed to be as clean as possible. Here is a view over the
completed PCB (sorry the photo is so crappy):

The circuit was originally designed to accompany my phasing-type SSB

exciter, but it can be used to amplify almost any HF signal from 2mW in the
HF band. Note that there is needed a Low-Pass filter between the amplifier
and antenna. This is a requirement for ALL transmitters.
L1 matches a 50-Ohm input to the 10-Ohm input impedance of TR1. The
output of TR1 is coupled via T1 to TR2 and TR3 bases. T1 also transforms the
impedance to the very low input impedance of these two transistors, so the
secondary winding must be quite thick wire.
TR2 and TR3 amplify the signal even further, up to about 12-Watts. They are
in a push-pull configuration so T1 secondary and T2 primary must both be
symetrical. T2 increases the output impedance to 50-Ohms.
TR1, TR2 and TR3 all have a 330R and 10n between the collector and base
terminals. This is needed for stability and is a crude form of neutralising.
Without these components then the stages would almost certainly oscillate or
generate spurious signals. An additional 180pf capacitor from both TR2 and
TR3 collectors to ground (not shown in the circuit diagram) give the amplifier
an even cleaner output signal. The 68pf capacitor across the output may be
increased if you never use the 30MHz band. It, too, increases the stability and
helps to keep the harmonic content low. My spectrum analyser did not show
any significant spurious or harmonic outputs from 0 - 100MHz when driven at
10-watts continuous, 14MHz. All spurious outputs were better than -60dBm (70dBc), which I though was pretty good!

TR4 is nothing more than a high-current constant voltage series regulator

using a 3v3 Zener diode for stability. It provides the base-bias voltage for TR2
and TR3. The 1K thermistor is thermally coupled to the tinned-copper
heatsink to reduce the bias a little when the PA gets hot. More about the
thermistor later.

Alignment
To align, set the 1K0 potentiometer to minimum resistance, apply power to the
'PA-12v' and 'DV-12v' terminals whilst monitoring the current drawn by the
'PA-12v' connection. The current should be next to nothing. Increase the
potentiometer until the current rises to about 50 - 100 mA. That's it!

Coils
I invariably receive loads of e-mail asking me all about the coils I use. "What
was the relative humidity when you wound ..." and "I don't understand what a
ferrite bead looks like ...". Ok, so to answer your questions:
L1 is 6 + 6 turns 28 - 36 SWG wire on two "small" ferrite beads super-glued
together - side-by-side. The two windings are connected in series to form a
single 12-turn coil with a centre-tap. The two ends of the coil are input to the
amplifier and ground. The centre-tap is connected to TR1 BASE via a
capacitor.

L2, L3, L4 are all about 10 turns 28 - 36 SWG wire on a single ferrite bead. If
you use thick wire then only 4 or 5 turns will do. You just can see one of the
coils in the above picture (top). L5 is just an 18 SWG link fed through a ferrite
bead. If you have a bead with a hole big enough to take a turn or two then by

all means add a few turns (it just makes the hammer that bit bigger when
smashing eggs).
T1 is 6 turns 24 SWG wire on two large ferrite beads superglued together side - by side like a pair of binoculars. Use larger beads for this, I used two
large ferrite slugs (8mm Dia.) robbed from an old valve IF can. The secondary
that feeds the final amplifier pair are 1 + 1 turn 18SWG enamelled wire. It
looks like this:

T2 has always given people a lot of problems. It is just 3 + 3 turns 18SWG

wire on a large twin-hole ferrite slab or even a "Pot-core", as I have used. All
work equally well. The secondary winding is 16 turns 22 SWG. It is not the
material that matters, but the physical size of the ferrite. If the ferrite gets hot
then it is too small. You could stack 2 or 3 1/2" ferrite rings on top of eack
other - twice, then place the 2 tubes formed side-by-side to form another
binocular ferrite shape, all held together with superglue. Here is the last T2 I
used, and the one I shall continue to use.

Thermistor (*1K0)

The thermister is connected between the BASE of TR4 and the copper
ground- plane on the top-side of the PCB. It's purpose is to reduce the bias
voltage to TR2 and TR4 when they heat up a little. This reduces the standing
(no-signal) current to the original setting and prevents "thermal stroll-away"
(it begins slowly). In my case I used a 1K0 thermister in series with a 330R
resistor. The 330R resistor damps the effect of the thermistor by lowering it's
dynamic range. The thermistor is placed in direct contact with the tinnedcopper heatsink for the power transistors. If you have problems with the idle
current (no signal) whether hot/cold, then reducing the 330R and reseting the
bias current will lower the "hot PA current". Increasing the 330R will increase
the "hot PA current". The PA standing current should be the same, no-matter
whether TR2 and TR3 are hot or cold.

Construction
The PCB is constructed on double-sided copper-clad board, but only one side
is etched. During assembly, insert each component with reference to the
component overlay. If there is a red circle with a green cross over a hole, then
the component lead must be soldered both top and bottom. If there is no
cross/circle then the top-copper should be countersunk a little with a 3mm
drill bit to remove copper from the edge of the hole. This is clearly shown in
this picture:

The decoupling capacitors in the circuit are somewhat "diagramatic". I really

used the first cap I put my hands on, and there are only two types of
decoupling capacitors; BIG ONES and small ones:
BIG - An electrolytic somewhere between 10uf and 10,000uf
small - A ceramic cap somewhere between 1nf and 330nf

I was fortunate enough to get my hands on a load of ultra-miniature lowvoltage electrolytics in a rectangular plastic case. Use whatever you have got
to hand, it is all the same and the decoupling component values are all VERY
FLEXIBLE. You will notice that in the circuit I show the places that must be
decoupled, but on the real board there are quite a few more. You cannot really
have enough decoupling. I love it!

PCB
A PCB foil drawing is available. Sorry about the quality, but I drew it around
the year 5BC (5 years Before Computers). At that time I used pens and
drafting film to draw all my circuits. Download the PCB foil for the ten watt
PA (38,712 bytes). The size of the copper foil is 134mm x 72mm, but I also
have about 1cm extra as a border all the way ariound the board for mounting
the complete PCB.
The last point is that the heatsink is made of 0.5mm tinned copper plate, bent
into an L shape so that it fits under the three power transistors. I have shaded
the bottom of the L in red on the component overlay. The vertical part of the L
is connected to a much larger aluminium heatsink, for example, the metal case
of your transmitter.

Have fun, from Harry SM0VPO, Lunda, Sweden.

Return to INFO page