Clackamas Superheterodyne CW Transceiver

FDIM 2010 QRP Challenge Entry by Jason Milldrum, NT7S

Introduction
When I heard in January 2010 that QRP ARCI would be
sponsoring a new design challenge at Four Days In May, I
got more excited about designing a new radio than I have
in a long time. I commend the creation of this 72 part
challenge, as it was one of the most difficult yet rewarding
experiences that I've ever had in my homebrewing and
radio design career. I'm pleased to present my entry into
the contest: the Clackamas superheterodyne transceiver.
Block Diagram
The topology of the Clackamas follows that of a typical
single-conversion superheterodyne. The signal coming
into the receiver front end passes through the transmitter
low pass filter and a DPDT T/R switch. A single-tuned
circuit provides bandpass filtering to knock down strong
out-of-band signals. The desired 40 meter signal is then
downconverted to the 4.032 MHz intermediate frequency
and amplified by approximately 8 dB. A simple Colpitts
crystal oscillator running at 11.059 MHz provides the local
oscillator signal to both the receive and transmit mixers.
Intermediate frequency filtering is provided by a 2-pole
4.032 MHz crystal ladder filter with a nominal bandwidth
of 450 Hz. This filter is coupled via L-network to a simple
bipolar IF amplifier with approximately 27 dB of gain. The
combination product detector/beat frequency oscillator
downconverts the IF signal to audio frequencies while
amplifying it by 10 dB. The audio signal is then fed into a
TDA7052 audio amplifier, which provides ample gain to
Low Pass
TX Filter

Bandpass
RX Filter

BF998
Mixer

drive headphones.
The heart of the transmitter is a N-channel JFET active
mixer, which takes the LO signal and mixes it with a
Colpitts carrier oscillator coupled to the gate. The output
of the active mixer is filtered with a double-tuned circuit
before being lightly coupled to the BS170 power amplifier.
Transmitter filtering is provided by a standard 5-element
low pass filter.
Design Strategy
Even though the rules of the Challenge limited us to one
integrated circuit (without penalty), I knew that I would
need to find a way to include circuits that resembled ICs, in
the sense that they would need to be able to perform or
combine functions in as few parts as possible. Immediately,
it seemed to make the most sense to try to work with dualgate MOSFETs. I've had a large store of BF998 dual-gate
2-pole
Crystal IF
Filter

Beat
Frequency
Oscillator

Variable
Crystal
Oscillator

TX Power
Amplifier

Bandpass
TX Filter

IF
Amplifier

JFET TX
Mixer

Carrier
Oscillator

Keying

BF998
Product
Detector

TDA7052
AF
Amplifier

MOSFETs in my junkbox for many years, courtesy of KE6F

(via W7ZOI's website).

Experimental Methods in RF Design have provided me with an

invaluable education in practical circuits that can get the
job done. His S7C design was especially influential in the
design of the Clackamas.

Over the last few years, I had experimented with the BF998
and was able to successfully use them as amplifiers but
never had much luck getting them to work correctly as
active mixers. Going back to many of my reference books
published in the 70s and 80s, it was easy to find lots of
dual-gate MOSFET circuits, but the biasing schemes
presented never worked correctly with the BF998.

Specifications
General
Frequency Range:

To make a very long story short, after much pain and

suffering I was able to tackle the two main impediments to
implementing my dual-gate MOSFET strategy: biasing the
BF998 for use as a decent performing mixer and figuring
out how to configure the BF998 to function as a
combination product detector and BFO. Once these
challenges were tackled, the rest of the transceiver fell into
place with only a moderate amount of difficulty. It should
be noted that I scrapped the whole idea of using dual-gate
MOSFETs twice so that I could try different design avenues,
but none of the alternative strategies could yield a parts
count even close to 72. Eventually perseverance paid off,
and I was able to get my original design idea to work after
a lot of experimentation.
One of the greatest advantages of using FETs is the
relaxation of the need for high oscillator drive levels. This
allowed me to cut all of the oscillators to bare-minimum
functional units, with no buffering necessary due to the
very light loading from high input impedances of the FETs.

7.029 7.0325 MHz

Receiver
IF Bandwidth:
MDS:
3rd Order DR (20 kHz):
IIP3 (20 kHz):
Blocking DR (20 kHz):
IF Rejection:
Image Rejection:
Current Consumption (13.8 VDC):

462 Hz
-126 dBm
80.5 dB
-5.2 dBm
102.6 dB
23 dB
48 dB
50 mA

Transmitter
Power Output (13.8 VDC):
Current Consumption (13.8 VDC):
Spectral Purity:

1.7 W
260 mA
<-40 dBc

Design Commentary
Front End
A light bit of bandpass filtering is provided by a singletuned circuit in the receiver front end. I was a bit
concerned about the ability of a single-tuned circuit to
provide any meaningful filtering of the very strong
adjacent signals one generally encounters in the 40 meter
band, but the filter worked surprisingly well. Of course, the
conditions on 40 meters tends to be easier here on the
West Coast of the United States than they are in other
places, so this design may not be as suitable in those areas.

I would have liked to have used FETs throughout the entire

rig, but the parts count limitation demanded that I use a
bipolar transistor amplifier in the IF, so I also stuck with
bipolar oscillator circuits since they are cheap and nearly
bulletproof.
Pedigree
It is very clear that the Clackamas can draw its lineage
from the series of minimalist superhet receivers that Doug
DeMaw, W1FB published in the late 70s. While I did not
strive to copy these designs, there is only a small number
of ways that one can configure a minimal parts count
superhet using dual-gate MOSFETs. Let there be no doubt
that the Clackamas was not a simple cut-and-paste job.
While its design may be similar to the W1FB classics, it
took quite a bit of work to modernize the circuits and pare
them down to an absolute minimum of parts.

The standing current in Q6 (the BF998 mixer) is very small;

only a few hundred microamps. When I was working on
the development of the mixer, I tried to increase the
standing current by reducing R9 and providing bias to gate
1. It wasn't difficult to get the standing current to a few
milliamps, but a large problem manifested itself.
Increasing the standing current also triggered a very
significant degradation in the noise figure (especially if
bias was added to gate 1). Reducing the standing current
made the receiver quite deaf. By quite a bit of trial and
error, I was able to find a value of R9 that provided the
sweet spot between excessive noise and insufficient gain.
This is not my favorite way to design circuits, but I still
don't have a firm grasp on all of the nuances of using the
BF998 as a mixer, so using a purely experimental method is
the best that I could do under the circumstances.

Perhaps the greatest inspiration for the Clackamas came

from the W1FB combination product detector and BFO.
Without this critical bit of parts-saving circuitry, I think
that my task would have been nearly impossible to
achieve.
Much credit also goes to the wonderful work of Wes
Hayward, W7ZOI. His work in the homebrewer classics
such as Solid State Design for the Radio Amateur and
2

leg of the carrier oscillator. Normally, directly keying an

oscillator is a big faux pas. However, the CO is so lightly
loaded by the transmit mixer that no noticeable chirp is
detectable during keying.

Crystal Ladder Filter

There is nothing particularly interesting about the 2-pole
crystal IF filter. The filter was designed for a 450 Hz
bandwidth with the xlad program from the Experimental
Methods in RF Design supplementary CD using common
4.032 MHz crystals. The terminating impedance of the
filter ended up being about 225 . T2 provides the
impedance transformation from the Q6 drain impedance
of approximately 1.2 k, while an L-network formed from
C10 and L2 match the filter to the 50 input impedance of
the IF amplifier.

It took a bit of work to find a workable transmit mixer. It

was always my desire to use an active device as a mixer,
but I encountered quite a bit of difficulty in taming the
mixers that I initially tried. Passive mixers were found to
be unsuitable because they provided conversion loss
instead of conversion gain and could not directly drive the
following stages. After quite a bit of experimentation, I
found that a J310 was the ideal choice for this application. I
was a bit concerned that the unbuffered VXO could not
drive the source of Q4, but it turned out to not be a
problem at all. The necessary bandpass filter is
incorporated into the drain of the transmit mix JFET.

Product Detector/Beat Frequency Oscillator

In my view, the greatest breakthrough in the design of this
radio was the completion of the combination product
detector/BFO. As I found though experimentation, the
only reason that this circuit can work is because the IF
signal at RF frequencies is converted down to baseband. In
order for Q7 to be able to oscillate at 4.032 MHz, the drain
has to be bypassed to ground at that frequency (just as you
would expect a Colpitts to be configured) while allowing
the AF signals to remain on the drain. As far as I can tell,
there is no way to configure this circuit to work as a front
end mixer (where you would move a RF signal to another
RF frequency).

Transmit Power Amplifier

I initially knew that I wanted to use a simple MOSFET PA
such as a 2N7000, BS170, or IRF510. The IRF510 was ruled
out almost immediately because the simple transmit mixer
could not drive the relatively large gate capacitance. The
2N7000 and the BS170 are nearly identical, but the BS170
can handle quite a bit more current than then 2N7000, so it
made more sense to use the BS170 as a PA (as evidenced by
many of the new QRP kits coming on to the market place).

Audio Amplifier
The audio amplifier is so dead simple there is very little to
say about it. The selection of the TDA7052 was a virtual nobrainer, since I knew that it needed very little supporting
circuitry in order work. It turns out that I was able to get it
to function reasonably well completely on its own, with no
decoupling on the VCC rail. There's more than enough
audio gain from this amplifier. You can't beat 40 dB of
audio amplification and 1 watt of available AF power for a
grand total of 1 part.

Although I wanted to drive the PA in class C mode, I found

that I could not extract much power from the amplifier
configured in class C in this circuit. Through some more
experimentation, I found that PA required a bit of gate bias
in order to produce a reasonable amount of power output.
Blue LED D1 provides a stable gate bias voltage under
varying power supply voltage changes. In order to achieve
maximum power transfer, the double-tuned transmit
mixer bandpass filter also needed to be adjusted to
compensate for the input capacitance of Q2 by reducing
the value of C22.

Variable Crystal Oscillator

There's nothing particularly interesting about the VXO. It
is a standard Colpitts VXO with polyvaricon tuning.
Because it only needs to drive high impedance FET gates,
there's no need for buffering. Drive levels to the receiver
and transmitter mixers were set by experimentally
changing the coupling capacitors (C20 and C29)

T/R Switching and Sidetone

It would be nice to have QSK, but realistically I knew that
wasn't likely in a minimal parts rig. Therefore, the T/R
switching is a simple DPDT switch. One pole switches the
receiver path to the antenna out during transmit. The
other pole switches power to the transmitter and receiver
sections appropriately.

Carrier Oscillator and Transmit Mixer

The carrier oscillator is nearly an exact copy of the VXO,
with two important exceptions. A large inductor was
needed in series with the crystal in order to pull the
oscillator onto the right side of the IF filter skirt. Since I
ran low on parts, I found that I was able to leave the
normal 100 nF decoupling capacitor off of the collector.
Once again, due to the fact that the CO is driving a FET
gate, no buffering was necessary.

The sidetone is a pure hack but it works surprisingly well.

Quite a bit of time was spent trying out different muting
schemes where I would attempt to let a small bit of the
transmitted signal back into the front end so the actual
signal could me monitored as the sidetone. I never could
get this level to anything lower than a roar, so I had to
approach the problem differently. In a brainstorm, I
figured out that I could just cut power to all of the receiver
stages except for the product detector/BFO and audio
amplifier. Now, the product detector picks up the stray

Transmitter keying is provided by grounding the emitter

3

carrier oscillator signal and downconverts it to a nice

sidetone which accurately reflects the offset between the
transmitted and received signal frequencies.

A Suggestion
I really enjoyed the FDIM QRP challenge, but would like to
suggest one rule change if QRP ARCI decides to conduct a
similar restricted parts count contest in the future. Please
allow for any tuned network in a transmitter output stage
to count for a maximum of three parts, no matter the
actual parts count. This rule comes from the Minimal Art
Session in Germany, and would allow the designer to not
have to worry about trading off spectral purity for a
critical feature elsewhere in the radio.

Attaining Zen
Not surprisingly, I was forced to evaluate the necessity of
every single part in the rig. Sometimes as designers we will
do this exercise on a circuit block, but rarely is it necessary
to perform it on the entire radio. One excellent side effect
of the requirement to do this is that I have learned much
about what is really absolutely vital to include in each
circuit, what provides nice but marginal performance
improvements, and what is superfluous. I have no doubt
that this hard won knowledge will be very useful in the
future.

Finally...
I wanted to spend a few moments to praise all of the
wonderful design tools that I used in the creation of the
Clackamas. All of the major tools that I used are free, open
source software. This includes the Ubuntu 10.04 operating
system, the wonderful OpenOffice.org suite, and TinyCad
schematic capture. Where I didn't have open source
software, I was able to run programs such as the extremely
useful EMRFD tools under WINE Windows emulation. It is
truly amazing how today's designer and builder has access
to professional quality tools for little or no money.

What Would I Put Back In?

Now that the Clackamas is cut down to the bare bones,
there are some things I might add back in if I wanted to
increase the operating conveniences and improve the
performance.

More effective decoupling would be one of the

first things on my list. I was able to remove quite a
bit of the decoupling with a minimal performance
hit, but I would feel much more comfortable with
the proper decoupling in the radio.

It would be great to stick with using the BF998

throughout the Clackamas. If the IF amplifier was
replaced with a BF998-based version, the receive
current consumption could be reduced below 10
mA.

Two poles of crystal filtering is a bit light for my

taste. The biggest problem with the 2-pole filter is
that the ultimate attenuation is only about 50 dB,
so you can hear strong nearby signals. A 4-pole
filter would be an excellent substitution.

Due to the very light bandpass filtering on the

front end and the fact that the mixer amplifies
everything that gets to gate 1, the IF and image
rejection is pretty poor. In order to fix this, I
would change this filter to at least a double-tuned
circuit, perhaps even triple-tuned.

The simple keying and muting in the Clackamas

works very well for the parts count, but semi-QSK
keying and muting would be a nice convenience
feature.

The tuning range of the VXO is a bit restricted, so

the addition of a second parallel 11.059 MHz
crystal, plus perhaps a series inductor, would give
a much greater tuning range.

ICs

Bill of Materials

U1
Subtotal

Resistors
(all 0.25 W)
R1,R8,R13
R2
R3
R4
R5
R6
R7,R11,R15
R9
R10,R14
R12
R16
Subtotal

X1,X2,X3,X5
49)
X4
49)
D1
S1

10k
1.2k
2.2k
22k
100k
150
1k
470
150k
100
220

Subtotal

1
1
1
6
2
7
1
1
1
2
1
2
1
1
1
1
--30

1u
220u
10n
100
65 trimmer
100n
180
150
22
220 (50 WVDC)
560 (50 WVDC)
68
270 polyvaricon
4.7
33
47

3
1
1
2
--7

2N4401
BS170
J310
BF998

1
1
2
1
1
--6

220u (molded)
T37-2 30T
T37-2 17T
FT37-43 10T
100u (molded)

1
1
3
--5

FT37-61 6:20
FT37-61 16:7
Toko 42IF123

Inductors
L1
L2
L3,L4
L5
L6
Subtotal

Transformers
T1
T2
T3,T4,T5
Subtotal

4.032 MHz (HC-

11.059 MHz (HC-

Subtotal

1
1
--7

Total

===
72

J1
stereo)
J2
J3
stereo)

Transistors
Q1,Q3,Q5
Q2
Q4
Q6,Q7

Blue LED
DPDT

Not included in parts count

(all 25 WVDC min. unless specified, all values pF

unless specified)

Subtotal

TDA7052

Miscellaneous
3
1
1
1
1
1
3
1
2
1
1
--16

Capacitors

C1
C2
C3
C4,C5,C13,C23,C25,C27
C6,C26
C7,C8,C9,C14,C15,C19,C30
C10
C11
C12
C16,C18
C17
C20,C22
C21
C24
C28
C29

1
--1

Phones (3.5 mm

1
1

BNC
Key (3.5 mm

t


t

t










t


!






( )t$t*

#

!

$%&'

!

$%&'


t

tt
t

t

t












t


t

!
0(t

 

+

#
t
t







,
t



"t












t


 
t,
t

0:?4$(A61$: <

t

t


 



#!
tt
,$%&'

6>$%=

#$#/


t


$ 1
t
#

tt
t



t
#

t
#
t
#

"
t!
t

( )! $t




-

t

+



t

+!










$64 46$$ ($$ 6<?

4:$9$: <

$!
0(t
$0($4<:$4>$4$64 46$:>?

!
0(t

t
t



t

t
t






#


.t

#$#/
t
#






t

t





t


4:$%=

t
#

31

.t-







t


!

 t

4$/64

t











!
0(t

#
!

$%&'

;$.96



445$14$/64$7#/8



-.t




t


!!
t 


"
t

"!
 )
$t 






"
 )
$t 

t

t
-$"2


!

"9$4$4:$(


,
!

#$#/



t








+

+

!!
t

6>$-4? 4$(

t!
t



t





?6$6;

-4$(A61$/64

-(,,

,
!

?$: <





+
!
t

-(,,

#t
!

$%&'





t


!!
t
t




t

+t

t

($ 4

( )t$*



"

 )
$




t


(0%$

t
$+$4$)$B6$464:4

4$%?:
 .
(
*C4:$(C1C(0%$

t
$4 ?
>
4
t

$ $

t

#$#/

14$"4??$(
!
$&'$-@

t
t


t
#

t


t


t

t

t
#

6:
.
t$<$t