This is the last construction section (Yeaaaaaaa!)
The last part of the kit is the final RF amp and output filter.


Transmitting without a resonant antenna or dummy load will cause high
SWR and can damage the final transistor. This is why we saved this
section for last ;-)

An outstanding description of this amplifier and it's function can be
found on Glen's home page at:

Before I read that page, I would have described Q6 as a class C
amplifier. Glen shows that the lines are a little blurry when defining
an amplifier in a class. Glen modeled Q6 as a simple switch, on for one
half cycle of the 7 MHz and off for the other. As you will see the model
comes really close to what was actually measured.

I'm not sure about this, but it seems to me that with Q6 either full on
or full off that it's power dissipation must be fairly low. This is also
reflected in the fact that Q6 has no heat sink. Operating my SW-40+ in a
normal QSO, Q6 barely got warm to the touch.

As for the output filters, they are described as a "5 element Chebyshev"
filter. Back in March Chuck posted and excellent tech. description that
I am going to include here. The following was done by Chuck Adams and
parts by Paul Harden.

****************** Forwarded text

March 26th posting by Paul Harden on low pass filters prompted this
posting. In fact, so that I don't have to type in the graphics
again, here is Paul's rendition of the Five Element Chebyshev Filter
as used by Dave Benson, NN1G, in the NE4040 and SW-40+:

-------------------------cut here-------------------------------

<From Paul Harden, NA5N>

For those who don't have the schematic of the OUTPUT FILTER ...

RF from PA ---C36--+---- L3 --+----L4 ---+-----> to antenna
| | |
C37 C38 C39
| | |
gnd gnd gnd

C36 is a 0.1uF and is a simple coupling cap, that is, blocks DC from
the final transistor (PA) from the filter components.

L3,L4 and caps C37, C38 and C39 for the output filter, which is
TWO low pass filters glued together (hopefully with solder!).
The values are:
L3, L4 = 1uH (16 turns on a T37-2 toroid)
C37,C39 = 470pF
and C38 = 1000pF

-------------------------cut here-------------------------------

At the desired frequency of operation, if you will make the impedance's
XC(37)=XC(39)=50ohms, XL(L3)=XL(L4)=50 ohms, and XC(38)=25 ohms
to get 50 ohms to 50 ohms impedance match. You can see why C38 has
twice the capacitance of C37 or C39.

Here is a table for L and C values that will give you 50 ohm
reactance at each of the most popular QRP frequencies.

Freq [MHz] L [uH] C [pF] for end caps and double value for center
---------- ------ -------
3.560 2.23 894.1
3.710 2.14 857.9
3.579 2.22 889.3
7.040 1.13 452.1
7.110 1.11 447.7
10.106 0.78 315.0
10.116 0.78 314.6
14.060 0.56 226.4
18.080 0.44 176.0
21.060 0.37 151.1
24.910 0.31 127.8
28.060 0.28 113.4

Now the C values are not exactly off the shelf standard values,
so you try for the closest you can get. Let's use the
40 meter frequency of 7.040MHz for an example. Dave chose
470pF for C37 and C39 and 1000pF or 0.001uF for C38 and
L3=L4=1.02uH. This because 390pF is close on the low end
and 470pF is closer to the desired value of 452pF and
0.001uF is closer than 820pF for C38. See Paul Harden's
book for the standard capacitor values. (Paul, I'll get
the commission check later. :-) ) These values are also
the same values recommended by Wes Hayward in the SSD book.
(Solid State Design for the Radio Amateur an ARRL book)

For the inductors, we get for the two popular toroid sizes
used in QRP work and the two most popular cores:


Core --> T37-2 T37-6 T50-2 T50-6

# turns [uH] [uH] [uH] [uH]

1 0.00 0.00 0.00 0.00
2 0.02 0.01 0.02 0.02
3 0.04 0.03 0.04 0.04
4 0.06 0.05 0.08 0.06
5 0.10 0.08 0.12 0.10
6 0.14 0.11 0.18 0.14
7 0.20 0.15 0.24 0.20
8 0.26 0.19 0.31 0.26
9 0.32 0.24 0.40 0.32
10 0.40 0.30 0.49 0.40
11 0.48 0.36 0.59 0.48
12 0.58 0.43 0.71 0.58
13 0.68 0.51 0.83 0.68
14 0.78 0.59 0.96 0.78
15 0.90 0.67 1.10 0.90
16 1.02 0.77 1.25 1.02
17 1.16 0.87 1.42 1.16
18 1.30 0.97 1.59 1.30
19 1.44 1.08 1.77 1.44
20 1.60 1.20 1.96 1.60
21 1.76 1.32 2.16 1.76
22 1.94 1.45 2.37 1.94
23 2.12 1.59 2.59 2.12
24 2.30 1.73 2.82 2.30
25 2.50 1.88 3.06 2.50
26 2.70 2.03 3.31 2.70
27 2.92 2.19 3.57 2.92
28 3.14 2.35 3.84 3.14
29 3.36 2.52 4.12 3.36
30 3.60 2.70 4.41 3.60

Dave went for the 16T on a T37-2 for a value of 1.02uH.
He uses 1.00uH in print and that is close enough due to variations
from core to core on the A(L) value.

I did a quick SPICE simulation and here is my first recommendation
for a SW-40+ mod. You knew the mods were coming, didn't you?????
Add one more turn to L3 and L4, thus 17T, and this will increase
the second order attenuation by an additional 4dB without
significant penalty at the fundamental. And I even modeled in some
additional distributed capacitance due to packed turns and it
seems to be for the good in this case. I get 34.85dB down at
14.080MHz for 17T vs 30.79dB down for 16T. Now as soon as some of
the people get theirs built and can get them to a lab with a
high-dollar spectrum analyzer they might take the time to try it
both ways and tell me if their is any difference. Probably not enough
to worry about, but it would be interesting to look at theory vs
real data. The SW-40+ etc. all work great as designed and built
from the instructions, so this recommendation may be ignored.
Due to other factors yet to be studied (see below) the input
impedance of the final PA may influence this effect.

Now those that are going to move their SW-40+ rigs to the Novice
frequencies on 40M will start to ask questions on what mods do they
need to do? The answer is none, as the filter is not critical for
the small change from the low end of 40M to the higher end.

In the latest issue of QRP ARCI Quarterly you will see a graph
of attenuation curves from a bunch of filter values that I took
from all the W1FB books and Dave's filter. NN1G's is the solid curve.
:-) ;-) The only one that beat it was the theoretical maximum.
Good job Dave. I'll try to put the curve on the web page for the
Elmer101 notes later on Friday.

Those of you that went to the trouble of getting SPICE up and running
on the computer can check out what happens when you vary some of the
values. I'm working on a series of what happens on the input side
due to input impedances not being purely resistive and 50 ohms.
Also what happens on the antenna side for variations also.
So much data and so little time. :-)

One more thing to look at is relative to postings that I see where
people 'squeeze' the windings together for these type filters and
see power output increase from a rig into a dummy load. Personally,
IMHO, this is a dangerous practice. Into a dummy load, the second
and third harmonics (which may increase) contribute to the forward
energy detected by the SWR bridge and or power meter. BUT, into
a real antenna you may see the reflected power increase at the same
time, thus showing that the effect is not what you really want.
Again, this is something else to experiment on. Remember, mileage may
vary. You have to take lots of data and that's what makes it all
interesting and challenging.

For those that like to read:

"Ferromagnetic-Core Design & Application Handbook" by Doug DeMaw, W1FB,
published by MFJ Publishing Company, Inc. Starkville, MS 39759,
MFJ-3506, $19.95

"Simplified Practical Filter Design" by Irving Gottlieb, published by
TAB Books, Blue Ridge Summit, PA 17294-0850, ISBN 0-8306-3355-3,
$16.95. (This one I picked up for $8.48 at Half-Price Books)

and of course the ARRL HB and Paul Harden's book.

************************* End forward

That's everything, so........

Wind L2, L3 and L4 torroids as per the instructions in the manual.

Remove the temporary jumper from J1 and install the proper 5k RF gain

Gather and install the following components:

L2 as per instructions
L3 as per instructions
L4 as per instructions
C36 .1 microfarad
C37 470 pf
C38 .001 microfarad
C39 470 pf
C113 .1 microfarad
D12 33 volt zener diode
Q6 2SC2078

You will have some capacitors left over. That's ok. They were included
by Dave for the VFO calibration.

Check your work. Be aware of the polarity of D12 and Q6. Don't install
them backwards.

Connect a 50 ohm dummy load to the ANT connection and ground. I just
soldered some wire to an SO-239 connector and soldered the other ends
into the ANT and ground holes. The antenna connection should go to the
center conductor and the ground should go to the shield. I then
connected the SO-239 connector to the dummy load in the shack. If you
have a QRP watt meter you can put it between the rig and the dummy load
to check your RF power out.

If you don't have a watt meter, connect your RF probe or oscilloscope
across the dummy load to view the output.

Turn it on. You should hear static from the headphones. Turn your RF
gain to maximum. Key the rig. You should see your watt meter move to
about 2 watts and hear a nice side-tone. O-scope users should see about
30 volts peak to peak of RF. You can adjust the R24 potentiometer to
vary the power out. I turned mine up all the way and the waveform got
distorted. I backed it off until it looked clean and then turned it down
some more. Now I'm measuring 30 volts p-p which is about 2.25 watts out.
Remember that peaking the power into a dummy load is not the best way to
get the most power. You may be increasing one of the harmonics and the
actual useable power may go down. Also your rig may not be in FCC spec
for spurious emissions. If I had a spectrum analyzer I'd like to see how
the harmonic output changes with R24 position.

In the forwarded text from Chuck (k5fo) he states that the spurious
outputs can be improved (reduced) by adding a turn to L3 and L4. Does
anyone with a spectrum analyzer want to do this and post the results?

Well, that's it. If your power out looks good, put it on your 40 meter
antenna and make a contact. My first two contacts were Texas, then Cuba.
Not bad for 2 watts ;-)

I put my board in the case from a computer A/B switch box. I mounted the
RF gain, tune pot, phones and key jacks in the front. On the back I put
the antenna SO-239 and the DC power jack. Right now I don't have an
on/off switch but I think I may add one to the RF gain pot.

I used a 10 turn Spectrol linear pot for tuning. I also got one of those
small 10 turn indicators. I thought it would be easy to calibrate, but
it seems that the varactor response is not linear. The frequency
difference between major divisions is 2kHz on the ends and 5kHz in the
middle. So all I did was make a little chart on the front of the rig
that had the operating frequency for every one of the 10 major
divisions. Enough to get me close. I can tune between 7.015 and 7.050
Mhz. I think the first addition to this rig will be a Small Wonder Labs
"Freq mite" for a CW frequency indicator.

This radio is a joy to use. It is small and uses very little power, yet
the receiver is very sensitive.

Well, that's it folks. Thanks for sticking with me. I can say personally
that I've learned a TON of stuff about this radio and QRP rigs in
general. I'll post a time and date for an "Elmerfest" on the air. Let's
get them finished up and on the air! Think of that as "graduation". Then
on to the Elmer 200 series.