A lot of postings about caps for Sierra band modules lately.

If you want an easier way and use the original band modules, do the mod
that's suggested by the rigs designer. Replace the rf choke feeding the
collector of the final to a 4 to 1 impedance ratio transformer. Replace
the final with a MRF237, or as I did with an inexpensive  RCA SK9618,
a 12.5 watt rf transistor which can be heat sinked to the back of the cabinet.

You can key down for as long as you want and doesn't even get warm.
Install the drive control to the back panel as described in the Sierra manual,
so you can run at any power you want.

Runs at 5 or more watts on 160,80 and 40, about 4
watts on 30 and 20, and 3watts on 15 and 10. Easily and smoothly
adjustable to zero out with drive control. All band modules original. I
love this little rig and use it most of the time. TS440 and TS130 haven't 
been used in over a month.

MORE THOUGHTS

It has been over a year since my article on increasing Sierra power
output was published in QRPp -- You do subscribe, don't you?

This Module cap thing is getting out of hand!  I conducted over 40 tests
on Sierra power with some guidance from Wayne, and reported that the S.M.
modifications made a minimal difference -- mostly on the lower bands
where large capacitance values are required.

If you are intent on replacing the original caps, be forewarned that on
80 & 40M, the values are high enough where a "normal voltage value" of 300V
won't fit on the module.  So -- the dilemma is this:  where a less lossy cap
is needed the most, the values and resultant size are the largest.

I personally think you would do just as well to replace the small caps
with larger voltage ones, even in ceramic.  I believe Dave Benson has
also made some remarks about all of this in the past, and Paul Harden has
also commented on usage of various capacitor types.  Don't get hung up on
silver mics!  (Which aren't manufactured anymore -- but that's another
story).

NOW THE GOOD STUFF:

As Dave Fifield noted in his post today, reworking the Xmit. Pi filters
on the modules made a BIG difference in power output.

Why is this?  Wayne designed the Sierra to produce 2W, average, output.
That was his design parameter.  When we muck around to increase power,
we changed his design.

Dave Meacham, in QRPp Sept. 95. stated the problem well.  " Output
filters are designed for a specific termination, usually 50ohms....the
proper resistive component of the load should approximate the
collector-voltage squared, divided by two times the power."

So...when we crank up the voltage, or increase the power output, the
resistance changes....and it goes lower.  One way to deal with this
mis-match is to reduce turns on L5 & L6 to find the best impedance for a
proper load to the final.  The results, as both Fifield and I have
stated, can be dramatic. The original work on this mismatch was noted by
others in early 1995 and served as the starting point for my own tweaking
experiments.

Is this the best solution?  No, probably not.  But its the solution I
originally reported.  Wayne has always stated the proper solution is a
matching transformer which replaces the rf choke feeding the final
collector.  I had already modified all my modules for maximum power
output for "my components and voltage choice", and didn't want to start
over from scratch -- so I didn't go back and pursue it, but others have.

If you have not played with module output filter tweaking, then a much
less time consuming option is the 4:1 matching transformer.  Roy, AB7CE,
published the results of his modification yesterday, and reported 5W on
the lower bands, 4W on 30 and 20M, and 3W on 15 and 10M. (With a final
change to an SK9618).

Vic Turner, on Jan 19th, posted a full explanation of building the step
up transformer.  "I simply removed the collector choke and substituted
10 bifilar turns wound on a T50-43 ferrite core.  This will step the 14.4
collector z up to near 50 ohms!  Also note that he had made other mods.
toensure adequate drive.  His solution was driver replacement with a 2N5109.
Other variations have been previously posted.  If someone needs more
details, I am sure he would be happy to reply.  His address is:

victnr@juno.com.

On last word.  If you made modifications, and still feel you are not
getting "proper" power output, re-adjust BOTH the XMIT BPF and PMO BPF
trimmer caps on each module for max. power out.  Are you getting two peaks on
the trimmers?  If not, the combination of component is too far out of spec,
and you have more work to do....

Got so many requests on this, Putting it on the list.
Change the rf choke feeding the collector of the final(L10) to a 4 to 1
impeadance ratio transformer. I used the same core material,-43, as the
original choke, but used a 0.5 instead of a .37 inch diameter core.Wind
with 12 bifiliar turns of #24 wire. Form a center tap by connecting the
start of one winding to the end of the other winding. Connect the tap to
the finals collector. Connect one of the other ends to the 12volt line
and the last lead to the input of the low pass filter. I installed in same
location as old choke. One lead to C46 on top side left lead as viewed
from frontof rig. Center tap to hole for collector trace.
Other lead to hole near C45.

IMPORTANT: installing in original position as I did requires cutting the
trace at C46 to collector, isolating C46 only.
You can use a MRF237 and sink it in original position. I used a RCA
SK9618, which has the emitter as the center leg and mouting tab. Since
the final has emiter to ground, no insulator is needed to heat sink
directly to back of cabinet with a little heat sink paste.
There is a convenient ground at PCB to solder emitter to, so only two
leads are needed to go to base and collector.

The MRF237 will run hot, but you can beacon with the SK9618
and maybe gets warm. I used a 500ohm pot to the rear panel, to replace
the drive control as described in the Sierra manual. I notched a hole big
enough for the two wires to it to run under the board.
I ran the two transistor leads on top.

You will need this control to back off of 6 to 7 watts on
160,80 and 40 meters, or to run at low current levels. With control peaking
transformer wound on the same core.  You can also experimentally stretch
or squeeze toroid windings in the low-pass filters--just check your work
with a spectrum analyzer, because you may compromise the return loss or
spectral purity of the signal by doing this.
(On the other hand, you may make it better.

Toroids seem to have as much as a 10% swing in permeability,
and that's *before* everyone winds them their own way :)

As I've pointed out before, the Sierra is designed to meet cost, size,
and ease-of-construction goals, and therefore does not have a complicated
transmit strip.  I designed for the 2-3 watt level at 20 meters.  Power
output is usually higher than this on the lower bands, lower on the
higher bands.

Of course the S/N ratio is typically higher on the high bands
(because they're usually quieter), so you might consider the lower power
and reduced current drain as a feature for portable operation   :)

But if you really want a 5-watts-on-all-bands Sierra with no tweaking,
you'll have to re-design the transmit strip.  I have built such a strip
on a protoboard; it takes about 30 to 40% more parts than the existing
transmitter design.  (It won't fit in the same space without doing
something I have sworn NEVER to do to the kit builders of the world,
some of whom have marginal vision, and that's use surface-mount parts.)

Basically, what it requires is more gain stages, each with feedback,
resulting in a more "level" frequency response from 2-30Mhz.  It's also
helpful to include ALC or some way of automatically controlling power
output in response to load or drive changes.  You also have to be very
careful not to cause problems with excess in-the-box radiation near
sensitive circuits.  Again, this would be dificult at the Sierra'a size
and price point; you'd probably have to add either a small additional PCB or
some shielding.

The NPO ceramic and C0G caps I specified for the Wilderness Sierra have a
very decent dissipation factor, measured at .005 to .001, which
translates to a Q of 200 to 1000.  Yes, S.M. caps are more consistent:
in my experience, usually around Q=700+.
C0G-rating caps are typically better than NPO ceramic disc.

However, even a capacitor whose Q is at the "low" end of this range (200)
is *not* going to make a noticeable dent in the output power, as Dave
Fitfield reminded us recently.  Just to convince myself of this, I have
on many occasions replaced 5% NPO or C0G caps with S.M. and -- true enough
--  no difference in power output. (Exception:  you may just happen to find
a cap that is closer to the optimal value for a particular filter.)
Now I *did* use some low-Q caps on the original NorCal (club project)
Sierra; they were 10%, high-K units, definitely to be avoided as they
have Q's as low as 40 and wider capacitance tolerance.  This was my mistake,
corrected on the very first Wilderness run.  Always stick with 5%, NPO
or C0G, and watch the voltage rating.

If you have a Sierra and want more power output, you have a few choices.
You might follow my suggestions in the manual, especially the two
easiest:

(1) increase supply voltage; (2) replace the output choke with a 2:1
transformer wound on the same core.  You can also experimentally stretch
or squeeze toroid windings in the low-pass filters--just check your work
with a spectrum analyzer, because you may compromise the return loss or
spectral purity of the signal by doing this.
(On the other hand, you may make it better. Toroids seem to have as much as a 
10% swing in permeability,
and that's *before* everyone winds them their own way :)

As I've pointed out before, the Sierra is designed to meet cost, size,
and ease-of-construction goals, and therefore does not have a complicated
transmit strip.  I designed for the 2-3 watt level at 20 meters.  Power
output is usually higher than this on the lower bands, lower on the
higher bands.

Of course the S/N ratio is typically higher on the high bands
(because they're usually quieter), so you might consider the lower power
and reduced current drain as a feature for portable operation   :)

But if you really want a 5-watts-on-all-bands Sierra with no tweaking,
you'll have to re-design the transmit strip.  I have built such a strip
on a protoboard; it takes about 30 to 40% more parts than the existing
transmitter design.  (It won't fit in the same space without doing
something I have sworn NEVER to do to the kit builders of the world,
some of whom have marginal vision, and that's use surface-mount parts.)

Basically, what it requires is more gain stages, each with feedback,
resulting in a more "level" frequency response from 2-30Mhz.  It's also
helpful to include ALC or some way of automatically controlling power
output in response to load or drive changes.  You also have to be very
careful not to cause problems with excess in-the-box radiation near
sensitive circuits.  Again, this would be dificult at the Sierra'a size
and price point; you'd probably have to add either a small additional PCB or
some shielding.

Change the rf choke feeding the collector of the final (L10) to a 4 to 1
impeadance ratio transformer. I used the same core material,-43, as the
original choke, but used a 0.5 instead of a .37 inch diameter core.Wind
with 12 bifiliar turns of #24 wire. Form a center tap by connecting the
start of one winding to the end of the other winding. Connect the tap to
the finals collector. Connect one of the other ends to the 12volt line and
the last lead to the input of the low pass filter. I installed in same
locationas old choke. One lead to C46 on top side left lead as viewed
from front of rig.

Center tap to hole for collector trace. Other lead to hole near C45.

IMPORTANT: installing in original position as I did requires cutting the
trace at C46 to collector, isolating C46 only.
You can use a MRF237 and sink it in original position. I used a RCA
SK9618, which has the emitter as the center leg and mouting tab.
Since the final hasemiter to ground, no insulator is needed to heat sink
directly to back of cabinet with a little heat sink paste.
There is a convenient ground at PCB
to solder emitter to, so only two leads are needed to go to base and
collector. The MRF237 will run hot, but you can beacon with the SK9618
and maybe gets warm. I used a 500ohm pot to the rear panel, to replace the
drive control as described in the Sierra manual. I notched a hole big enough
for the two wires to it to run under the board. I ran the two transistor
leads on top. You will need this control to back off of 6 to 7 watts on
160,80 and 40 meters, or to run at low current levels.
With control peaking drive
at about 90% I get 3 watts out on 10 and 15 meters and 4 watts on 20 and 30
meters. All band modules original. The SK9618 is the RCA replacement for
MRF260. Also ECG342. Same specs all around, but got SK for $4.25

A while back I posted a output power mod for Wilderness Sierras.
Left out a couple of details than many have asked about. I used a
FT-50-43 ore, but the original FT-37-43 core works just fine. I have lots of
both, the larger one was just a little easier to work with.
The bifiliar transformer is wound with parallel turns, no twist.
Should have put this in original posting. Sorry about that.

The past Friday I completed the output mods on my Sierra which you
recommended...I replaced the L10 collector choke with a 4:1
transformer (I used 10 bifilar turns on a FT- 50-43), replaced Q7 (PA)
with an NTE 342 (couldn't locate an SK9618 at a reasonable price), and
mounted a 500 ohm drive control pot on the rear panel.

I did modify two modules:  I reduced the number of turns on L5 & L6
from 26t to 23t, & installed silver mica caps on my 10 meter module
(120 pF for C47, C49 & 220 pF for C48) - I'll eventually swap out the
rest of the caps.

The results (at 13.8 V measured with an HP8565E Spectrum Analyzer with
a 5W, 20 dB attenuator on the input):  7 W on 80M, over 5 W on 40M,
30M, 20M & 15M and 3 W on 10M.  [It helps to work in a Design
Engineering Lab]  The external drive control makes power adjustment
very easy.

Note:  the NTE 342 (cost around $5.25) is the same configuration as
the SK9618 - the center pin is tied directly to the tab and, thus can
be mounted directly to the inside of the rear panel to provide both
grounding and a heat sink.  I used one of the existing ground "holes"
on the Sierra PCB for the center pin.

Roy, thanks to you, Bob Follett, Vic Turner (for sharing your ideas)
and to Brad Bradfield (for the caps).  I look forward to a Sierra to
Sierra QSO with you.