Think about Poorman's TRX (filter edition) https://translate.googleusercontent.com/translate...
Think about Poorman's TRX (filter
edition)
June 2005
Added in July 2005
Next is the SSB filter. Ladder type is well-
known as a self-made filter, but here we will
try a half-lattice type filter.
In the case of the half-lattice type, a
combination of crystals with slightly
different series resonance frequencies is
used, but I tried an experiment on how to
insert a trimmer capacitor in series with the
crystal, which was previously taught by
JH2ECU Numajiri.
First, try using a half-lattice type two-stage
connection using four 10MHz crystals, each
bought at 50 yen in Akizuki. Trimmer
capacitors for adjusting the series resonance
frequency are as high as 20 to 30 yen each,
so I attached them only to the upper two
crystals in the figure.
The phase transformer between stages is a
ferrite bead from Akizuki and bifilar 4t.
The input and output are directly connected
to 50Ω.
This is a state where the passband is
adjusted to be the flattest while observing
the input and output characteristics. The
vertical axis is 1dB / div. The -6dB band has
only about 2.0kHz. And no matter how I
went up, it did not spread well.
1 of 13 2/2/20, 10:21 PM
Think about Poorman's TRX (filter edition) https://translate.googleusercontent.com/translate...
This sets the vertical axis to 10 dB / div and
the span to 20 kHz. The skirt is quite wide.
The -60dB band cannot be read from the
figure. Nor is it symmetrical.
I set the span to 100kHz. Is the -60dB band
about 25kHz? The shape ratio is about 12.
Then, I attached trimmer capacitors to all
four crystals.
2 of 13 2/2/20, 10:21 PM
Think about Poorman's TRX (filter edition) https://translate.googleusercontent.com/translate...
When the band is narrow, the inside of the
pass band becomes flat as in (1), but when
trying to widen the band, ripples inevitably
occur. The figure on the left shows 1 dB / div
on the vertical axis, so there is
approximately 1 dB of ripple, but the -6 dB
band has been expanded to 2.8 kHz.
The left figure is vertical axis 10dB / div,
span 20kHz. The spread of the skirt has not
changed.
The span is 100kHz. The -60dB band is
33.3kHz. In this case, the inside of the 7MHz
telegraph band will be completely inside the
-60dB band, so it may be a little hard. But
even if you make a ladder type with four
crystals, the skirt characteristics will be
similar.
3 of 13 2/2/20, 10:21 PM
Think about Poorman's TRX (filter edition) https://translate.googleusercontent.com/translate...
When I bought the crystal in Akizuki, there
were two brands of 10MHz, so I bought 5
each. When measuring both characteristics,
the SIWARD display used in the experiments
so far has fs = 9.99543 MHz and Cp = 4.84
pF, whereas the other MEC AV9B display has
fs = 9.99736MHz, Cp = 3.64pF. In other
words, both fs are shifted by 1.93kHz.
Therefore, if this difference in fs is used, a
band for SSB is likely to be made without TC
for fs correction. The left figure is the
circuit. X1 and X4 are for MEC AV9B display,
with 5pF TC for Cp correction. X2 and X3
are for SIWARD display.
The left figure is 1 dB / div on the vertical
axis. The ripple in the band is as good as
about 0.5dB. The -6dB band is about 2.3kHz.
Although it is a bit narrow, it seems that it
can not be used with SSB.
The left figure is vertical axis 10dB / div,
span 20kHz. TC for Cp correction is adjusted
based on out-of-band attenuation. The
spread of the skirt is not contrasting.
4 of 13 2/2/20, 10:21 PM
Think about Poorman's TRX (filter edition) https://translate.googleusercontent.com/translate...
The span is 100kHz. The -60dB band is
26.3kHz. Out-of-band attenuation is about
65dB.
I moved the TC to bring the notch as close
as possible to the passband. Unfortunately,
the higher notch does not reach -60dB. In
addition, the rebound is quite large, and the
out-of-band attenuation is getting worse.
5 of 13 2/2/20, 10:21 PM
Think about Poorman's TRX (filter edition) https://translate.googleusercontent.com/translate...
Another problem turned out. When the span
was expanded to 500 kHz, it was found that
a considerably large sub-resonance
appeared.
The main culprit of the above sub-resonance
was the crystal with SIWARD display. This is
a view of the resonance characteristics of a
single crystal. I thought it worked, but I'm
sure I should give up using this brand of
crystal.
This is a MEC AV9B display crystal. There
are secondary resonances, but they are
modest.
6 of 13 2/2/20, 10:21 PM
Think about Poorman's TRX (filter edition) https://translate.googleusercontent.com/translate...
This is the situation when X1 and X4 are
extracted and only X2 and X3 are used, and
Cp is canceled by TC1 and TC2. Although it
has a single peak characteristic, it seems to
be quite usable with a simple CW receiver
(aside from the problem of secondary
resonance).
Actually, I don't know how these days, but
the IF filter on the narrow band side of the
spectrum analyzer has the same
configuration.
I bought 100 crystals of 10.695MHz at
Yahoo! Auction for 2000 yen, so I tried half-
lattice three-stage. Since it is not easy to be
full of trimmer capacitors, C for the series
resonance frequency shift was fixed at 100
pF (hidden in the picture behind the crystal).
When examining the relationship between
the series additional capacitance and the
series resonance frequency shift width for a
single crystal,
10pF + 9.05kHz
22pF + 5.34kHz
47pF + 3.02kHz
100pF + 1.43kHz
It was like that.
For canceling the parallel capacitance, we used a 5pF trimmer capacitor that costs 60 yen
each.
Here is the schematic.
7 of 13 2/2/20, 10:21 PM
Think about Poorman's TRX (filter edition) https://translate.googleusercontent.com/translate...
The left figure is 1 dB / div on the vertical
axis. The frequency characteristics in the
band are not flat because the capacitor for
the series resonance frequency shift was
fixed and fixed. Also, the -6dB band is
narrow and only about 1.4kHz. After all, a
trimmer capacitor for adjusting the series
resonance frequency was necessary.
The left figure is vertical axis 10dB / div,
span 20kHz. The area around the skirt looks
like noise, so I can't measure the -60dB
band. Also, despite the three stages, the
shape ratio is not good.
* The measurement was re-measured with
another spectrum analyzer and the figures
were replaced. Since the -60dB band is
7.54kHz, the shape ratio is about 5.
The span is 100kHz. Something like
secondary resonance is coming out.
Adjusting the TC for parallel capacitance
cancellation changes the condition of the
skirt considerably, but as described in the
SSB handbook, it was not possible to
properly set the shape ratio depending on
the position of the notch.
* The figures have been replaced.
8 of 13 2/2/20, 10:21 PM
Think about Poorman's TRX (filter edition) https://translate.googleusercontent.com/translate...
* Something like secondary resonance seems
to be caused by the measurement system.
It was difficult to adjust the notch position
using the TC for canceling the parallel
capacitance, so we decided to insert two
shunt crystals and then explicitly insert the
notch due to the series resonance. One half
lattice. The four crystals use the same
10.695MHz crystal as the three-stage filter
above.
Here is the schematic.
Type Type 4 (and I named it) and the way to
insert TC for parallel capacity cancellation is
a little different, but both work.
9 of 13 2/2/20, 10:21 PM
Think about Poorman's TRX (filter edition) https://translate.googleusercontent.com/translate...
The left figure is 1 dB / div on the vertical
axis. When the ripple in the pass band was
allowed to about 1dB, the -6dB band
expanded to about 2.0kHz. Spread more
than this will make the ripple even bigger.
The left figure is vertical axis 10dB / div,
span 20kHz. The notch by the TC for
canceling the parallel capacitance is set
above the pass band (right side in the
figure), and the notch by two shunt crystals
is set below the pass band (left side in the
figure). The out-of-band attenuation is not
very good because it is one stage, but this
shunt crystal method is FB in the sense that
adjustment is easy to understand.
* The figures have been replaced.
The span is 100kHz. It is not symmetric at
all. hi
With this, the half-lattice part has two or
three stages, and if you add about four shunt
crystals, it seems that good characteristics
are likely to be obtained.
However, the conclusion of this series of
experiments is that the ladder type is likely
to provide simpler and better
characteristics.
* The figures have been replaced.
10 of 13 2/2/20, 10:21 PM
Think about Poorman's TRX (filter edition) https://translate.googleusercontent.com/translate...
When I met Numajiri the other day, when I
reported the results so far, he commented,
"If it is not about 4 steps, it will not have
practical characteristics." At the same time,
I got about 2,000 cheap crystals at a price of
about 1 yen each, so I took the plunge (?)
And tried a half-lattice four-stage. 10MHz
crystal is used.
Here is the schematic. The cost is 86 yen,
but most of it is the price of a trimmer
capacitor.
The left figure is 1 dB / div on the vertical
axis. The -6dB band is about 2.4kHz. There
is a little ripple.
The left figure is vertical axis 10dB / div,
span 20kHz. I think it's a reasonably good
characteristic. The -60dB band is 7.66kHz
and the shape ratio is 3.18.
11 of 13 2/2/20, 10:21 PM
Think about Poorman's TRX (filter edition) https://translate.googleusercontent.com/translate...
The span is 100kHz. It's pretty neat, and you
might be able to use it.
When the capacitance was measured with
the trimmer capacitor removed, they were
all around 35pF, so replacing these with a
fixed capacitor of 33pF is the diagram with
the wider BW in the left figure. The -6dB
band is about 2.52kHz, and the -60dB band
is about 7.03kHz with a shape ratio of 2.79.
The cost was reduced to 26 yen, less than
1/3, because the fixed capacitor was used.
When examining the relationship between
the series added capacitance and the shift
width of the series resonance frequency,
Cx (pF) Δf (kHz)
220 +0.34
100 +0.83
68 +1.14
47 +1.72
33 +2.29
22 +3.26
10 +5.94
6 +7.94
3 +10.74
It looks like
The narrower BW in the figure on the left was tested, considering that if the additional
capacitance is 150pF, the bandwidth will be around 500Hz. The -6dB band is about 650Hz,
12 of 13 2/2/20, 10:21 PM
Think about Poorman's TRX (filter edition) https://translate.googleusercontent.com/translate...
and the -60dB band is about 2.65kHz with a shape ratio of 4.08. When you superimpose it,
you can see that the slope on the right side is translated nicely.
With this method, the band can be used only in the case of USB, but if you add a series
capacitance of about 22 pF to the crystal on the opposite side and put a getter of about 3
kHz on the frequency in advance, LSB It is also possible to shift to the side.
When I tried to switch the capacitor with a diode SW, the shoulder of the pass band of the
filter was drooped and it was BF.
13 of 13 2/2/20, 10:21 PM
