Author: Tom Rauch - W8JI
Clicks are often a problem on congested bands. They are most
problematic when we try to copy weak signals next to moderately strong
signals. While a fast rise and fall time guarantee excessive
bandwidth, a long rise and fall is no guarantee a radio will be
"click-free". Some radios switch into transmit while the synthesizer (VCO)
circuits are still settling to a new frequency. These radios generally
produce a loud "thump" on key closure that happens to be right on the
DX station, when the operator is working split. If the operator is
using QSK, VCO switching thumps can be particularly annoying. The
thumps will occur every time the VCO moves from the receive frequency
to the transmit frequency.
Many radios have rise and fall times that are much too fast, but
how fast is much too fast? For now let's ignore VCO switching
problems, and consider envelope shape.
The ARRL recommends 5 mS rise and fall times for CW, based on data
in section 2.202 of FCC rules and CCIR Radio regulations. According to
professional sources, 5 ms rise and fall times are not harmful to
readability at 35 wpm under marginal (fading) conditions, and 60 wpm
when signals are reasonably far above noise floor. This rise and fall
results in a occupied bandwidth of 150 Hz, although unwanted transient
energy caused by the shape of the waveform slope appears at wider
bandwidths.
Two things come into play; the slope of the envelope rise and fall
at any point controls bandwidth of keying sidebands, and the amount of
voltage change during that slope controls the power level of the
sidebands (clicks). The shape (bandwidth) and amount of signal level
change in a slope area (level) combine to determine how offensive the
transmitted signal is. Very subtle changes in envelope shape can have
a profound effect on key click amplitude. This makes it difficult (if
not almost impossible) to determine whether our radios are as clean as
they should be when considering only overall time required for a CW
envelope to reach full power levels.
Reference Data for Radio Engineers, in the section of Radio Noise
and Interference, addresses key clicks in a manner the ARRL Handbook
does not. They give an example of multi-pole shaping of waveform.
While this would amount to perhaps a $5 parts addition, virtually
everyone ignores the roll-off we could have on all CW signals!
Here are the bandwidth curves of three basic envelope shapes, one
rectangular (some radios are this bad!), one for a proper single pole
R/C filter with slightly rounded shape (the ARRL suggests this shape
probably because it was practical in the early years and "stuck" even
though it is not ideal), and one for a filtered rise and fall (this
would be a sine-shaped rise and fall from a multi-pole filter). We can
clearly see a large difference in bandwidth in the curves below:
From Ref Data for Radio Engineers 29-10 1977 Edition
FT-1000MP Measurements
I measured two sample FT-1000MP's (an early and a late model) by
operating them into a high power fixed 30 dB attenuator pad. The
output of the 30 dB pad was connected through a 3-way splitter to a
step attenuator and conventional receiver, a spectrum analyzer, and an
oscilloscope. The receiver used a 300 Hz eight-pole filter, the
spectrum analyzer used a 50 Hz filter, and the scope was triggered
from an external keying signal. Power was measured on a conventional
Bird average reading meter.
At 1kHz spacing clicks from the stock FT1000MP's were about 15 dB
worse than clicks from my old test bench radio (a well-worn ICOM
IC-751A) and more than 20dB worse than the clicks from my
click-reduced FT-1000D!
Here is a spectrum display of my stock IC-751A using 30 Hz analyzer
bandwidth and ten second sweep:
The 751A is approximately -58 dB at 1 kHz, and rolls off smoothly.
Here is the nicely sloped (but too fast) 751A rise:
Rise approximately 3 mS....and the fall (which is too sharp at the
upper corners):
Now the stock 1000MP spectrum:
The FT1000-MP is approximately -50dB at 1kHz. It is 8 dB worse than
the already "hard" 751A, and has a "click plateau" below the carrier
frequency that hovers around -55dB for around 500Hz bandwidth.
In direct comparison, here is a "de-clicked" FT-1000MP:
The modified FT-1000MP rivals any of the better radios I have
tested, including my "de-clicked" FT-1000(D).
The modified FT-1000MP is around -85 dB at 1 kHz, over 30 dB
improvement from the stock MP! Rise time is close to ARRL standards of
5mS, while fall time is around 3 mS. FT1000MP modified rise:
Rise 6 mS. The upper edge is a little sharp, but why
worry....clicks are reduced 30dB or more!
Modified MP fall:
Fall time is around 3mS. While it has much more rounded edges, the
slope is still not very "round". Unfortunately we are limited by what
is possible to do, and this mod is already difficult enough for
laymen.
Some concerned was expressed over the "power" of dots when using
long rise and fall times. One simple solution is to turn up the weight
control slightly. Keep in mind, even without ANY external weight
adjustment, the change in average dot power at 45 WPM was only a few
percent! On the air tests with VK3ZL and ZL3REX on 160 meters with
fading signals and noise, revealed both could tell absolutely no
difference between having the click filter in-line and out-of-line at
40 WPM CW speed. This waveform meets FCC and CCIR specifications for
60-WPM CW modest strength signals, and 35 WPM weak fading signals.
Doing The Mod
The ideal CW radio would use a high-order filter with controlled
group delay, and a reasonably linear attenuator or modulator to
control the envelope shape. All other stages should be fully on just
outside the output window of the CW signal. I initially hoped a CW
"modulator" could be added on, but for now it appears modifying the
1000MP to ideal circuitry would be too involved. My only option was to
"hunt and peck" and find a modification that would be reasonable to
do, and inexpensive. This is the best solution I could find, reduction
of clicks was excellent. The only drawback is two resistors need to be
hand-selected, and the radio needs some minor disassembly to reach a
connection point on the RF board.
I mounted this mod on a separate terminal strip under a screw on
the left front corner of the IF board. This allowed me to experiment
with component values while watching bandwidth and other parameters.
This is the basic circuit I used:
C3 was a .1uF disk capacitor. This component's value turned out to
not be especially critical, it mainly prevents fast rise and fall of
the low-level RF amplifier stage that is driven by a gate. There was
no combination of resistance across or in series with this capacitor
that reduced clicks in any of the radios I tested.
C1 and C2 are also .1uF disc capacitors. In all units tested, I
could find no better combination for reducing clicks.
The only critical components appear to be R1 and R2. R1 ranged from
120k to 470k in the units I tested. R2 ranged from 1k to 10k ohms. I
initially clipped in potentiometers, so I could listen to the output
and adjust the clicks at 1kHz spacing. Both pots were adjusted for a
null in click amplitude. That null is rather sharp, and turned out to
be around 30 dB deep. This takes the 1000MP from being one of the "clickiest"
radios I have found to one of the cleanest!
The best method of nulling clicks is by listening on another
receiver with a narrow filter. Make sure you are well below overload
on the receiver, and set that receiver so the carrier from the MP is
just outside the passband of the test radio filter. It is almost
impossible to use any other method for proper adjustment, including
watching the envelope on an oscilloscope.
Work in a clear uncluttered location, I like to work with the radio
on a clean small carpet on a well-lit bench, and have a container for
all the hardware I remove.
Here's how to make connections to necessary points:
- Remove the top and bottom covers of the radio and set them
aside.
- Invert the radio, so you have the heatsink exposed.
Four main screws hold the heatsink mounting bracket. Two screws
are shown above (one under the screwdriver and one afew inches to
the right of it). Two more screws are on the side of the radio
chassis. You might want to remove the long screws holding the fan
bracket, although I got by without doing so.
- Lay the fan and PA assembly out of the way, you may have to open
some of the wire harness clips to get more wire. Unplug the fan so
it is totally free from the unit. It should look like this now:
- The RF board is the green-colored board you see above. There are
several screws holding that board down, and two screws on a rear
panel DIN connector that is mounted on that board. The board will
freely move when you remove ALL the screws. Do NOT pry or force the
board out, if you have to pry you missed a screw!
- Flip the board over, you might have to unplug a wire harness or
two...but I managed to work without doing that. The board should
look like this:
- I added the green wire you see above. I tacked it on a foil pad
by laying the wire across the point where two chip components
soldered in, as shown below:
You can see the little black FET (Q1034) and the slightly
triangle-shaped foil trace that connects to Q1034. I bent an "L" in
the small wire I used. The wire will route topside to a terminal
strip, so it needs to be several inches long.
- The new wire routes under the RF board to an oval slot in the
chassis, where if feeds to the other side and emerges near the IF
section unit. Be careful not to pinch any wires when remounting the
circuit board. The opening on the right is best for getting the
added wire topside:
- While re-installing the RF section, fan, and heatsink inspect
the wiring carefully. Be sure nothing is touching moving parts of
the fan, and be sure no wires are pinched or left unplugged. It
might be advisable to check the radio quickly on a dummy load to be
sure the transmitter section works properly.
- Flip the radio over, and remove the two ribbon cables connecting
to the IF section:
- Remove the mounting screws and flip the IF board exposing the
bottom:
This is the area where the wire attaches to the IF board:
Note the FET above and to the left of the pen, and the IF
transformer (two transformer shield connections and five electrical
pin-outs, with an unused pad-set for a surface mount device in the
center of the transformer leads) below the pen. The connection point
for a new small insulated wire is the chip capacitor (C2148) pad
that also connects to the center pin of the top three in-line pins
of the IF transformer.
- Attach a small wire to the point mentioned above. It will route
to the new circuitry.
- Re-install the board. Be sure you do not pinch any wires. Be
sure you do not forget to plug in the ribbon cables, or any other
wires you removed.
I mounted a small four lug (plus ground) terminal strip at the left
front corner of this board, and mounted the components on that strip.
The lead lengths and dress are not critical, so the additional
circuitry can be added where you like.
As mentioned earlier, all you need to do now is adjust the two
resistor values for minimum clicking in another receiver tuned
slightly off frequency (be sure you do not overload it) while sending
a string of fast dots. You should be able to obtain a very large
reduction in off-frequency clicking.
|
