± 0.5 dB on band 8. This is due to the built-in AGC circuits
of the MC1648. I did not do an FFT on the output, but after
looking at scope sine wave presentations for over 40 years I
have developed a pretty good eye for distortion, and would
judge spurious and harmonics to be a good 30 dB below
the output level. That is a clean sine wave and very
respectable for this type of generator. LC oscillators do not
have the long term stability as their crystal referenced “big
brothers” do, and usually will drift 500 ppm or more over a
given time. Although this one is no exception to that rule, it
out-performs those figures by quite a bit.
After a one hour warmup, I performed many tests for
short-term stability of 15 minute periods. I did this for each
band at the low, mid, and high ends of their ranges — 24
tests in all. The best case was 2 ppm drift in that time and
the worst case was 98 ppm. The rest varied from 20 to 80
ppm, with 50 ppm being a good overall average. So, I could
conservatively spec the overall short-term stability at 100
ppm which is quite good for an LC oscillator. I was also
pleasantly surprised that I could easily inject a signal into the
input of a narrow band land mobile receiver at 155 MHz
and hold that input for quite a while — although it did show
a sensitivity to load changes in those frequencies.
The external modulation input was set arbitrarily at 800
mV P-P for all inputs. In the AM mode, 50% modulation can
be obtained before soft clipping and then hard clamping
occur. This function has a good linearity figure of 3% and
about 15 mV P-P for each percent of modulation sensitivity.
The clamp circuit prevents the MC1648 from damage due
to accidental over-voltage of input here. Without the clamp,
it has the same linearity up to 90% mod. Unless you are
feeling lucky, do not bypass this feature. The standard test
signal for AM is 30%, so this is achieved and then some.
The FM mod input is attenuated quite a bit in the
summing amplifier and is not so prone to over-voltage. At a
full 800 mV P-P, it will produce a deviation of approximately
0.5% of the carrier frequency. A calibrated FM deviation is
very difficult to achieve and would require twice the
circuitry as the entire generator with specially machined
switches, so it was not even considered for this design.
However, if you have a few “pet” frequencies that you
would like a known deviation for, there is one method of
achieving this. Set the generator to the carrier frequency
desired and carefully inject a DC voltage of zero at the
junction of C7 and R7 on the control board, with the mod
control set for FM operation. Note the frequency and slowly
increase the DC voltage until the carrier frequency increases
by the amount of deviation desired. An AC P-P voltage that
is the same as that DC voltage will now give the exact
amount of deviation you want. If you do this often, a rear
panel jack connected to that point would make things
easier. Just keep in mind that the DC eventually winds up
on the Vt line to the varactor because there is no blocking
capacitor in that path.
There are a couple of changes that I made that are not
shown in the images. On the RF board, I extended the leads
of R10 to 1/2” on both ends so as to mount it higher off the
board, and bent it slightly away from the board. This resistor
dissipates one watt, and was dissipating a lot of that heat
into the copper foil. This is the one exception to the “short
leads” rule. The added inductance of the leads will only
affect band 8 by increasing the MMIC’s load impedance.
The effect is so minimal that it can be ignored. It still
couples some heating to the board and — along with the
upside down mounting of the MC1648 adding yet more
board heating — they will raise the board temperature about
13 degrees F above the work area’s ambient.
Of course, some of that heat is generated by
transformers, regulators, and such. This is why it takes about
a full hour warm-up to stabilize. However, after a couple of
minutes, the generator is ready for use. Just let it warm up
for an hour before doing critical work such as IF alignment,
etc. The other change was adding a 1” square of thin
aluminum to the 7824 regulator tab in the power supply as
it tended to run too hot without this.
I did not design an RF attenuator in this unit to keep its
footprint as small as possible. Years ago, I bought a really
low priced HP attenuator on eBay. It will attenuate 0-130 dB
in 1 dB steps, and with HP quality. I find that I use this
attenuator even with equipment that has one built in due to
its wide bandwidth and precision.
In my sweep generator article mentioned previously, I
show construction of an attenuator that won’t bust your
wallet. It has decent performance to upwards of 200 MHz
and about 5% accuracy. This is a four step 0-40 dB in 3 dB
steps. This could be expanded to 0-100 dB in 1 dB steps by
using 27 5% 1/4W resistors and nine DPDT mini switches,
plus some double-sided circuit board for a housing at a cost
of maybe $30. The switches could be mounted vertically in
a single column on the left side of the panel.
Either way, you would have to add another 1-1/2” to
the panel width to accommodate these. (I could include this
in the email packet.) If you are desperate for an attenuator,
the “cheap and dirty” version shown in Figure 2 will suffice
as far as more power/less power goes. Also, there will be an
insertion loss of 3 dB (for the MMIC protection) and it will
not be calibrated. For five dollars in parts, it will get you by.
As to what I might change in the future: replacing the
fine-tune pot R2 with a three-turn pot for even finer
adjustment; and adding an internal tone oscillator with 800
mV P-P output. Then, I’d connect this to the front panel jack
via a miniature SPDT toggle switch.
I am always changing designs for the better as new
ideas and components become available. I do have an
outboard clock generator design for this generator included
in the email packet. It runs at exactly half the frequency of
the RF output feeding it, and has a perfect square wave with
rise and fall times in the low nanoseconds. When doing this
project, take your time and double-check connections as
you proceed. Most of all, have fun! NV
34 June 2014
The frequency counter was omitted from the images for clarity and is an article in
itself. N&V have given me the green light on this and I am in the mid stages of
design. So, hopefully, that article will follow on the heels of this one. Stay tuned!