CMOS RF SYNTHESIZER
by Harry Lythall - SM0VPO
One of the most interesting projects to work with
is that of a synthesizer that can accurately gnerate desired frequencies. This
synthesizer uses just three CMOS chips and one PNP transistor. It is a fact that
the transistor may be the hardest component to get in some parts of the world!!
The synthesizer will generate a 3:1 frequency range anywhere from 300Hz to 4,000,000Hz
(4MHz). Using the components shown you can feed the synthesizer with up to 12vDC
but with only 9vDC the synthesizer may not achieve more than 3.5MHz. Tye lock-range
of this synthesizer is about 10:1 or so.
The Reference Oscillator
The CD4060 reference oscillator divides a crystal
reference oscillator sucessively by two. From each stage we can select a frequency
that is half the value of the previous frequency. The reference frequency of the
synthesizer is multiplied by the programmeable divider divide rate to give the
final frequency. If our reference frequency is 1KHz then the final output may
be programmeable in 1KHz steps. This is quite convenient for a decimal programeable
divider. Divide by 1000 = 1000KHz, divide by 1001 = 1001KHz etc. You could also
select a crystal that would give you a 100Hz output then you could select frequencies
in 100Hz steps. Divide by 1000 = 100KHz, divide by 1001 = 100.1KHz etc. Here are
reference frequencies you could use for different oscillator crystal frequencies.
CRYSTAL/REFERENCE FREQUENCIES (FROM 4060)
Remember that the programable divider will only
divide up to 15999 and this limits the maximum frequency you can get out of
this synthesizer. With a 100Hz reference frequency you can only get a maximum
of 1.5999MHz. With a 1KHz reference frequency you can program up to 15.999MHz
which is beyond the range of a CD4046. Incidentally, a 32,768Hz wrist-watch
crystal will give you 2Hz so the synthesiser will generate from 6Hz up to 31,998Hz
in 2Hz steps.
The 4046 has a wide-range Voltage Controlled Oscillator
(VCO) which uses C1 and R3 to determine the frequency. C1 is selected by LK2 and
I used 33pf, 47pf, 68pf and 100pf. R3 is selected by LK5 and I used 4K7, 10K,
33K and 100K. The 10K resistor from pin 12 is an offset resistor which may be
omitted. I only used it to push the VCO up to 4MHz which is about the maximum
you will realistically get from a CD4046. The 74VHC4046 is supposed to go up to
someting like 20MHz.
There is no reason why you cannot use an external
VCO, such as a VHF oscillataor with a divide by ten prescaller. This should
enable you to generate up to about 159.99MHz in 10KHz steps.
The output of the VCO is divided by a programmed
figure in the 4059 IC. I just used two 8-pin DIP switches on the PCB but there
is no reason why you cannot use "BCD thumbnail" switches. Each digit is in the
form of Binary Coded Decimal (BCD) and so the range of each digit is 0 to 9. The
1000's digit however may be programmed up to 15 (F in hexadecimal) so the maximum
divide rate is 15999. The minimum divide rate is 3 so that with a 100Hz reference
oscillator the minimum output is 300Hz.
The reference oscillator and the output of
the programmable divider are both fed into the twin phase detectors in the 4046.
One of these outputs (pin 13) gives a HIGH, LOW or TRISTATE (high impedance) levels
which indicates whether the programmable divider output is above, below or equal
to the reference frequency. This is used to control the VCO frequency and complete
the loop. There is a second phase detector but this is little more than an Exclusive-OR
gate. I have used this to give a loop locked indicator via the LED and the BC557
The output of the phase detector is fed via R1 (10K,
33K, 100K and 330K selected by LK3) to a capacitor to form the loop filter. If
the loop filter time constant is too short then FM modulation will be removed.
If too long then it will take too long for the loop to lock onto a new frequency.
There is also a dampling resistor R2 (1K, 3K3, 10K and 33K) which is selected
by LK4. The loop damping resistor should be included to prevent loop over-correction.
This would result in the VCO constantly swinging above and below the correct frequency.
The PCB does not give any particular output socket
since it was designed for training dealer staff and was only to be viewed with
an oscilloscope. A high-level buffered output is available from pin 10 of the
4046 VCO but this must have a load resistor of 10K to ground. The output is a
square-wave with odd harmonics extending throughout the HF spectrum. Here is a
spectrum display from the prototype - 0 to 20MHz:
As you can see, there are both odd and even harmonics
throughout the HF spectrum with the odd harmonics being the strongest. This
means that you will need to filter the output if you intend to use it for transmitting
or receiving. A closer view of the spectrum, also at 1.8MHz centre frequency
shows the output to be relatively clean:
The completed synthesizer PCB is 108mm x 64mm and
has only battery (and output?) connections. I fitted all three ICs to sockets
even though modern sockets cost more than the CMOS chips (I am VERY old- fashioned,
I even like to see girls in black seamed stockings!).
Because this unit was used for training I needed
to adjust some of the settings so I used header sockets and U links. In this
way I could select any clock frequency, VCO parameters and loop-filter settings
I set the frequency by means of two eight-pin DIP
switches. Note that the place values of the switches are 1 2 4 8 from left to
right EXCEPT the fourth (most significant digit) which is backwards: 8 4 2 1.
Let us assume I want to generate the frequency
1.815MHz. With a 4.096MHz crystal I set the following links and DIP switches.
Note that the MHz switch is backwards.
|LK1 = 1KHz (Q11)
||LK2 = 33pf
||LK5 = 4K7
||LK3 = 100K
||LK4 = 10K
||1 & 3 = ON
||2 & 4 = OFF
||1 = ON
||2 3 & 4 = OFF
||8 = ON
||5 6 & 7 = OFF
||8 = ON
||5 6 & 7 = OFF
LK3 and LK4 are entirely optional. If LK4 is omitted
and R2 is made a total short circuit then the circuit will never stabilise but
oscillate up and down the band. LK4 should be a mirror of LK3. LK3 sets the loop
time constant. If you add DC blocked AF to the VCO voltage (4046 pin 9) via a
270K resistor then you will be able to generate an FM signal. With LK3 set to
10K then AF below about 1KHz will be removed (the loop feedback will see modulation
as a frequency error). With LK3 set to 1M0 then the loop will take an eternity
to stabilise but will give good quality modulation down to to about 100Hz.
A printed circuit board is available but I drew the
artwork before I had a computer so I used drafting pens with old fashioned ink
on good old tracing paper. I have scanned my PCB Artwork
which works, but I have not bothered to tidy it up. I have not yet drawn a component
overlay as I think that it should not be too difficult for anyone of modest intelect
to identify the component locations. Anyway, you can not complain, it's for free!!
Did You Know that ...
To make an RF VCO that works in the VHF spectrum,
forget about buying those expensive VariCap diodes - you don't need them. Ordinary
high voltage Zener Diodes work. A 20v zener diode operated at 1v to 13.8v exhibits
a capacitance of typically 10pf to 2pf as the voltage is varied. I have even used
this effect to make a 5.0MHz - 5.5MHz VCO for this synthesizer, just as an experiment.
Have fun, de HARRY
Lythall - SM0VPO
[ About me
| Acronyms | CW
| Data Sheets | Docs
| Download | E-mail
| HOME | Ham
projects | Hobby circuits
| Photo galery | PIC
in my guestbook | View
my guestbook ]
© 2001 - YO5OFH, Csaba Gajdos