SOME VXO AND "SUPER VXO" EXPERIMENTS
6/DEC/2004
UPDATED 17/12/2004
click over the photo for a higher resolution one
I needed a VXO for a simple CW XCVR for the 20m band, so I built the following circuit:
C1, C2 - 150pF
C9 - 300 pF air variable
L1 - 5 uH wound over T50-6 toroidal core
This circuit is known as"Super VXO", invented by JA0FAS and JH1FCZ in 1980, it's nothing more than using 2 or more equal xtals in parallel.
For a start I chose R1 to R5 for 10 mA quiescent current on each transistor. I used OrCAD PSPICE SV, but there's plenty of software to help with this, it can also be done with a hand calculator, it 's really very simple. The choice of this value of current came after having a look at BC548 and J310 transistors' datasheets. BTW, a 2N3904 can replace the BC548. Later the quiescent currents can be modified; in the oscillating transistor the current can be lower to decrease internal heat thus enhancing frequency stability. The FET buffer can have its current adjusted by means of R4 and seeing the output under load with an oscilloscope.
This basic circuit worked well. L1 = 5 uH and C9 = 300 pF, I tried 1, 2 and 3 crystals (14060 kHz) in parallel.
Results:
1 crystal 14052 a 14063 delta F = 11 kHz
2 crystals 14045 a 14065 delta F = 20 kHz
3 crystals 14033 a 14065 delta F = 32 kHz
I also tried a 10 uH toroidal coil and compared it to a small RF choke. the choke gave awful results, output decreased from 10 dBm to -4 dBm and stability was compromised. As this inductor is part of the tank circuit, it must be of high quality. The problem with this small choke is the high series resistance (I measured 13 ohms on DC, it's more on RF) and the use of a high permeability ferrite core, which is bad regarding changing inductance with temperature, so compromising stability.
I used a BB112 varicap, as it allowed the needed frequency excursion. The maximum reverse voltage for this part is 12 Volts, I chose to feed the circuit with 10 Volts, using a LM317. Beware! the circuit must be fed with no less than 12 Volts, otherwise the LM317 won't regulate! Another alternative is to choose a low dropout regulator.
It's interesting to note that we can lower the frequency from the marked crystal value, but we cannot make it higher (but only a tad amount, as a crystal is specified for a given load capacity, decreasing this can make it oscillate a bit higher on frequency). What we really can do is to put series inductance and parallel capacitance, lowering the frequency.
The 68 ohm resistor gave a better spectral output and less power variation at the output. This resistor controls the positive feedback, without it we can only act on the two equal 100 pF capacitors.
I used 2 varicaps facing each other on the main tuning. When only one is used, for half of the cycle it conducts, this increases phase noise. Moreover, output power changes with tuning. These were all measured facts. The RIT circuit is in a point under lower RF voltage, that's why I put only one diode there.
The 10 uH inductor was made with 46 turns of # 30 wire over a toroidal core T50-6. Frequency shift was 13,998 to 14,052 MHz. With a 50 ohm load I measured 8 Vpp on the drain of the J310 and 10 dBm (10 mW) at the output link. The broadband transformer was made with 10 turns on the primary and 3 turns on the secondary over a FT50-43 ferrite toroidal core.
For the intended aplication there will be two outputs. One for feeding a NE602 mixer, the other going to a 74HC240 buffer. Instead of the broadband transformer, RC coupling wil be used and resistors values will be chosen to suit apropriate excitation levels.
