Tuning VFOs With A PN-Junction
Introduction
Air variable tuning capacitors are becoming increasingly expensive and are only sold by a few vendors. In addition, they are often bulky and may require reduction gear for fine tuning. In a high performance project, it may be silly to skimp and not use an air variable capacitor for tuning the VFO as they normally offer excellent frequency stability and linearity for tuning an LC circuit. Alternately one may use VVC diodes which offer the best frequency stability performance of all the diodes shown on this page. Some authors are developing methods to even better stabilize VVC diode tuned VFO circuits and let us hope they publish their data so that we can all benefit with more stable, high grade VVC tuned VFO or VCO designs. However, in many trivial and popcorn projects, using silicon diodes or the PN-junction of cheap BJTs to tune the local oscillator is very practical.
This web page will explore using the PN-junction as the variable capacitance in a VFO tank circuit. bipolar transistors, silicon, zener and VVC diodes will be discussed.
Please note these were simple experiments and do not reflect scientific methodology or results. Your results may vary and this page is really meant to get fellow Home Builders thinking about and performing their own experiments. The examples may not best reflect practical circuits, however are used to illustrate concepts.
A supplemental page with some practical examples can be found on the PN-Junction Tuned VFO Supplemental Page.
The Test Oscillator
All of the various PN-junction tuning configurations were evaluated in the test oscillator shown in Figure 1. The basic principles and rules of stable low-drift VFO design are still required for the best possible frequency stability. I have covered many of these at the bottom of the following web page from the 40 meter superhet project VFO schematic. All the VFO stability hints except for number 6 apply.
When building a PN junction tuned VFO, it is probably best to build the entire VFO, ensure that it is working and then add your tuning circuit. This way you can ensure the circuit oscillates and is stable before you add more circuitry which may make troubleshooting more difficult in the event of problems. I built my test VFO with just the first and second 51 pf NP0 capacitor soldered in place and watched the output in the scope and on the frequency counter. I saw a sine wave and observed the frequency to be ~8.6 MHz. I then soldered the third 51 pf NP0 ceramic capacitor in place and watched the frequency drop to ~7.1 MHz. This is perfect as I wanted my VFO for the 40 meter band and knew that the added tuning circuitry should drop the frequency to somewhere around 7.0 MHz. I listened to the VFO output as an audio tone on my commercial superhet tuned to 7.1 MHz after connecting a 1 meter piece of wire to the emitter of Q3 for an "antenna". The audio tone after about 5 minutes was quite stable considering the oscillator was not encased in an air tight enclosure. This was also confirmed with the frequency counter. Is is unlikely that the exact same frequency would be measured in a reproduction of this circuit, however it should be close.
The tapped Hartley was chosen for reproducibility and because it is a rare-to-fail design. The voltage across the tank circuit in a Hartley oscillator is many volts peak to peak and I wonder if there is a better VFO design for using tuning diodes with. This web page created much more questions than answers I am afraid.
L1 was wound and then boiled in water for ~6 minutes. It was glued to a single-sided circuit board with Polystyrene Q-Dope from GC Electronics. The coil does not lay on copper as I used a motor tool to grind away the copper in the shape of a circle with a diameter slightly larger than the wound coil. The inductor was doped face down on this circuit and covered with 2 separate coats. It is vital that your inductor is secured so that it is unmovable. A small trimmer variable capacitor for setting the lower band edge frequency would normally be added to this circuit. This was emitted for control purposes when testing the various semiconductors for frequency range in the Figure 1 circuit. The Q2-Q3 buffer has too high an output voltage for 7 dBm diode ring mixers. For diode rings, a more suitable buffer may be used and two examples may be found on this web site.
Device Tuning Linearity
In a PN-junction, the P layer of the junction is the anode and the N layer is the cathode. Consider a normally biased diode. Current flows easily through the diode when the anode is positive and the cathode is negative. Tuning a VFO with a PN-junction means that we actually change frequency by changing the voltage applied to that PN-junction with a potentiometer. This voltage is applied as reverse voltage, meaning that the polarity of the voltage is such that it makes the P-type semiconductor material from the PN-junction negative and the N-type material positive. The resulting low current flow is referred to as reverse or leakage current. In VVC diodes this leakage current is very very low, making them the best choice for voltage variable capacitors, however almost PN-junctions will work as a voltage variable capacitors if the builder is willing to experiment and match the device used to the performance and cost required.
A great article by Hugh Wells, W6WTU appeared in the now defunct HAM Radio for June 1990. Hugh tested many types of diodes and charted their capacitance versus reverse voltage from 0.5 to 16 volts. His diode capacitance test circuit was shown in his article and I used it last year to conduct some of my own experiments. Hugh's original article spawned many of the experiments that eventually evolved into this web page.
The reverse voltage versus capacitance response curve for a given device is often graphed on a piece of graph paper. The most linear portion on the graph is then determined and the control voltage is aimed to fall between the lowest and highest voltages from the graph that gave the most "linear" part of the response curve. Many home builders do not have a sensitive capacitance meter and cannot plot a reverse voltage versus capacitance response curve, however, I found it may be practical to plot a voltage versus frequency graph for a given VFO and use that instead. All that is needed, is a frequency counter and VOM. The response curve graph can be done for any PN-junction device you use as a variable capacitor in your VFO circuit. The advantage of a voltage versus frequency graph is that the device under test can be measured in the circuit you are using it in and factors that influence the capacitance of the device may be accounted for. The disadvantage is that the diode will rectify some of the oscillator AC voltage and what you will most likely be measuring is the total DC voltage, not just the applied DC voltage.
The reason you want to aim your control voltage within the most linear portion of the response curve is so your your tuning will be smooth and linear like an air variable capacitor. If you do not do this, quite often you may have most of your effective tuning change of the VFO in a small portion of your potentiometer's total turning range. This may make calibration or fine tuning exceedingly difficult.
Changing or adding resistor values to set your minimum and maximum reverse voltage values is an easy way to put your VFO in the most desirable operating portion and help ensure linear tuning. In some cases, you may simply desire a large frequency range and may not care about how linear the tuning is. The best part of homebrew radio is the design choice is up to you. Using good quality linear taper potentiometers or more expensive 10 turn versions may be used for best results.
In Figure 2, is a voltage versus frequency graph for the Figure 3 circuit which uses the base and collector of a 2N3904 as the PN-junction. The 47K resistor that connects the 50K pot to ground and the 11.7 volt zener diode were removed when conducting the plot to allow the minimum and maximum tuning voltages shown to be used. You can see from the Figure 2 graph, that from a control voltage of 3 to 4, there is terrible linearity. The best overall linearity falls in the 7 to 12 volt range. Thus, for the best tuning linearity in the Figure 3 circuit, your minimum and maximum control voltage should therefore fall somewhere within this range. This principle will become more self evident as you read through this web page.
Bipolar Junction Transistors
VVC diodes may not be in your junk box. For popcorn VFOs, if you really want to save money on a parts order, bipolar transistors such as the 2N3904 may be used as voltage variable capacitors for a popcorn receiver project. Only two leads on the transistor are ever used and the redundant lead maybe cut off. Voltage variable capacitors using one, two or four BJTs may be used depending on the tuning range and popcorn factor you desire. It is obvious that the Q of a BJT tuning diode will not be as high as those seen in better VVC diodes, but they are practical alternatives for very low-cost, simple receiver projects. I would probably not use a BJT-tuned VFO to tune a transmitter project as the potential for drift is high, however reasonable stability is possible with careful design and a bit of luck.
Consider Figure 3. For brevity sake, the VFO buffer stages will not be shown any more but were used in all of the Figures to follow. A single 2N3904 is used as the variable tuning capacitor in this experiment. Note that there are two voltage regulators used on this schematic. If you were really going frugal, only one could also be used. The voltage regulator on the nJFET of the Hartley oscillator is used to lower its operating voltage to minimize the effects of internal heating on this device which can cause drift. Also keeping the RF amplitude down in a PN-junction turned circuit is probably a good idea as large AC voltage swings may cause the PN junction of our tuning device to conduct and reduce Q and waveform purity. You can always build up the oscillator output voltage with buffer/amplifiers.
The second zener diode voltage regulator is used on the voltage control circuit of the tuning circuit. Unregulated voltage for the tuning device whether it is a transistor or diode is not recommended as fluctuations in DC supply voltage will change the frequency of the VFO. Additionally, your DC tuning voltage must be free of any ripple, hum or noise riding along the DC voltage. AC noise if present, can sweep the oscillator frequency at a rate consistent with the ripple frequency. Clean DC power may be achieved by decoupling your VFO power supply from RF with RF chokes, resistors and capacitors. It is also good practice to ensure that correct polarity to your VFO circuit is present at all times.
You can not eliminate the series 100 pF NP0 ceramic capacitor connecting the 2N3904 to the tank circuit or the VFO will not oscillate. You can however, experiment with this capacitor value.
Figure 3 contains the minimum tuning voltage and frequency measured in the test VFO. The capacitance of the 2N3904 will be at its maximum level at the minimum control voltage. The frequency and voltage with the pot turned all the way to the left is 7.036 MHz and 5.82 volts respectively. The minimum control voltage is set by the 47K resistor connecting the 50K pot to ground. Raising this value will raise the minimum control voltage and lowering or eliminating this resistor will lower the control voltage. When the 47K resistor was removed, the measured voltage was 3.42 volts which gave a minimum frequency of 6.977 MHz. From Figure 2 we learned that the minimum voltage should be 7 volts for maximum linearity in the VFO tuning response. I decided to accept 5.82 volts as a trade off to illustrate that in many cases you may compromise the minimum tuning voltage to suit the standard value resistors you have on hand or to attain a bigger tuning range for your VFO. In many design scenarios, compromises may be made to suit your needs and parts collections! The maximum control voltage is largely set by the 11.7 volt zener diode. If you wished to lower this value series resistors or a voltage divider could be added as well. The 11.7 volt zener was chosen as I may want to use this VFO with a 12 volt battery and this zener diode would facilitate such operation. The maximum control voltage and frequency with the pot turned all the way to the right is 7.053 MHz and 11.88 volts. The 2N3904's capacitance would be at its minimum value for this circuit at 11.88 volts. The Figure 3 VFO tuning range is 17 KHz, with the control voltage going from 5.82 to 11.88 volts.
The stability of this VFO is fair. I have a commercial kit receiver that has worse drift, but again I would hesitate to use this VFO design for a transmitter.
One of the drawbacks of using just a single tuning diode or BJT is that if the device is forward biased by the RF signal during part of the AC cycle, it's reverse leakage will increase momentarily and it's Q will be reduced. Also, harmonic energy is produced as the tuning diode or BJT is alternately biased positive and negative which results in reduced VFO output waveform linearity. The solution is to connect two tuning diodes or BJTs back to back with the reverse DC voltage applied to both devices simultaneously. The two tuning diodes will be driven alternately into high and low capacitance and the net capacitance will remain constant and not be affected by the AC signal amplitude. The circuit in Figure 3 could be improved by adding another 2N3904 back-to-back with the existing one, but the tuning range will be reduced. This occurs because you have now connected two "capacitors in series" and the total capacitance has been reduced and may be analyzed by the classic capacitors in series equation we all learned in radio school.
Consider Figure 4.
Experiments were conducted on this circuit and the minimum and maximum frequency and corresponding measured DC voltage on the BJT emitters for each was recorded. In Figure 4 as shown, the lower frequency/DC voltage was 7.054 MHz and 1.08 volts and the upper frequency/voltage was 7.062 MHz and 6.52 volts.
In the red insets are a schematic showing a four BJT version of this circuit and a plot of the reverse voltage versus frequency of the Figure 4 VFO without the series 22K and grounding 10K resistor ( This accounts for the frequency differences between the plot and the final VFO ). As shown, the best tuning linearity is above 3 volts, however, that would limit the tuning range too much and I chose components to give a voltage swing from 1.08 volts to 6.52 volts. With this particular arrangement the maximum voltage was limited with the 9.1 volt zener and series 22K resistor as the zener (breakdown) voltage on the EB junction of a 2N3904 is close to 8 or 9 V and you never want to lose the reverse bias and go into breakdown.
Silicon and Zener Diodes
Silicon diodes are widely available and inhabit almost every junk box. I have tried using many types of diodes as voltage variable capacitors, but some types, such as rectifier diodes have a better range of variable capacitance than others. As stated before, reversed-biased diodes will decrease in capacitance as the voltage applied to them is increased. Do not expect the Q and thus the stability of a tuned circuit using silicon diodes to be that of better VVC diodes in a similar circuit. One particularly practical example of an application for using silicon diodes is for BFOs. In a BFO, there is typically a crystal and the desired frequency change is relatively small making silicon diodes attractive as the tuning device.
Earlier, I referred to W6WTUs HAM Radio Magazine article. From my own experiments, I confirmed Hugh's readings and found that rectifier diodes, are probably the most practical and easy silicon diodes to use as voltage variable capacitors. As far as rectifiers diodes go, the 1N4001 is probably the most commonly used in this application and a web search of HAM Radio electronic sites found its use in three different oscillator projects. From my experiment, the 1N4001 may exhibit a capacitance swing from 35 pF to 4 pF with a corresponding minimum and maximum reverse DC voltage of 0.5 and 12.0 volts. If you increase the voltage above 12.0, a lower capacitance may be realized but its significance is limited as it is getting into the non-linear area of the voltage versus capacitance curve and above normal QRP operating voltages. In addition, the linearity of the change in capacitance versus reverse voltage change is quite poor under ~5 volts for the 1N4001. I plotted the response curve for the 1N4001 and the results indicated that the best control voltage range to use for VFO tuning linearity is 5.0 to 12.0 volts.
Experiments with some diodes in my junk box were tried in the Figure 6 circuit. The Figure 6 circuit is the identical test oscillator circuit illustrated in Figure 1 and is used throughout this page. The only difference is that the 2 zener diode voltage regulators have been added. The various diodes were connected to the oscillator by a series 100 pF NP0 ceramic capacitor at point A.
In Figure 7, a 1N4007 is used as the tuning diode. For all the diodes tested, the minimum and maximum frequency and corresponding DC tuning voltages were recorded. In my experiments, the 1N4007 exhibited a capacitance range of 23 to 9 pF with the corresponding minimum and maximum reverse voltages of 0.5 and 12.0 VDC. The most linear portion from my measurements was in the 5-10 volt region of the capacitance versus voltage response curve graph. Thus in Figure 6, a 47K resistor was added between the 50K pot and ground to set the minimum voltage near the desired minimum 5 volt value. In fact, the 47K ohm resistor provided a minimum voltage of 4.60 volts. In my test oscillator circuit, this gave a minimum frequency of 6.826 MHz. Turning the pot all the way to the right provided a maximum voltage and frequency of 9.33 volts and 6.917 MHz respectively. This gives a total frequency swing of 90 KHz.
The frequency swing of a PN-junction tuned VFO is dependant in many factors other than the minimum and maximum tuning voltages. Some of these include the device used, its connection method ( series or parallel ), the L to C ratio of the tank circuit, the value of the series connecting capacitor and oscillator AC amplitude. The frequency range may be influenced by where and how the PN junction is connected to the tank circuit. To elaborate on the last point, by tapping down the inductor of the tank circuit, you can decrease the tuning range of the tuning diode. In one VFO, I connected the tuning diode via a series NP0 capacitor to the normal 25% from ground Hartley tap point that connects the coil to the source of the nJFET oscillator transistor. This reduced the tuning range and the tuning diode was used as an RIT. As you tap towards the inductor's grounded end, the frequency range of your tuning PN-junction should decrease.
In Figure 8, two 1N4007s are connected back-to-back for the reasons stated previously in the text. The 47K resistor from Figure 7 had to be raised to 68K to set the minimum voltage near 5 volts. It gave a minimum voltage and corresponding frequency of 5.17 volts and 6.995 MHz respectively. The maximum tuning voltage and frequency was 8.98 volts and 7.007 MHz. The tuning range is now 12 KHz as the two tuning diode capacitances are now in series.
Note the 100K series connecting resistor in the Figure 7 and 8 schematics . Changing this resistor from 100K to the 470K value would actually change the tuning voltages and frequencies very little. In circuits which use tuning diodes, many designers use higher value resistors for R1 to minimize the effects of stray capacitance in the voltage source which may shunt the diode. The high value resistor also serves to block RF from getting into the DC supply voltage. In circuits which use VVC diodes, the resistor value will often be several mega ohms. There actually is some science involved which factors in the time constant of the tuning diode capacitance and the resistor value and is beyond the scope of this basic discussion.
In fact, whole books have been written on tuning with diodes and the data presented on setting the minimal DC voltage so far is also greatly oversimplified this function. Like transistors, tuning diodes or PN-junctions may be correctly biased with DC voltage so that the effects of the AC circuit voltage are accounted for. Thus when an AC voltage is superimposed on the DC voltage the bias will swing above and below its set operating value. The tuning diode may be biased along its operating curve to provide square-law, linear or other types of response depending on the application. All of this is the stuff of electronic engineering and is left for the textbooks.
In Figure 9, a 1N5408 is used as the tuning diode. This device has a lot of capacitance and is a 1000 volt, 3 amp diode I use to repair bridge rectifiers in tube amplifiers. Note the minimum frequency and voltage is 6.315 MHz and 1.95 volts. The maximum frequency and voltage is 6.577 MHz and 9.31 volts. I did not plot a response curve for this device, however it has more maximum capacitance than any other diode in my junk box.
In Figure 10 is the common 1N914 diode and its voltage and frequency measurements. Note that the frequency swing is just ~6 KHz. Actually this circuit is not set up to use the 1N914 very well as the 1N914 diodes I tested exhibited around 5 pF of capacitance with 0.5 volts DC control voltage and drops below 1 pF with control voltages greater than ~ 4.5 volts. I did not bother changing the circuit resistor or zener diode regulator values to target these control voltages as there is little point. The 1N914 is a very non-linear tuning diode across is response curve and I found the 1N4148 to be similar. These diodes are not a great choice for tuning VFOs from my experiments.
In Figure 11, a 33 volt zener diode is used as the tuning element in the Figure 6 test circuit. A zener diode should be operated below it's breakdown voltage to function as a tuning diode. The measured frequencies and voltages are shown in Figure 11. A voltage versus frequency plot was not made. One good example is W7EL's Optimized 40 M Transceiver which uses a zener diode configured very differently to act as the the RIT tuning element. ( See Bibliography Page )
Voltage Variable Capacitance (VVC) Diodes
As we have learned, every semiconductor diode has some internal capacitance. This capacitance is deliberately exploited in a class of special-purpose junction diodes called VVC diodes. VVC diodes are also called varactors, tuning diodes, silicon capacitors, electronic capacitors and voltage variable capacitors. Various companies have marketed them with the trade names of Capsil, Epicap, Paramp Diode, Semicap, Varicap and Voltacap. In addition, I have seen at least 5 different schematic symbols for varactor diodes.
The capacitance of a VVC diode varies inversely as the reverse voltage and directly as the forward voltage is increased. The reverse voltage capacitance is what we experimenters are interested in as when forward voltage is applied, leakage current is high and Q is low in the VVC. Conversely, when reverse voltage is applied the Q tends to be high and the leakage current very low. The VVC capacitance versus voltage response curve varies in a non-linear fashion over most of its range like the other devices mentioned so far.
Consider Figure 12. I built the circuit for the purpose of determining the voltage versus frequency response curve for a MV2107 in my standard oscillator circuit. Cx represents the series NP0 ceramic capacitor which connects the oscillator to the MV2107 and control circuitry. The voltage was varied from 2 to 12 volts and the resultant frequency was charted with Cx at 51 pF and then 100 pF. The voltage was adjusted by turning the 50K pot until the desired voltage was read with my voltmeter. To get the measurements below ~3 volts, I connected a 100K resistor in parallel with the MV2107 to ground. The numbers were recorded and 2 graphs were plotted on as shown. Although, it is open to interpretation, my results showed that the MV2107 is terribly non-linear when the DC control voltage is less than 4 volts. I found for both cases ( Cx = 51pF and Cx = 100 pF ), the most linear frequency change versus voltage change was probably in the 4 to 9 volt region for my circuit. Thus when I design a real VFO using the MV2107, resistors and a zener voltage regulator will be used to set the tuning voltage variation somewhere between 4 and 9 volts, in order to get the most linear tuning possible.
Of great interest to me was that changing Cx did not seem to greatly change the voltage versus frequency response graph as shown above in Figure 13. However, I found that changing Cx does change the minimum voltage measured in the Figure 12 circuit. With Cx at 100 pf and the 50K potentiometer turned to give the lowest voltage and frequency, the measured voltage was 2.88 volts. When I changed Cx to the 51 pF capacitor, the measured minimum voltage was 2.29 volts Thus changing the value of Cx has an effect on the control voltage. Presumably, the voltage across the VVC diode depends on oscillator AC amplitude and leakages which may be affected by the Cx value. In fact, some may argue that voltage control circuits found in Figure 12 and the earlier schematics were sloppy designs as there is no DC return path to ground in the voltage control circuits which help to better define the control voltage. This has been fixed in Figure 14 by adding the 470K resistor in parallel with the MV2107. Additionally, the 50K potentiometer was grounded via a 100K resistor to place the minimum control voltage above 4.0 volts as determined by the voltage versus frequency change graph. This VFO now tunes a range of 62 KHz by going from 4.06 to 6.66 volts.
The control voltage for a VVC diode may be DC, AC or both. For the simple purposes of this web page, we will only consider the DC control voltage as this is not an electronic engineering web site. When a constant DC voltage is applied, the capacitance will remain constant for a given voltage. The tuning voltage is usually applied by a variable resistor connected to the VVC diode through an isolation resistor which eliminates the effects of hand capacitance and of stray capacitance in the voltage source which may shunt the VVC diode. DC current flowing across the isolation resistor is almost nil at reverse voltages.
If the varactor is connected to a circuit that contains a DC voltage, a blocking capacitor must be used or the DC voltage will serve to tune the circuit or may burn up the varactor. This blocking capacitor may also affect the maximum capacitance attainable by the varactor in the circuit. If you want the maximum varactor capacitance possible, the blocking capacitor must be significantly higher in capacitance value than the maximum capacitance of the VVC diode. Alternately a smaller capacitance may serve to limit the maximum capacitance of the VVC diode in the circuit that you are tuning.
Figures 12 and 14 shows the a MV2107 used as the tuning element in the Figure 6 circuit. This device is a general purpose abrupt tuning diode made by Motorola. It has a Q of 350 at 50 MHz and is sold in a T0-92 package. This component is supposed to give 22 pF at 4.0 volts dc. The pin out is seen in the red inset of Figure 12. This is shown from the bottom view with the leads point to the ceiling. You can look in parts catalogs or even the ARRL Handbook for the Q and other characteristic listings for a given VVC diode. The higher the Q, the better the circuit selectivity. The Q for a given tuning diode varies with the reverse voltage and frequency of the oscillator and this is usually stipulated when the VVC diode is referenced.
Popular VVC diodes used in QRP work may include such diodes as the MV2107, MV2109, BB104, MVAM109 or the MV104 which features two back to back varactors in a single package. All the principles discussed in the earlier text, such as placing two diodes back to back applies to the varactor diodes.
A web site with some VVC diode characteristics listed..
My experiments also confirmed that the higher Q VVC diodes give the most stable VFO of all the devices I used on this web page. I plan future experiments with these devices and hope that you will study the serious designs of other authors. One particularly fascinating application of VVC diodes being used in a VFO can be found at the Elecraft web site under the schematics for the K2 transceiver..
It is possible to frequency stabilize these circuits by using techniques to counter the drift such as polystyrene, N750 or silver mica capacitors and other mechanisms.
Conclusion
The information presented on this page will hopefully get some builders experimenting with whatever diodes they have on hand for their popcorn receivers. Some builders have constructed very stable VFOs using VVC diodes and these have been published on the web and in the HAM literature. Do not be daunted by experimenting. You can also use other linear taper potentiometer values than the 50K version I used. Some operators use 3 or 10 turn potentiometers to provide finer tuning of their VVC diode tuned VFOs. Also to keep costs down, you can use a 50K pot for the main tuning in series with a smaller pot such as a 1K for fine tuning
This web page will be updated in the future as more experiments are performed and comments are received.
If you see any mistakes, have comments, criticisms or stuff to add please email me. Have Fun!
It is probably better to measure the DC voltage on the pot side of the isolation resistor. This will reduce measurement of any rectified AC voltage which will add to the applied DC voltage. I now do this any time I measure applied varactor DC voltage.
Note Figure 5 was omitted.
This page last updated---July 7, 2003.
Supplemental page last updated---Jan 6, 2003.
