Stacked Toroid VFO Experiments 2008

Introduction



I have noted a newer trend in VFO design is to stack 2 powdered iron toroid inductors. For a given inductance, this allows the builder to use less windings and larger gauge wire than would be possible for a single toroidal inductor. In a VFO, the goal of these builders possibly might be to reduce heating effects, increase unloaded Q, or perhaps to reduce core magnet flux density. For me the goal was far simpler, I just wanted to make compact, larger value inductors for VFOs at 3 MHz and less.
This page contains some basic experiments with stacked toroidal inductors and includes a prototype VFO I built for 3 MHz. In addition, during these experiments, low-budget lighting techniques to enhance my close-up photography were developed and a stab at basic html and JavaScript programming was undertaken.



In the photograph above are 3 rapidly wound FT50-43 ferrite toroids. This was my genesis experiment. I measured the inductance of the 2 six turn inductors and then the stacked one. Their values were 19 and 20 uH for the singles and 43 uH for the stacked toroidal inductor. Permeability variation, wire spacing, meter tolerance and other factors need to be considered, but the basic idea is that stacking increases the inductance per turn (~double for 2 toroids, ~triple for 3 toroids etc.) and facilitates the use of thicker gauge wire if so desired.

In the photograph to the left is the T68-6 hamburger. The two T68-6 cores were epoxy glued together and compressed lightly in a vice for several hours. In VFOs, low noise and minimal frequency drift are essential. With respect to temperature-related drift, the builder must consider the thermal stability of the circuit components, RF shielding and the mechanical rigidity of the PC board, anchors and chassis. To address inductor thermal stability, #6 material is chosen by most builders as it has the lowest temperature coefficient of the commonly available powdered iron toroids.

To my surprise, I learned from Wes, W7ZOI that I had an incorrect understanding of the wire gauge versus temperature stability relationship. I used to think that heavier gauge wire had greater frequency stability than smaller gauge wire wound on a given toroid. As it turns out, smaller wire is often better for thermal stability. The smaller wire will get right on the core during winding. Stiffer, larger gauge wire is more difficult to wind so that the wire is right against the core and so air gaps are created that will expand and contract during temperature changes. Smaller gauge wire will have a reduced Q, but it won't be as significantly lower as you might guess. The thermal stability characteristics of large wire is surprising, however, it can be mitigated somewhat by annealing the wire with temperature cycling or by dunking it in boiling water. Roy, W7EL first reported annealing coils in 1980 and this has been confirmed during experiments by builders using temperature controlled ovens. One of the initial tests I performed for this web page was to see if boiling the stacked coil affected the epoxy glue. The glue was not effected by annealing the stacked coil with 5 or even 10 minutes of boiling in water.


A Good VFO Topology


Shown above in Figure 1 is the final VFO, buffer and frequency controlling scheme. The tuning range of this VFO was from 2.998 to 3.027 MHz. The VFO is a tapped Hartley which lacks the usual gate-clamping diode. The main oscillator and buffer are based upon articles written by Jacob Makhinson, N6NWP which were published in QST for Dec 1996 and Communications Quarterly for Spring 1999. This is a stellar VFO in turns of spectral purity because it was designed by N6NWP to be so. Spectral purity refers mostly to reduced phase noise; short-term, rapid and random fluctuations in the phase of the output sine wave which in turn, contributes to receiver noise. Although I bread boarded a couple of common-base BJT oscillators and tested them, I went back to this design and recommend you at least consider it for your next VFO project. You do not need to use a stacked toroids.

VR1 is a L78L05 in a TO-92 package. Using a simple zener diode as the voltage regulator might also be a good choice. I have read that zener diodes below 5.6 volts have a negative temperature coefficient and zener voltages above 5.6 have a positive temperature coefficient. Accordingly, the optimal zener voltage for a minimum temperature coefficient is probably between 4.7 and 5.6 volts. Zener diodes tend to be noisy devices, but a bypass capacitor may mitigate this problem. The L78L05 is a lower noise device compared to a straight zener diode voltage regulator.  I have observed better temperature stability using the L78L05 device rather than 5.1 - 5.6 volt zener diodes in my VFO experiments, but by no means can I scientifically quantify this. Very low temperature coefficient voltage regulation devices are available, but the cost versus benefit is highly questionable in a simple VFO project.

Most builders would tune this VFO with a good quality air-variable capacitor either directly or capacitively connected to the hot end (top) of L1 and be done with it. Depending on the capacitor, this would also provide a good tuning range if desired. Finding cheap and/or good air-variable capacitors is becoming increasingly difficult. In keeping with the "popcorn" nature of this web site I wanted to investigate if just varactors could be used to tune this VFO. I tried connecting varactors to the hot end and also at different tap points along L1. The result was increased noise and decreased frequency stability. The best varactor connection point for noise and frequency stability was to connect them to the lower tap per the original N6NWP VFO schematic. The original schematics called for a inductor with a inductive reactance of about 150 ohms or so at the desired frequency The problem with this is that the tuning range of the varactors was limited to as low as 4 KHz. If I connected the varactors to the hot end of L1 via a small value coupling capacitor such as 68 pF, the tuning range was also low. While this low frequency swing is perfect for use in a VFO frequency drift correction circuit, it is not practical for tuning a VFO for use in a receiver.

This is where the stacked toroidal coil comes in. With a stacked core inductor, one can easily build VFOs for 3 MHz or lower with an inductor XL of 300-400 ohms or more. With a higher inductance (and thus lower C), the potential frequency swing you can obtain is better suited for tapped L1 connected varactors. Note that 2 back-to-back varactors were used to optimize Q and phase noise; you the builder could elect to just use one varactor to increase tuning range with a compromise in the for mentioned parameters.

I chose an inductor consisting of 36 turns of #22 magnet wire on 2 stacked T68-6 trroids. This turns number was chosen arbitrarily - I knew I could get 36 turns of #22 wire on the stacked toroids. The inductor was boiled for 5 minutes after the taps had been created and the enamel was stripped off the wire. The formula used for the taps is as follows: Divide the total turns by 7.25 to get the first tap and by 1.45 to get the second tap. Note that zero temperature coefficient C0G ceramic caps were used as indicated in the schematic to reduce frequency variations from changes in ambient temperature. I now only use newer multilayer, monolithic C0G ceramic caps purchased from Digi-Key. I have found NPO and C0G to be similar, but a salient discovery was that some older and surplus NPO/C0G capacitors currently being sold are terrible for drift in VFOs. Once the basic circuit was built, I temperature compensated the tendency of this VFO to slowly drift up in frequency by replacing a 100 pF C0G capacitor with a 100 pF polystyrene cap.

Temperature compensation is almost always required. The best way to test for drift is a frequency counter, but if you do not have one, you can use a stable receiver set in the SSB/CW mode. I use both. Consider keeping a small assortment of newer low to midrange value RF temperature compensation caps in stock. On capacitors other than NPO (which uses the letter O instead of the normally stated ppm value), the temperature coefficient is listed as P for positive and N for negative, followed by a 3-digit value in ppm/°C. For example, N220 is - 200 ppm/°C. and P100 is +100 ppm/°C.
If your VFO is drifting upward (more commonly seen on my bench) you need to insert one or more N or negative coefficient capacitor(s). If your VFO is drifting downward, then use positive coefficient value(s). On hand, I have polystyrene, Philips and Murata 5% N750 ceramic disk capacitors and some high Q N750 trimmer caps for negative coefficient parts. For positive coefficient compensation, I would personally try to using a silver mica cap or 2 as I don't really have any true positive coefficient caps in stock. Most builders minimally use 4 or more C0G/NPO capacitors to reduce heating effects and to average out temperature coefficient variances plus compensation cap(s) for resonating their VFO. I did not follow this rule-of-thumb in these experiments.
The VFO is lightly coupled to a JFET source follower which is biased with the current source comprised of Q3 and Q4. RFC1 was wound on an FT37-61 ferrite, although a small RF choke could have been used. Anecdotally, I found this home made RFC may provide temperature compensation, although I cannot verify this. I do this for all of my VFOs. Small RF chokes are getting expensive and I tend to wind my own as I have 100s of ferrite cores in my parts stock.

To the left is the next inductor I wound. This coil has 50 turns of #26 wire and is approximately 19 uH. The taps were at 7 and 34 of the total turns. It was annealed and soldered into the same circuit as the previous inductor. I went with a larger inductance, as I was disappointed with the small 37 KHz tuning range of the first VFO circuit. I was hoping that by using a a larger L and thus requiring less total C to resonate at 3 MHz, my tuned frequency range would increase for the same varactor pair. Note that, I have used two different varactors. The combination of these 2 varactors gave the best frequency swing and lowest frequency drift. Experiment with all aspects of your VFO! My collection of varactors is considerably limited. Some varactors, such as the MV2109 pair I tried, caused the oscillator to stop and/or behave in a weird fashion during tuning. A BB104 worked well, but had even less of a tuning range than the pair of varactors shown in the schematic.


To the right is the Figure 2 VFO schematic containing the above 50 turn inductor and the necessary capacitors to resonate it around 3 MHz. It tuned from 2.996 to 3.047 MHz; still a little disappointing. I was hoping for at least 100 KHz of frequency swing. Such is the life of the amateur experimenter! I had to use 2 polystyrene capacitors for temperature compensation this time and the circuit was not quite as thermally stable as the original.
To build this VFO for other frequencies, choose an XL (such as 150 ohms) for the inductor at your frequency of choice. A simple program to do this basic, but boring arithmetic is located on the bottom of this web page. For the 5 pF coupling capacitor, I put two 10 pF caps in series. For the L1 and JFET gate coupling cap, you can use 10 pF for frequencies at or greater than 2 MHz. For frequencies of 10 MHz and greater, consider using a 2 pF coupling cap to keep loading on the inductor light. Capacitors to resonate your inductor are typically just found experimentally using a frequency counter. If you are using tuning diodes, consider not connecting up your varactor pair until you have a functioning, relatively stable VFO running. Tuning resolution is not great with a 100K pot. A large size knob can help, but the more you spend, the better tuning becomes! A big 2 watt Allan Bradley pot with a big knob worked well on a receiver I built a few years ago, but a 10 turn-pot proved better. Ten-turn pots are expensive and it may be best to bite the bullet and just purchase a air-variable capacitor with built-in reduction. This also eliminates the varactor pair and the 1000 uH RF choke. The choice of how to tune your VFO is yours to make.

Above are photos of my frequency counter measurements of the VFO(b) at start up, 1, 5, 10, 40, 60 and 80 minutes. When you consider that my circuit was lying uncovered on a bench, the inductor was not anchored by anything but its wire, the furnace was cutting in and out and that it had survived 4 days of bread boarding and had many long-leaded components (for re-use), this is actually pretty decent. I listened to the VFO on a nearby receiver while using its 500 Hertz crystal CW filter and I did not have to make any receiver tuning adjustments after the 10 minute warm up period. To me, the ultimate way to test VFO frequency stability is by listening to it in a receiver. Your ears more sensitively perceive frequency changes than your eyes do when staring at a frequency counter sampling every 1-2 seconds or so.

I have taken some flack on the World Wide Web and by email for using my ears as a VFO stability tester. Perhaps this is well deserved criticism, as it does not quantify drift. However, the last time I checked, receivers are meant for listening to signals and almost any drifting oscillator beat note will not stay centered in a narrow IF pass band of a stable receiver. If a VFO stays put in a narrow pass band, I am pretty sure it is stable enough for use in a home-built transmitter or receiver. From my experience, albeit limited, any drift you can measure you can also hear. I sure wish some of my critics would publish their work so I wouldn't have to perform so many experiments to try to improve my hobby projects! The target audience of this web site is people who want to have some fun and perhaps do not have hundreds of dollars worth of test equipment. It is okay to use a receiver as a piece of test equipment if you want to or don't have anything better to use.
Apart from digitization, miniaturization and the demise of HAM radio in general , I posit some of the other reasons that analog hobby electronics is dying is lack of mentorship, imagination and fear of failure. Every design or method generally has good points and bad points. This web site is truly for people who like to experiment with and enjoy building simple electronics circuits. This is the "popcorn" niche I aspire to. I have found that it is very easy to criticize, but far more difficult to contribute. Hopefully I am in the latter group!
Thanks for all the feedback; good or otherwise.


RF Amp and Low Pass Filter


To the left in Figure 3 is the RF amp I built for the 3 MHz VFO. The VFO and buffer output is too low to properly drive a 50 ohm load, so an amplifier is required. The output power is 7 dBm, which was chosen to drive a diode ring mixer. A 50 ohm pi LP filter follows the RF amp. The 3.88 MHz cutoff frequency was chosen so that standard value capacitors could be used. A VFO output LPF is strongly recommended.
In more comprehensive receiver designs, consider using a better LPF such as a half-wave type.

In the photo above is the VFO output into my oscilloscope. Note the camera reflection.


Miscellaneous Images and Chatter

Above. The single-sided copper circuit board ready for bread boarding. Copper was removed to create a B+ strip, 5 volt section, a pad for the inductor resonating caps and a bare area for the inductor to lie on. Currently, for build and keep VFOs, I anchor my coil with 3 small zap straps around the inductor and through holes drilled in the circuit board. The circuit board is secured into its chassis with 4 thick bolts.

Shown above is the VFO bread board looking quite messy from all the various components present and also those which had been soldered in and later removed. This board was used over 4 days and 3 different VFO topologies were built and tested. By far, the tuned Hartley presented on this web page was the winner. A red LED can be seen and flashed when the circuit was powered up to warn me. It is easy to forget that you have power on when experimenting otherwise.

Above are 2 high quality air-variable capacitors. Either of these would be great for tuning a VFO. Most builders use a main tuning cap plus a second, smaller air variable trimmer capacitor to set the band edge. In the experimental VFOs, I did not use a trimmer cap and tuned them close to 3.0 MHz with appropriate type fixed value capacitors and luck.



A 3.5 MHz VFO for Diode Ring Mixers

Shown above in Figure 4 is a simple VFO that spans from 3.5 to 3.86 Megahertz. Note than the main VFO output is from the Q1 gate. Additionally, the Q2 source is not AC bypassed. This provides a very clean output waveform that does not require low pass filtering. The paralleled 390 + 39 ohm source resistors on Q2 set the AC output power  to exactly 7 dBm, The 39 ohm resistor (labeled RS) is probably not necessary as any output value from 5 - 10 dBm will be fine for a diode ring mixer. In another version, I just used a 470 ohm resistor.

Shown above is the Figure 4 bread board. Tuning is achieved with a high-quality air variable capacitor with a built in 6:1 brass geared reduction drive system. This particular main tuning capacitor allows precise tuning even though the band spread is high. I am going to keep this VFO for my lab. One hour drift is under 30 Hertz without temperature compensation or a cover. This was more luck than good design. The basic VFO topology (minus the buffer) as shown in Figures 1, 2, 4 and 5 may also be found as Figure 4.15 in EMRFD. The Figure 4 buffer is pure popcorn, but works very well. This is a clean, stable and simple VFO. You need not use a double stacked, number 6 material toroid for the inductor unless you want a lot of inductance.

Update Nov 16, 2009: I have received an email regarding Figure 4 above. The builder did not observe a nice sine wave output in his oscilloscope. The signal leaving Q1 was pure, however, it was being distorted in the Q2 buffer. There was clearly a build or measurement problem. This buffer was designed specifically to give an undistorted output. To ensure reproducibility, today, I built yet another Figure 4 VFO at 7.00 MHz and used a very similar buffer. The oscilloscope output of this breadboard is shown above. In the buffer, I used a 1 megohm Q2 gate resistor and a 33 ohm (anti-oscillation) drain resistor instead of the 51 ohm used originally in Figure 4. The sine wave has low phase noise and no distortion as shown above. Ensure you take the LO signal off the Q1 gate and do not  bypass the Q2 FET source resistor(s). Q2 should be run off 12 volts or so. During testing, connect a 51 ohm resistor from the T1 three turn output link to ground. Measure from the 51 ohm resistor to ground. He might have a bad FET as well.


Photographic Technique

The above image illustrates how I currently light my photographic subjects. In this case, I am photographing my radio station keyer with a background consisting of recycled, shiny, white paper. Two full spectrum Ott-Lites were used and I find they provide adequate brightness and good depth and color for close up shots. These were purchased last year when a local store had them on sale at 40% off their regular price. I use a modest, lower Megapixel camera set for Macro photography. For every picture that is kept, 10 are discarded. The usual problems are motion artifact and lighting issues.


Applications

Calculate Inductive Reactance (XL) from Frequency and Inductance
Enter Inductance (uH)
Enter Frequency (MHz)
= XL (Ohms)
Calculate Inductance from XL and Frequency
Enter XL (ohms)
Enter Frequency (MHz)
= L (uH)
Tapped Hartley VFO Inductor Tap Points Calculator
Enter # of Turns
= 1st tap
= 2nd tap


These are just written in html - I also wrote JavaScript versions, but these seem to work okay.


Epilogue - December 19, 2009

Shown above in figure 5 is the new QRP/SWL HomeBuilder preferred VFO topology. Please see Figure 4.15 from EMRFD. This VFO topology somewhat outdates my previously presented designs. The older designs work fine, but you should probably replace the 6.2 volt zener diode with a L78XX voltage regulator. I asked Wes, W7ZOI if the the gate clamping diode used on these designs significantly adds to VFO phase noise. He informed me, that it does increase phase noise, but not significantly for most applications. 

I am hopeful that this web page was informative and interesting. Good luck with your own VFO experiments!