|ZL2PD Hunts for Varicap Diodes|
|Republished from Jan/Feb 2007 issue of “Break-In”, New Zealand’s amateur radio magazine, with the permission of the editor|
|Like most people who have electronics as a hobby, I’ve found some components increasingly hard to find. I keep a list of the really hard to find parts on a notice board, adding to it from time to time, and removing the occasional item when I finally track them down, or when I find a suitable substitute. Most recently in the shack of ZL2PD, I really needed to find some high capacitance varicap diodes.
My problem was simple. I’m in the process of building an 80m (3.5 - 3.9 MHz) SSB transceiver using a Collins 455 kHz mechanical filter in the IF chain. Such designs need some good front-end receiver selectivity in the form of at least a couple of selective tuned circuits. In addition, I’m planning to use a simple transistor VFO. This design approach requires three tracking tuned circuits - two in the preselector filter ahead of the mixer, and one in the oscillator. The simple approach calls for a three gang variable capacitor. Thirty years ago, a trip to a local parts supplier would yield such a device reasonably readily, probably along with a nice 10:1 reduction vernier drive.
More recently, say ten years ago, I would buy a set of varicap diodes. A common choice for HF band designs was the excellent Philips BB212. Three such high capacitance varicaps would be ideal for my transceiver. Guess what component has been on my ‘rare and endangered’ components list for more than three years?
|A selection of many different types of diodes that were tested in the search for a part that could work as a varicap diode|
|Varicap (i.e. variable capacitance) diodes operate using a reverse bias voltage applied to the P-N junction of the diode. As this reverse bias voltage increases, the depletion layer at the P-N diode junction widens. With normal operation of a silicon diode, it will take about 0.6V to force electrons across this depletion layer, the reason for the this voltage drop across a conducting diode. In the case of a reverse biased varicap, since minimal current can flow in a good reverse biased diode, the result is an insulating layer between two conducting silicon layers. This layer widens with increasing reverse bias voltage. The result is a device which can be used as a voltage variable capacitor.
Varicap diodes are also produced with a ‘hyperabrupt’ junction in which the P-N junction is doped to enhance this effect. Capacitance can readily range across a 2.5:1 ratio across the specified voltage range in a standard varicap, and up to 10:1 (and often beyond) for hyperabrupt devices. Typical VHF and UHF varicaps range from 1 to 50 pF while HF varicaps range from 50 pF to 500 pF.
Two common circuit arrangements are used. Figure 1 (a) shows the arrangement for a single varicap diode while Figure 1 (b) shows the more common dual varicap diode configuration. This latter arrangement has some practical benefits, ensuring RF voltages across the tuned circuit will not cause the diodes to conduct, but it does reduce the available capacitance by a factor of two. To reduce inductance effects of device leads, and to reduce oscillator noise, some designs also feature multiple paired varicaps in parallel.
Since bias currents are tiny, the resistor can be as high as 100k without any problem, although RF chokes are also commonly used.
|Figure 1: Single and double varicap tuned circuits|
|While low capacitance varicaps for VHF and UHF circuits are reasonably easy to find, high capacitance varicaps for HF and AM band designs are as rare as unicorn’s tusks. Those wonderful Philips BB212 dual varicap diodes appear to have become nigh on unobtainable. A few can occasionally be found, but at a price to make your head spin. Well, mine, at least.
Other alternatives exist – Toko make some nice varicaps in the KV1500 series, On-Semi (previously known as Motorola) still make some of the MV series, while Zetex also make some useful devices. The problem is, you’re not likely to find anything like these down at your local parts supplier, assuming you’re lucky enough to still have such a store! If you could buy them, most now seem to be made only in eye-strainingly small surface mount device (SMD) packages. The biggest problem? Unless you’re up to buying a few thousand at a time, you may not be able to buy these anyway. I found I could order a few over the internet, but the process is slow, and to get just a few parts, the cost can be very high.
I had heard rumours of workable alternatives to these devices. Somewhere I had read that zener diodes could be used in their place, but I couldn’t find any information about typical performance. Others suggested that high voltage rectifier diodes made great varicaps, while one source suggested LEDs.
I’ve even used rectifiers (like 1N4004 power diodes) in the past for the odd circuit, but I suspected these would not provide the wider capacitance range needed for my transceiver. As it turned out, a few quick tests proved this to be the case.
|The results for the first set of diodes I tested (shown in the photo at the top of this page) are plotted in Figure 2. I’ve not seen many graphs of capacitance for typical diodes used in this way, so these may also prove useful to other homebrewers. The diodes shown include many of the diode types which others suggest are useful as varicaps:
- A selection of small and large LEDs
- A standard 1N4001 1A rectifier diode
- A large 4A rectifier diode
- The largest power diode I could find in a plastic package (about 10 mm in diameter!)
- A variety of small 400 mW zener diodes, and
- A Motorola 1N759 12V 1W zener diode.
Incidentally, some care may be needed if you use LEDs as varicaps. Some have reverse breakdown voltages as low as 5V. I’ve tested a number of LEDs at with reverse bias voltages beyond 12V without apparent difficulty, but longer term use might encounter reliability issues.
These diodes were all measured using the test rigs and methodology described in more detail in the appendices at the end of the article. Figure 2 shows the results.
|Figure 2: Diode performance for a variety of common diode types|
|Because my SSB transceiver uses a 9V supply for most receiver and transmitter stages, I was most interested in varicap performance over the 1 – 9V range shown here. There is little point in tuning varicaps below 1V. In many cases, the device’s Q, such as it is, seems to fall below 2V, and any oscillator voltage across a tuned circuit can rapidly drive a single varicap into forward conduction. This completely ruins the reverse bias varicap effect.
Most of the these diodes fall into either a 10 – 20 pF or 15 – 30 pF varicap class across the 1V to 9V reverse bias range. These will be ok for most fine tuning applications or in huff-and-puff oscillators, but they were useless for my transceiver. I needed much greater capacitance ranges.
By the way, the largest rectifier diodes had negligible capacitance variation, and those results are not included in the graph.
Clearly, these diodes were completely inadequate for my higher capacitance requirements. Just as despair was beginning to close in, I tested a Motorola zener diode I found lurking in the corner of my junkbox. It showed some of the right features, but….
It was like seeing a murky photo of the outline of a rare mythological beast – A sniff of a clue to justify a full scale hunt.
|In sharp contrast to the other small zeners tested, that Motorola 1W zener diode I found delivered a useful 40 to 100 pF range. I’ve plotted this in Figure 3.
It was not adequate for my requirements, nor could I find any more of these in my workshop, but it suggested where to focus my varicap hunt. I headed off to the local parts store and bought a set of 1W zener diodes, one of each voltage across the range available. A quiet spring afternoon of measurement and analysis in the ZL2PD shack produced the results shown in Figure 3.
|Figure 3: Capacitance performance of 1W zener diodes and the Motorola (On-Semi) 1N759A 1W zener diode|
|My first reaction was disappointment. I had thought that 1W diodes, perhaps because of their larger silicon wafer structure, would deliver the same response as the Motorola 1W zener. Sadly, while useful, not a single 1W zener reached the halcyon heights of the 1N759A. Nor was there any apparent relationship between their voltage rating and their test results. But, on the positive side, the results were better than the general mix of diodes I’d measured earlier.
Well... If 1W zeners were better varicaps than 400mW zeners, could 5W zeners be even better?
Samples of 15V, 24V and 75V 5W zeners were quickly obtained. By the way, these parts are big. The 15V and 24V zeners measured almost 10mm in length and over 3mm is diameter, while the 75V part, although slightly shorter, has even heavier duty copper conductors, and a body nearly 5mm in diameter. Serious zeners! Testing these also required a change of test oscillator, details of which can be found in the appendix. Figure 4 shows the results.
|Figure 4: Graph of 5W zener diode varicap performance|
|These zeners have capacitance, serious capacitance! Not only did the 75V zener feature a solid 190 – 310 pF range, a ratio of 1.6, but I was stunned to find that the 15V 5W zener had an amazing 380 – 760 pF range, or a useful ratio of 2:1. Results predictably follow voltage ratings, even though the parts were from a mix of manufacturers. While smaller 1W zeners offered capacitance ratios of up to 3:1, equally useful for other homebrew applications, these big zeners certainly seemed to match my varicap needs.
At first glance, some may feel these very large capacitance values are too high to be useful for anything beyond AM and low band HF use. In fact, they open up a wide range of uses when combined with a lower value series capacitor or used in centre-fed pairs across a tank circuit. Both arrangements reduce the impact of RF voltage swings on the diodes.
|RF Performance Tests|
|A vital issue was to determine if these diodes had adequate Q to support real RF applications such as front end tuned circuit and oscillator tank tuning. Given the noise floor in lower HF bands, the target of my transceiver, I could accept modest Q values. While the test oscillators (See Appendix B for the circuit details) gave me some confidence that the diode performance was probably OK, I decided to try building a simple toroid tuned circuit at 3.5 MHz. This would also allow me to adjust the voltage on the diodes in circuit while evaluating the Q using a grid dip meter.
This worked perfectly. 45 turns on a T37-2 toroid gave me the required 8uH inductance to resonate with the 75V/5W zener on 80m, and tuning the diode from 2.5 to 7V covered the 80m band perfectly. The coil/diode seemed to deliver the typical selectivity of a single tuned circuit.
A more precise test was required. For this test, I built a variable bandpass “ultra-spherical” filter. The simple version I tested is shown in Figure 5. Developed by Wes Hayward W6ZOI, and first published in Ham Radio magazine in June 1984, this is a tunable LC bandpass filter which provides up to 20dB of selectivity at the band edges of 80m when tuned mid-band. Although Wes used a standard variable capacitor, I set out to make and test a version using these diodes.
|Figure 5: Variable bandpass filter
(from Wes Hatward W7ZOI)
|The results exceeded my expectations! The filter tuned from 3.5 – 3.9 MHz as bias was adjusted from 1V to 9V, and when tuned mid-band, to 3.7 MHz (about 2V bias), I measured the performance shown in Table 1. This table also compares these results with those published in the original article for the same filter using conventional capacitors and high-Q inductors.
Parameter Varicap Filter (Fig 5) Original W7ZOI Filter
Insertion Loss -1.9 dB - 2 dB
-6dB Bandwidth 85 kHz (–40/+45 kHz) 80 kHz
Attenuation @ 3.5 MHz -15.5 dB -20 dB
Attenuation @ 3.9 MHz -16.5 dB -20 dB
Table 1: Varicap Tuned Ultraspherical LP/BP Filter
The original W7ZOI filter was designed using inductors with Qu=250 while T37-2 toroids I used typically reach Qs around half these values. I expected increased losses and poorer band edge results. While my filter had slightly reduced band edge attenuation, the overall performance was quite acceptable. More importantly, the test confirmed the viability of the zeners as varicaps.
|These tests indicate that 5W zeners can be used as high capacitance varicaps, although the physical size of these devices could be a problem in some cases. With care, this should not present a major limitation.
I should caution against assuming that any 5W zener diode will give the results in line with those shown here. I’ve tested several types, and they do seem to give similarly useful results. However, as those television adverts say, “Individual results may vary” and I would recommend testing any zener diodes before committing to buying larger quantities.
My next step? Having tracked down those elusive varicaps, I’ve taken that part off my ‘big game’ list, and I’ve returned to building my transceiver front end and VFO.
|1. Wes Hayward, W7ZOI, ‘The Peaked Lowpass: A Look at the Ultraspherical Filter”, ‘Ham Radio’ magazine, June 1984, p 96 – 104.
(See Figure 14 and 15)
|Appendix: Measurement Method|
|An HF test oscillator was built to cover 10 – 50 MHz (Figure B.1). A set of fixed capacitors ranging from 4.7pF to 820pF were used to produce a graph of frequency versus tuning capacitance, and Microsoft Excel was used to plot the graph and to derive a simple curve-fit equation. The same method was adopted for a second oscillator using larger values of capacitance covering 250 – 550 kHz. The plot for this second oscillator is shown in Figure A.1.|
|Figure A.1: Oscillator frequency vs tuned circuit capacitance for the oscillator in Figure A.2|
|Figure A.2: 10 - 50 MHz test oscillator|
|Figure A.3: 500 kHz test oscillator|
|Each varicap was measured using the HF oscillator, with the oscillator’s frequency measured as the varicap bias was varied from 1V to 9V. The higher capacitance 5W zener diodes were measured using the second (500 kHz) oscillator. Measured frequencies were converted to equivalent varicap capacitance using the curve-fit equation obtained from the initial test, again using Excel.
Varicaps with sub-50pF capacitance were measured using the HF oscillator and a parallel 220pF fixed capacitor (C1 in Figure A.2). A similar graph of frequency versus tuning capacitance, and a curve-fit equation, was used to more accurately derive these relatively small varicap values.