Choosing L-network Capacitors

Anyone who works on high power amplifiers and antenna matching networks know that suitable capacitors can be expensive. They must be stable, highly efficient and withstand high RF voltage and current. Choosing lesser components will only result in failure and replacement. Air dielectric capacitors, whether variable or home brew with a fixed value, work well but are large, usually far too large for where we want to fit them.

In my junk box I have a modest quantity of capacitors that are rated for high power RF applications. Of course one never has enough or of the desired values if antenna experimentation is to be done. Because I have been using no more than 200 watts since returning to the hobby several years ago I have often taken the easy way by using unsuitable capacitors in matching networks, knowing that they would most likely survive at these power levels. So far I've had no spectacular failures or excessive heating.

Since my plan is to run QRO eventually, probably next year, taking a shortcut on RF capacitors is no longer an option. I have done some reading and research to find alternatives that are not expensive and will do the job. This is important because I do like to experiment with antennas, which requires a stock of capacitors covering a large range of values.

As already stated, variable capacitors are too big. For the low bands fixed padding capacitors are needed in any case so variables are no panacea. Years ago variable capacitors were commonly used because there were few affordable instruments to measure antenna impedance and design a network to convert complex impedances to 50 Ω, and the math is difficult for most hams.

My 160 meter antenna L-network

We have made progress. Now I measure the feed point impedance with an accurate analyzer, plug the values into software tools and -- presto! -- out pops a matching network. I design coils again using software, pick one more capacitors and measure their value and built the network. It is now quite easy to get a near perfect SWR in one pass of this procedure. An example is my article on designing and building the L-network for my current 160 meter antenna.

For that network I needed a capacitor of around 2150 pf. From my junk box I pulled out an ancient 2200 pf disk ceramic capacitor. It worked well despite being of uncertain voltage rating and RF efficiency. But it was not without problems. There was a noticable temperature rise during extended contest operation on 160 meters, which in turn altered the capacitance and SWR.

To compensate I preset the rig's ATU for the "hot" capacitor. When the SWR started its inevitable rise I would switch in the ATU and continue operating without further problem. Happily the capacitor suffered no catastrophic failure as I feared.

This fall I attempted to improve the situation. The original capacitor (see above) was replaced by a chain of higher value disk ceramic capacitors. Power handling of small capacitors can be increased by connected higher value capacitors in series or lower value capacitors in parallel by decreasing voltage across or current through each capacitor, respectively.

It didn't work as expected. Although the photo is fuzzy you may be able to make out the "Z5U" rating symbols. I foolishly paid no attention. When put into service the heating effects were far worse than with the the original capacitor. Within 15 seconds of CW transmission at 200 watts the SWR swung wildly upward. Wait a minute and the low SWR returned. It was almost impossible to keep the SWR low without frequent re-tuning of the ATU.

Not all capacitors are equal

After a month of this I did what I should have done at first, which was to find out what Z5U means. In brief, the dielectric properties are such that the capacitance is acutely sensitive to temperature and efficiency is poor. These capacitors are suited to RF bypass usage where the capacitance value isn't critical. It was time to find something better.

For those who want a ham-oriented introduction to RF power capacitors I highly recommend an online article by I0JX. The technical depth is just enough to explain capacitor design and construction that the average ham should be able to understand. He surveys the variety of capacitors in the market and how to identify which ones are appropriate for high power RF applications, and how to pick the right capacitor at each point in a QRO amplifier. This is directly applicable to antenna matching networks.

It turns out that the 500 pf doorknob capacitor I used in the L-network of my short 80 meter vertical in Ottawa several years ago is not rated for carrying QRO. Rather it is only adequate for RF bypass. Fortunately it is big enough that the 100 watts maximum I ran at that time let it run cool. Not all of those large ceramic doorknob capacitors are the same.

Scrounging

While visiting a friend in early December he regaled me with the woes he's had with finding replacement capacitors in the pi-network of his Eimac 8877 HF amplifier. These are switched in for 160 meters because the air variable capacitors for the plate and load control need padding. Unfortunately the capacitors he had that were properly rated were too big to fit in the small space where the existing ceramic "doorknob" capacitors are situated.

My eyes lit up. I calmly asked to see these overly large capacitors that he can't use. He pulled out a cardboard box full of large and ancient mica capacitors. It was a treasure trove for a scrounger like me. Having done him a few favours he let me go through them and find what I needed. I was fortunate to find one with a value of 0.002 μF, which is 2000 pf. It measures 2300 pf, near perfect for my 160 meter antenna.

A week before Christmas I cut out the problematic chain of disk ceramic capacitor and popped in the mica capacitor. As you can see it's quite large. The screw holes turn out to be the same size as on my stock of doorknob capacitor, making it easy to attach to the existing fitting on one end. On the other end the screw holds the severed lead of the removed capacitor.

160 meter bliss

Back in the shack I turned on the transceiver and tested the modified L-network. The SWR was back to 1 at resonance, and less than 1.5 from 1800 to 1840 kHz. After 10 minutes of hitting it with 200 watts of CW the impedance was rock steady. In subsequent operation it has continued performing beautifully.

I am hopeful the capacitor will work well with 1000 watts. There is some reason for doubt. These ancient "sandwich" style mica capacitors are considered unreliable over long use due to the mechanical construction. New mica capacitors of equivalent rating are extremely expensive. My hope is based on the RF stability of all varieties of mica capacitors despite my failure to locate specs on the internet of this specific capacitor type. It's just too old, I guess. What little I could learn tells me that the RF current rating should be over 10 amps, more than enough to handle 1000 watts in an L-network designed for a moderate impedance transformation.

One further note on L-network design as it relates to capacitors. For every impedance transformation there are typically four topologies: shunt L or C and series L or C, with the shunt placed on either the generator or load port. The one I am using places a shunt C across the load port, with the L in series between the ports. Tools such as TLW, which comes with the ARRL Antenna Book, allow you to select the network topology.

I chose this topology because it produced values of C and L that were convenient to implement. Topology choice also determines whether the network rejects harmonics (low pass filter), an important consideration for SO2R and multi-op contests. At this time I did not bother to do so. I will when I decide on a permanent 160 meter antenna. My L-network designs for the 80 meter vertical yagi use a low pass filter topology.

Going forward

I continue to keep my eyes open at flea markets for high power RF ceramic capacitors. You have to get there early because they go fast! That is, the ones with high voltage and current ratings, and those with high capacitance values. The electrical boxes I use for matching networks have enough room for putting a couple of mica or doorknob capacitors in series or parallel to get the values I need. I can probably get a few more of the large mica capacitors from my friend to add to my stock.

For the two L-networks in my 80 meter vertical yagi I will conduct another experiment. I ordered a large number of small valued high voltage, high stability disk ceramic capacitors that I can add in parallel to get the values I need. These capacitors are very cheap when bought in bulk so the financial risk is negligible.

We'll see how they do in practice, first with 200 watts and later with 1000 watts. If it doesn't work out, well, it may be back to hunting for surplus mica and ceramic doorknob capacitors.

If you've never given much thought to the capacitors needed for building matching networks I hope this article has given you the incentive to learn about capacitors and encourage experimentation. For those who enjoy playing with HF and MF antennas you will see the difference when the proper capacitors are used.