Prevent Transistor Failures With Science
Ever wonder what
causes a solid-state RF amplifier to fail? One answer is heat; the
other major contributor is transient voltages. Most of us know to
keep equipment cool, but few understand that voltage spike-induced
transistor failures can be prevented with pure science.
Here is a shorted quarter-wave stub with N T adapter ready to
create voltage spikes in transmitters. Because of that, I highly
recommend a shorted quarter-wave stub after any solid-state amplifier
feeding a tube input. They normally go at a 50-ohm point in the
system and provide a DC short to voltage spikes, while passing RF
with little or no loss.
Electronics has long suggested these for its FM exciters. I build
them for clients and you can, too. (Just contact me via my website if
you have questions.)
RF theory and
practice tell us that a quarter-wavelength of transmission line will
ignore the frequency of interest if it is on a T adapter in the RF
path and is shorted at the far end.
That’s right —
you short the center conductor to the outer conductor at the end of
the cable. The short will be what DC or transient voltage spikes will
see. The trap will short everything but the frequency of interest.
In the case of an 88
to 108 MHz FM system, a 50 ohm RG-58, RG-8, RG-213 or similar coaxial
cable is just fine for the use. The important part is that the cable
needs to be sized in length for the frequency you want to pass.
As you know, each
coaxial cable type has a VF (velocity factor). So a quarter
wavelength of line would not be the same physical length as a quarter
wavelength in free space. Let’s say you are operating at 98.1 MHz.
A wavelength at that frequency is 120.4 inches (about 10 feet). A
quarter of that is 30.1 inches.
RF travels much
slower when going through a transmission line. A typical
example is Belden 8259 RG-58A/U. Its velocity factor is 66 percent.
To achieve an electrical quarter wavelength at 98.1 MHz, that line
needs to be 66 percent shorter, which turns out to be 19.87 inches.
A spectrum analyzer displays return loss at and near the frequency of
Making one of these
will require a connector on one of the cables. I wrote an article
about that subject in
Radio World last
year (“Installing the Connectors
the Right Way,” Oct. 10).
Hopefully that will
help. You can do fairly well by calculating the line length for a
particular frequency, but there is the length of the T adapter to
contend with. I use a return loss bridge to trim the cable length
experimentally so that it is exactly on frequency.
Put a 50 ohm dummy
load on a return loss bridge connected to a spectrum analyzer with
tracking generator. Then sweep the frequency of interest. Your test
arrangement should show 30 dB or more of return loss when the dummy
load is attached.
Install a T adapter
in series with the dummy load and the result should be the same.
Connect a piece of coaxial cable to the open port on the T, and your
return loss will become terrible. Use wire cutters to cut into (but
not through) the line, at a length longer than you calculated or
think it will take to make the trap.
Cut through just
enough to short the center and outer conductors together. The
spectrum analyzer should show a return loss dip at a frequency lower
than the desired one. Then cut again making the line shorter by maybe
1/8 inch. The frequency should go up.
When you get close
to the desired frequency, cut through the cable entirely. Then
carefully bare the center conductor near the cable end and twist the
outer braid onto it.
Here’s the test
setup with a return loss bridge and dummy load connected for testing
a shorted quarter-wave stub.
trimming just a bit at a time until you are on frequency. Then whip
out your soldering iron and solder the end of the center conductor to
the outer conductor at that point. Verify that the best return loss
happens at the desired frequency. If the frequency is too high, throw
the cable away and start over. If it is too low, just keep trimming
the cable shorter.
When you are done,
you can confidently install the shorted quarter wave stub trap with T
adapter in a transmitter or on the back of an exciter. There should
be no change in VSWR.
When I build these
traps, I usually use RG-58 or RG-8 type cable. You will more or less
be pushed into one of those two cable sizes because the equipment
will have a BNC connector for up to 250 watts or an N connector for
up to 1000 watts.
I even built one for
a 10 kW FM transmitter, using 1-5/8 inch rigid transmission line. In
that case, the shorted quarter-wave stub trap was used for
attenuating the first harmonic (twice the carrier frequency), which
is sometimes misnamed as the second harmonic.
Another was built
for 950 MHz, but it was less than two inches long. The goal was to
protect an STL receiver from lightning surges.
In conclusion, I’d
like to say that the more we know about basic theory, the more we can
use it to make stations better.
WØMH, is a professional broadcast engineer, certified
by the Society of Broadcast Engineers. He has more than 30 years’
experience and has written numerous articles for industry
publications over the years. His website is www.mwpersons.com.