Duplexers
HomeWhat are they and how they do a most incredible job..?
Modifying 6" VHF cavity filters to 2 metre notch filters
Modifying low band cavity filters to 2 metres
| Duplexers operate at very close frequency
pairs, in the same band. Diplexers operate at widely separated frequencies such as VHF - UHF Duplexers really do an amazing job. They allow a transmitter and receiver that are very close in frequency to operate into the same antenna. And on 2 metres, with the receiver and transmitter using frequencies only 600KHz apart, they have to provide radio frequency isolation of 100dB. That is a very large number to say the least. The closer the frequency pairs the greater the isolation required. Doubling the frequency separation reduces the required isolation by about 12dB. |
WARG's eleven sets of 2M duplexers, 66 cavity filters in total.
| A prototype duplexer was made from the
article in QST's Repeater handbook based on the W1GAN design and it
worked. http://www.repeater-builder.com/antenna/pdf/w1gan-duplexer.pdf This original duplexer is now located at Walliston attached to VK6RLM. |
WARG's first 2M duplexer
| WARG club members knew little about cavity
filters and Duplexers from a practical construction point of view. How
important was the type of construction and if assumptions were made how
would the performance be affected..? The construction article in QST
were detailed but did use some materials and type of construction that
was not easy to duplicate. For example the "finger stock" used at the
movable tuning point of the center resonator. Also the insulators used
at the top of the cavity for the entrance of the coupling loop. Could
we do it a simpler way not requiring difficult to find materials and
fabrication...? The answer after constructing a single cavity filter and doing some tests on it showed that simpler materials and construction could be used. These changes were minor with the knowledge gained, but at the time much was a new frontier. Temperature drift
Tests were done on this cavity filter to see if the all important notch frequency drifted with temperature. The cavity filter was set for a notch of 146.1 and placed in the fridge. Once it had reached 5 degrees C the notch frequency was measured and it had drifted only 20 KHz higher. Next the cavity filter was placed in the Sun and was too hot to move without gloves. The notch had only drifted 30 KHz lower. This was a bit of a surprise, that so little temperature drift was found over such an extreme temperature range. It could be that as the center tuning element lengthens pushing the cavity filter lower in frequency. However the outside tube increases in diametre and hence with less capacitance between it and the center tuning element, raises the frequency. The two tend to cancel each other out to some degree. Construction It was decided to construct eleven (11) duplexers. With 6 cavity filters in each duplexer this required 66 full size 2 metre cavity filters to be constructed. A very ambitious project to say the least. The outside tube is 4" copper and the inside tube 1 and 3/8". The inside tube has a smaller tube (1") inside which has a threaded rode attached and this protrudes through the top of the cavity filter and is the frequency adjustment. On top is an aluminium box that contains the BNC in/out connectors and the notch tunning components. The large amount of materials were purchased and work began. Job number one was to cut the copper tube into the correct lengths. This was done on a lath and special wooden holders had to be constructed to hold the copper tubes. Aluminium boxes were drilled and BNC connectors installed. The top plate has the centre 1 and 3/8" soldered to it and the 4" copper tube is soldered to a brass base plate. All the soft soldering was done on Trevor's (VK6MS) gas stove. All went well and we began assembly of the filters. All the inside surfaces of the cavity filters were highly polished and varnished. Silver plating would have been nice but way outside our budget. The outsides were painted with a green blue marine paint. All cavities were then screwed onto a wooden base and carry handles added as a duplexers weighs a lot.  Duplers
ready for alignment
Next came the tunning and this does require a good understanding of cavity filters and the point to appreciate is this duplexer is a notch duplexer. The standard 2 metre cavity filter in bandpass mode does not have enough attenuation 600KHz away, only 10dB. A duplexer for 2 metres at 600KHz separation (RX/TX) is required to have 100dB isolation between receiver port and transmitter port...! And that is a lot. You could place ten cavity filters connected in series to achieve the 100dB but as each cavity filter has 0.5dB loss on the centre pass frequency this would add up to a 5dB loss on the wanted pass frequency and a total of 20 cavity filters required for a duplexer. Remember what we want is a filter that has as little loss on the pass frequency and as much attenuation on the frequency we want to reject 600KHz away.  2m cavity filter band pass A
standard cavity filter has a typical bell curve response but if you
look
at the response over a wide frequency range an interesting result is
observed. The response is not symmetrical. Above the centre resonance
frequency there is a deep notch. On 2 metres this is typically 20 MHz
or so higher and the notch depth can be 100dB...! This notch can be
moved closer to the centre frequency by adding capacitance across the
in/out connectors, typically 5pF. However the closer you make the notch
to the centre frequency the less is the notch depth. At 600KHz away the
notch depth for a 4" cavity is around 35dB. Larger diametre cavity
filters with higher Qs would be greater
Adding the notch components inside the alumnium box produces a very sharp notch, some 35dB 600KHz away from the pass frequency. However it is important to know that once the notch components are added the cavity filter has very little attenuation on frequencies other than the notch frequency. These notch components are either a capacitor for the notch to be on the low side or an inductor for the notch to be on the high side. The capacitor or inductor are soldered between the in and out BNC connectors. Yes this seems strange to connect anything between the in and out connectors.  Cavity filter with inductor added (wire shown in red)  Cavity filter with capacitor (~5pf) added (shown in red) Putting it all together Once the cavity
filters were completed the all important connecting together 6 cavity
filters to make a duplexer. Alignment was the critical requirement and
using the ARRL repeater handbook article on the W1GAN design alignment
was undertaken.
Put simply each cavity filter is aligned on its own, to the pass and reject frequency. The cavity filters can not be aligned when connected together, as there is interaction between cavities, and one cavity filter tries to correct for mis-alignment of another cavity filter, and the end result is poor overall performance, the pass attenuation and notch depth is not as good as it could be. Without a tracking spectrum analyser the process of alignment is tedious and does require patience and an understanding of what you are doing. I put in many hours of trial and error aligning the cavity filters. I could not get the published results. I came close but not the 0.4 dB pass attenuation along with the 35 dB rejection notch. More like 0.6 dB pass attenuation and 33 dB notch attenuation. After much playing around I discovered the coupling loops were over coupled and on moving them slightly out from the centre tuning tube obtained results closer to the published figures. After many hours I was able to tune the cavity filters and connect them together to make a working duplexer. I used various different methods and they are presented below. An interesting point....How do you measure the dB loss on the pass frequency and the attenuation on the notch frequency of each cavity filter if you don't have a tracking spectrum analyser or calibrated signal generator, neither of which I had. Explanation...What is a tracking spectrum analyser.....? A tracking spectrum analyser is a device that uses a signal generator that can be swept over a given frequency range, meaning shift frequency back and forth over the range you want to look at on a spectrum analyser. The spectrum analyser has to be synchronised to the sweep rate of the RF signal generator, so the resulting trace is viewable on the spectrum analyser.  Typical tracking spectrum analyser display of single cavity in notch mode (low pass) The
spectrum analyser display above makes for easy tuning of a cavity
filter in notch mode as the effects of changing various situation
with the cavity filter, such as
In / out loop coupling Varying C and L to produce the position of the notch And the frequency to which cavity filter is set to can be seen instantly on the spectrum analyser. Without the use of a spectrum analyser and associated signal generator, the simpler methods of tuning up a duplexer is a lot more labour intensive. Some basics
A single cavity filter with no added C or L
to produce a notch is know as
A band pass cavity filter A single cavity filter with the C added between the in and out connectors is know as A high pass cavity filter (notch on low side) A single cavity filter with the L added between the in and out connectors is know as A low pass cavity filter (notch on high side) If you look at the spectrum analyser display of cavity filters that have been modified with a capacitor (C) or inductor (L) it is most important to note the basic bandpass characteristics of the cavity filter has been lost. No longer does the cavity filter act as a, pass only a narrow RF frequency and greatly attenuate all others, but a notch out a narrow RF frequency and pass with little attenuation all other frequencies. This is important, as connecting a duplexer to a repeater does not offer attenuation of unwanted frequencies, such as any spurious RF transmissions that a repeater's transmitter may have, or offer any attenuation of strong RF signals to the repeater's receiver, other than the notch frequency. Cavity filter alignment using simple
equipment
Over the years I used several
methods. The first difficulty was how do you know what 0.5 dB is if you
want to know that the single cavity filter is producing the expected
loss on the wanted pass frequency...? And how do you measure the -35 dB
attenuation on the notch frequency...?
Simple test gear
From my first attempt to convert an ex
commercial FM transceiver to 2 metres, there was no way around
constructing some simple test gear.
A 50 Ohm 25W dummy load A signal generator And a SWR meter - power meter. The Dummy load was thirteen 680 Ohm 5W resistors wired in parallel and placed in a tin can with a coax socket. The resistors were chosen for having low reactance and did make a good 50 Ohm 50 watt shielded dummy load. My first signal generator was a crystal harmonic generator. A 4 point something megahertz crystal produced about a 50 uV signal on 2 metres due to the harmonic content of the oscillator and could be FM modulated. A normal carbon 100 Ohm potentiometer made a crude but effective variable attenuator, so the 50 uV 2 metre signal could be wound down to zero. The SWR - Power meter was constructed from an article in "Electronics Australia" and we had the basics for aligning and testing 2 metre radios. This equipment also was later used to align cavity duplexers. A tunable RF signal generator was later constructed using a commercial amateur VFO, shifted in frequency to 72 MHz and the 2nd harmonic made for a good tunable signal generator, along with a carbon pot for RF attenuation. Also added to the tunable signal generator was a sin wave audio generator to FM modulate the RF. Also constructed was a switchable low power RF attenuator that could select RF attenuation in 1 dB steps from zero to 60 dB. So with the home brew test gear, constructed over a couple of years, testing and aligning FM radios, and later cavity filter duplexers was made possible, as no body had access to spectrum analysers back in the 1960s and 70s. Calibration
-0.5 dB
-0.5 dB is a a power loss of 11%. If you start with 10 watts a 0.5 dB loss means you end up with 8.9 watts. Lets call it 9 watts, near enough. Hence to test the pass attenuation of a cavity filter, what ever power level you put in, expect 90% to pass through for the expected 0.5 dB pass attenuation per cavity. -35 dB
-35 dB is a big attenuation, how do you
measure, with simple equipment, - 35 dB...? What I did was construct a
35 dB RF attenuator using normal carbon resistors. This meant I now had
a substitute I could replace a notch cavity filter with and see if the
cavity notch and attenuator were the same, or there abouts.
Alignment
The
cavity filters were fitted with an adjustable capacitor for the high
side ones and the inductor ( a straight piece of wire 2.5" long) for
the low side ones.
The inductor ones did not require any changes in the length of the inductor, 2.5" had been worked out by W1GAN to be the correct amount of L. Using my signal generator and an FM receiver the cavity filter was tuned in frequency to maximum for the pass frequency. Then the signal generator was shifted in frequency to the notch frequency and a fine adjustment of the notch depth was made for minimum signal strength on the FM receiver's S meter by adjusting slightly the tuning frequency of the cavity filter. The pass frequency is not as sharp as the notch, so small variations to achieve the best notch had little effect on the pass frequency.  Simple setup for alignment of cavity filter pass and notch frequency As
said, the low pass cavity with the inductor did not require any
adjustment of the inductor. However to obtain the lowest pass
attenuation and the greatest notch attenuation often required a bit of
fiddling. The amount of coupling of the two coupling loops to the inner
resonant part of the cavity filter had to be adjusted sometimes to get
the best pass and notch results. This was fiddly, as it meant pulling
the cavity filter apart and making the adjustment then putting the
cavity filter back together.
The high pass cavity filter, with the variable capacitor, required learning about where the variable 5pF capacitor should be set before beginning alignment. If the variable capacitor was set to high in value, the correct pass and notch frequencies could not be achieved. After much experimentation I found, set the 5pF variable to about 3pF, then adjust the frequency of the cavity filter to the pass frequency, then increase the variable capacitor value for the best notch. The connecting coaxs
Once
the individual cavity filters had been aligned, one at a time, the next
job was to connect them together to make a duplexer. The article in the
ARRL handbook had dimensions for the interconnecting coaxs between each
cavity filter and the combining coax at the output of the RX and TX
filters to be joined and hence to the antenna. These coaxs shown in red
below.
All
coaxs are 50 Ohm
 Important interconnecting coaxs shown in red There was no explanation as to how the
lengths were chosen. The only coax that had a relation to 2 metres was
the coax between the high pass side of the duplexer and the antenna T
connection. The length of 26" is a half wave length at 2 metres.
A duplexer is a 50 Ohm in out system but at the notch frequencies this no longer applies. Each cavity filter is only a 50 Ohm device at the pass frequency. At the notch frequency each cavity filter is almost an open circuit and hence no longer presents a 50 Ohm impedance to this frequency. My only explanation as to the reason for the odd coax is to cancel out the impedance mismatch at the notch frequency. |