Tuning a ferrite Circulator …. thoughts on application Mike Pinfold ZL1BTB
May 2022
I had the need of a circulator for 144.650 MHz , our local branch 2 m linear repeater site has the dreaded infestation of growing pine trees such that if you stand on the radio hut roof you can no longer see the town or Lake Rotorua for that matter. The repeater runs two antennas of two five eighths in phase with a RX TX vertical separation of about 3 meters providing a measured isolation of -40 dB . If you climb up the antenna pole near the top where the separate receive antenna is mounted , you can see the lake and the town . The transmit antenna is now below the growing tree line and there is reduced distant signal strength around the repeater coverage area , the simplest way to attack the problem is to put transmit and receive function on the one clear antenna at the top of the pole in the open 'However with only 75-80 dB Tx Rx isolation in the duplexer system ,we are on the borderline for clean operation of the repeater if we share one antenna for Rx and Tx , We will need more Tx Rx isolation and the only easy way to provide that with the system we already have, is the use of a Ferrite circulator . ACirculator is a ferro magnetic RF device which does what the name suggests in that it circulates signal around the three rf connections of the device in mainly one direction at low loss . .Typically it has 3 ports and their is an easy rotation of all rf signal around in one directon and a much harder if they try to travel in the opposite direction ..It has directivity ! So it is able separate signals in traveling indifferent directions with low loss.
I had among my junk, a circulator that would perform at 144 Mhz provided it was tuned . This is the catch cry for a circulator to work well at a specific frequency or frequencies , it needs to be well matched to the three loads on its ports . Nominally they are designed with matching networks on each port to see 50 ohms resistive at the designed frequency , though other impedances can be accomodated .. If you look at the performance data of an rf circulator. its reverse isolation (back feeding) in dbs is very much related to vswr it sees on the ports. If there is a good match between load ( antenna) and ( circulator) then this port to port isolation may be very good, in excess of -20 dB (backwards feed) with a forward loss of under +1 dB. As the VSWR climbs so does the ( leaking backwards signal ) hence isolation begins to fall .
I used my VNA , Agilent E5061 for
the tune up, though a little nanoVna would be more than up to the task. Or maybe
a simple VSWR meter and low power transmitter ???Normally you pass a signal
through the circulator in the low loss , forward direction of S12 looking at the
loss while you adjust only the matching network on the 50 w Load , you then
rotate the 3 connections by 120 degrees the retune the next port (only at the 50
w load ) for good through signal and so on rotating around the circulator a
couple of times , you improve the match each time. Intermittantly checking the
reverse isolation ( the backward leaking signal) ( N to BNC adapters would have
been less hassle) You should start to see a difference between the low loss
forward signal of under 1 db compared to the isolation signal of >20dB. I found
this technique a pain and didnt get far at all, other than wearing out N
connectors on the circulator .|
performance graph from M2Global website
Gotta be a better way
..Now Circulator performance is directly connected to the port matching hence
resultant VSWR ( see the above graph) , so I tried another approach . I would
tune the matching of each port for minimum VSWR using S11 or S22 and just go
round and round the terminated device with one cable , until the matching showed
a low VSWR .
I marked the frequency of interest ie 144.650 MHz on the VNA and then tuned for
minimum VSWR on that frequency and after a few port rotations , it did it in no
time . Reset to maglog and S12 or S21 I checked the reverse isolation figures on
each port all terminated in 50w ( easy on the E5062 with a press of a button)
and they were getting at the -20db level , and then went around and around
tweaking the “isolation” notching only on the 50 w load port for maximum
isolation , I could easily exceed a narrow band isolation of over +30dB or more
, but be wary , I dont know how thermally stable circulators are . They may
actually drift over time so dont get greedy with the isolation figures. A broad
band -20 dB is a 100 times power reduction , thats a lot of signal suppression ,
Now we can hope to have a minimum of 95 to 100 dB Rx Tx isolation , I hope
that's enough without having to add more coaxial 2m cavities into the mix ,I
will sneak up the hill when the weather drys up a little and Try it out . Plug
the RX feed into one port of the circulator , the shared antenna into the
“middle” port and the TX into the remaining port (with respect the the rotation
of directivity) and see what happens .
Fingers crossed if all goes well, our repeater coverage received signal levels
at a distance , will be back to normal .
The next experiment will be to
measure the circulator port impedance with no matching circuit then
hopefully derive a matching network with the smith chart for
the frequency of interest