How do you set up a cavity filter using a NanoVNA?
XMIT----[F1]---[F2]---[F3]---| | |--- ANTENNA | RCV-----[F4]---[F5]---[F6]---|
What do we want? We want to be able to send and receive at the same time and using only one antenna. This sounds like magic but it is actually not only possible but it also works very well.
What do we ACTUALLY want? We want two “50 ohm frequency isolated impedance paths”. One between the antenna and the receiver and one between the antenna and the transmitter.
A cavity tube filter is a wonderful thing. 🙂 The ones I have have two N-connectors, a rod to set the PASS frequency and a variable capacitor to set the BLOCKed frequency. So, the rod is used to set the frequency we want to PASS the filter, and the capacitor is set to BLOCK another frequency (not far from the PASS one). The PASS frequency will be somewhat attenuated (less than 0.5dB) and the pass band is not very narrow. The BLOCK frequency will be attenuated by about 35dB and is a quite sharp dip.
Another beautiful thing is the NanoVNA. You can buy them on many sites, most of them having SMA connectors. My latest one, however, has N-connectors, N-cables and N calibration sockets. Almost ideal for cavity tuning.
Ok, let’s say we want to calibrate our filter for using a transmit frequency of 145.750Mhz and receive frequency of 145.150MHz.
- Calibrate the NanoVNA, CABLES CONNECTED, for the frequency 144.5MHz to 146MHz or thereabout.
- Connect the VNA to each cavity filter in turn. Tune F1, F2 and F3 to have a PASS frequency of 145.750MHz and a BLOCK frequency of 145.150MHz. Tune F4, F5 and F6 to have a PASS frequency of 145.150MHz and a BLOCK frequency of 145.750MHz. Now we have a good starting point.
- Now we can connect F1-F2 and F2-F3. Connect the VNA to the remaining ports on F1 and F3. Carefully fine tune the PASS frequency to get the lowest possible loss on the PASS frequency. You should also see a deep dip on the BLOCK frequency. Unfortunately the NanoVNA cannot see the depth of it since -70-80dB is its lower limit. Never mind for now. Do the same fine tuning with F4-F6.
- Now we should have well tuned “50 ohm frequency isolated impedance paths” through F1-F2-F3 and F4-F5-F6. Connect F1-F2-F3-[T-connector]-F6-F5-F4. Connect your antenna (you checked it has a good SWR at your frequency, right?) to the T-connector. Put your 50 ohm calibration plug on F4 to act as the receiver impedance.
- Switch the VNA to show SWR. If the antenna is (very) good, you should see it getting close to 1:1 at your PASS frequency. If not perfect, adjust the rods on the three involved cavities. A little on each rod, adjust until it cannot get any better. You should get close to 1:1.
- Connect the VNA to F4 and put the calibration socket on F1. Adjust the PASS frequency of F4-F6 in the exact same way as F1-F3.
- We most probably changed a few rods. Unfortunately we must now disconnect all cables, adjust all 6 BLOCK frequencies again (be very careful to get it spot on). Connect all cables (and the antenna) again and check the SWR on both paths again. Not perfect? Go back to 5. Don’t forget the 50ohm plug.
Are you having problems getting it to be perfect? The cables connecting the T connector from F3 and F6 are a bit critical. When the filter is working as we want the impedance for the PASS frequency of F1-F3 through F4-F6 is very high. This means that when the transmitter is transmitting all goes well up to the T-connector. Then, because F4-F6 have high impedance for the TX frequency, the cable between the T-connector and F6 appears to be a loose end! Or, not quite…. The cable PLUS the internal ‘antenna’ in the cavity appears to be a loose end. If this is an issue for you (never had to investigate, my cables were ok) you will have to experiment using different length cables here. One cable plus internal cavity parts should be a 1/2 (right?) wavelength to cancel itself out as an open end coax. And this goes for both cables to the T-connector. Fixing only one of them will give you problems with the other one.
[The Smith chart of the VNA should be able to help you (I have not tested this, I will). Setting the BLOCK frequency of the three cavities and measuring reflection with VNA connected instead of the T connector should give you a good indication when the cable length makes it cancel itself out.]
If everything seems ok connect the transmitter and receiver and have a go. Don’t blame me if things blow up. 😉
The problem with using only the NanoVNA is we cannot for sure see that the BLOCK frequency of F1-F3 and F4-F6 is correct when we connect three cavities in series. The three (if perfectly tuned) cavities brings the BLOCK signal down by 105dB or so, and the NanoVNA can only cope down to -70dB or thereabout. However, we CAN note that the BLOCK frequency is at least 70dB down and the “walls” of the graph tells us we are probably down a lot more than 70dB at the BLOCK frequency. But it would be nice to be able to do this last fine tuning. If anyone has a good solution please let me know.