In the words of the Monty Python team, “Now for something completely different!”

The tuning of Duplexers and cavities filters is not an experience to be taken lightly. The professionals have a luxury of sufficient bandwidth and spacing between their Transmit and Receive frequencies. It is only the amateur radio fraternity that would consider using only 600 kHz shift on VHF, and it is only thanks to the available width on 432 MHz that we can accomplish something relatively simple. However being a shared band in Region 1 it is none the less important.

Nothing so complex should be attempted without the correct equipment, which in technical terms means a spectrum analyser and an injected frequency source.

The object of a cavity filter is to block unwanted emissions and permit the wanted frequencies with the least loss, therefore in the example that follows I shall attempt to tune a Procomm duplexer for the UK Amateur 70 centimetre allocation for a 9 MHz split, with a transmit frequency of 430.4625 MHz and a receive frequency of 439.4625 MHz. What this means is to block the TX frequency side from the RX frequencies and the RX frequency side from the Tx Frequencies. I do not have the luxury of a spectrum analyser, but I do have two other pieces of equipment that will serve almost as well.

The spectrum analyser with some sketches of successful notch filtering

The Minature Spectrum analyser pictured works very well, but is difficult to manipulate as there was never any manual, and the settings are largely guesswork. But the unit I shall show in this demonstration is the NanoVNA, its name indicating its size.

The NanoVNA with attachments for N-Type

It is a fairly simple and inexpensive device, that arrived with all the necessary cables and attachments except I had to purchased the SMA to N-Type plug to use with most of the cavities to which I have access. Before use a full charge on its lithium-ion battery is required using a USB charger equipped with a USB-C connector.

An essential piece of equipment is an HB pencil to use on the touch screen when turning it on, as the semi-circular spring switch at the top is a bit hit & miss. Leaving the screen protector in place is advised.

Turning it should open the device to all the traces turned on. With the pencil on the Display Menu turn all the traces off except one. On FORMAT leave LogMag as your standard method for these procedures. On SCALE set ScaleDiv as 10 – tap 1-0-x1 and Reference postion as 10 – tap 1-0-x1.

Now go to the STIMULUS menu and select CENTRE and enter the frequency 439.4625 not forgetting the decimal point, then touch M. As we do not know exactly where the current notch is set we need to observe the whole of the band so select SPAN and select 20 M for 20 MegaHertz. We are wanting to put this notch on the TX side of the Duplexer, so with the 50 Ohm load on the unused side of the duplexer I attach the channel 0 side to the antenna socket and the channel 1 side to the remaining connector.

The NanoVNA connected. Note the 50 Ohm load on the RX Side. We can already see two notches.

If we see no visual indication of the transmission by the NanoVNA on the screen, widen the SPAN. Ultimately we will see a peak and a trough or more than one trough somewhere in the bandwidth displayed. The object now is to move these troughs into the same position as the frequency 439.4625, which we do by adjusting each of the screw ends on the cavities on this side of the duplexer.

Three cavity notches – Note the span is 20 MHz, so we have a way to go yet.

Eventually we see three troughs in close proximity of the frequency which should eventually dip into the bottom of the trough in the very centre. By gently adjusting each of the screw, eventually these troughs converge to make as near as possible a single trough. By reducing the SPAN on the screen, we can make the appearance of the single trough more defined and thus more accurate. We should be aiming for 91db as an optimum for this type of duplexer. However we note that each manipulation changes the wave form often in the wrong direction, so it is a very careful balancing act to maintain the notches together at the same time maintaining the most profound notch we can achieve.

Nearly there – aiming for the most profound around 91 dB, 86.87 db isn’t bad

A finished tune with the span set at 10 MHz, should show the minimum of loss on the desired frequency, and the maximum of notch on the rejection frequency. It may not indicate a good standing wave ratio without a proper load, but rest assured, if the results are that good we can’t go wrong. Tuning the other side of the duplexer is of course the reverse, procedure permitting the transmission of the higher frequency and notching out the 430.4625 MHz.

To the left of the trace is the band pass and to the right the frequency notch.

Setting the SPAN to the whole 10 MHz band width and rotating the frequency with the fixed wheel switch we can clearly see a profound notch on the 439.4625 MHz and minimal losses at 430.4625. A successful tune.

Basically we have a miniature transmitter which we are seeking to attempt to view the waveform, with its own in-built receiver, at the same time as we are attempting to block with tuning the cavities to the same frequency. Easy really.

This is an example of tuning a single cavity on a VHF repeater for 145.00 MHz Receive-side. You will notice the band-pass at our desired frequency 145.000 MHz, and a very strong rejection notch at the unwanted frequency of 145.600 MHz. This permits 2 Metre repeaters to work in the limited band-width available on this band. The cavity being tuned is beneath the meter. In this repeater installation there are three receive-side cavities and two transmit cavities connected by tuned lengths of cable.
No-one in the professional world has this inconvenience with which to work. I mean that the 144-146 Band is very restrictive for duplex devices. Most professional repeaters have Several Megahertz between Transmit and Receive. It’s fortunate that the 430-440 MHz band is a bit more friendly in this regard.

Leave a Reply