A Tale of Two ‘Plexers

‘Twas the night afternoon before Christmas New Year, when all through the house
Not a creature was stirring, not even a mouse;

Though, come to think of it, we likely have a mouse or two… or likely more than two…
Hard to keep them out, here in “the country”…

That aside though, with everyone getting nauseous at the sight of more eggnog I thought it’s time for some Hard Science!
Preferably Science of the Amateur Radio kind!

This is going to be rather lengthy with lots of pics and screenshots, and this story is absolutely irrelevant to our Kemptville Club, so feel free to hit the ‘delete’ button on your E-mail reader if this is not your cup-of-tea. It works better through the groups.io Web site to view long messages with pictures by the way.

First, let’s talk about “The Problem”

As some of you know I am the proud owner of a Yaesu FT-736R, dating back to ca. 1990, in fact, this is the very same one Rick (VE3CVG) was using. I’ve added some band modules, it now supports 2m – 1.35m – 70cm – 23cm, fixed a broken control board, and re-calibrated everything. Better-than-factory at this point! The issue with all those band modules is that they each have their own antenna connector, and I have just one single antenna for all of them (a discone, covers 2m through 23cm no problem).

Moving on to “The Solution”

By this point many of you are shouting “tri-plexer” or “du-plexer” (or diplexer, the difference between them is rather vague). I actually would need a “quad-plexer” but have not found such a device yet. So I got these:

Comet CFX-324 Triplexer
Diamond MX-37 diplexer

One was going for small money on eBay, the other was purchased new (could not find it on FleaBay). The antenna runs into the Diamond MX-37 first, that duplexer splits off 23cm, going directly to the 23cm module in the FT-736R. The lower-frequency output feeds the Comet CFX-324 Triplexer, where it gets split into 2m, 220MHz, and 70cm to feed those modules. What we want from these ‘plexers is that they have very low pass-through losses, while at the same time very high isolation between the ports. But do they?

That is the question I’ve been looking into, and I bet you want to know too! (One of those ‘too much time and too little to do’ situations)…

The two ‘plexers were connected up the same as they are for the Yaesu, with 50 Ohm loads terminating all the open ports. I had to get the Heavy Metal involved since I didn’t have enough small 50 Ohm terminators, this is what the test setup looked like:

Test setup di-triplexer

The spectrum generator was set up for 0 – 1.5 GHz, with the tracking generator switched on and normalized (the latter just means all the cable and connector losses were taken out of the equation, as is any variability in the tracking generator output vs. frequency).

Pass-through losses

First measurements were to determine the losses from the various ports to the antenna (and vice-versa). These ‘plexers only contain passive elements, inductors and capacitors, so they work equally well in both directions.

2 meters pass-through:

2m Pass-through

This picture shows what gets through from the antenna side to the 2m port. The Comet triplexer lists “0-150 MHz” and they are not kidding! It drops off a cliff right at 150. The 2m band shows around -0.74 dB worst-case loss through the two ‘plexers combined.

220 MHz pass-through:

1.3m Pass-through

For the 1.35m band the Comet triplexer lists “200-320 MHz”, that is what it shows on the spectrum analyzer too. Worst-case pass-through losses are around -0.65 dB for the 220 MHz band.

70cm pass-through:

70cm Pass-through

The third port on the Comet triplexer is meant for “390-500 MHz” and this picture roughly bears that out. For the 70cm band the worst-case pass-through losses measure in at -0.86 dB. Keep in mind that all of these are combined losses of the duplexer and triplexer connected together.

23cm pass-through:

23cm Pass-through

This shows a classic high-pass filter that is cutting off right around 900 MHz, though pass-through losses up to just about 1200 MHz are a bit large. After 1200 MHz it flattens out, with the 23cm band clocking in at -0.65 dB in pass-through losses.

Those values are not too shabby! The Diamond MX-37 claims -0.45 dB though, and the measured loss (for 1300 MHz) is a little more. Some of that are additional adapters and a bit of coax that was not taken into account with normalizing. The Diamond duplexer claims -0.25 dB for the lower-frequency port in pass-through, with 45 dB in isolation between the ports. That isolation is what we’re going to look at next!

Isolation, or inter-port attenuation

Isolation between ports, or how much energy gets from one port (say 2m) to the other (say 70cm) when transmitting, is at least as important as the pass-through losses. We really want to keep the magic smoke inside those hard-to-get gallium-arsenide MOSFets that are in the input stages of the various band-module receivers! To measure the isolation between ports the tracking generator was hooked up to one port, and the spectrum measured on another. All other ports were once again terminated with 50 Ohm.

2m to 1.3m isolation:

148-to-220MHz Isolation

The Comet triplexer claims “more than 40 dB” isolation between its three ports, but as the trace above shows it does not quite live up to that. Signals put into the 2m port show up with just 33 dB attenuation on the 1.3m port. That’s not great, but we will look at the practical implications later. A transmitter sending into the 1.3m port shows up -49 dB attenuated on the 2m port, which is OK.

2m to 70cm isolation:

430-to-146MHz Isolation

As the screen-shot shows, isolation between those two ports is pretty good! Basically 60 dB (or more) as that is the noise floor of the spectrum analyzer. I didn’t bother teasing a lower noise floor out of the Siglent analyzer, there’s no point in going lower than 60 dB, it is more than enough for isolation! This is in both directions, so 2m to 70cm as well as 70cm to the 2m port are down at least 60 dB.

1.3m to 70cm isolation:

450-to-224Mhz Isolation

This is where things get a little rough! As the spectrum analyzer shows there is a big ‘hump’ in passing signals through between those two ports around 350 MHz, and both 1.3m and 70cm are really still on the flanks of that hump. 70cm signals will pass through to the 1.3m port at 42 dB attenuation, while 1.3m signals show up on the 70cm port at 55 dB attenuation. Still within spec for the triplexer.

1.3m to 23cm isolation:

225-to-1300Mhz Isolation

This too is within specs, any 1.3m signals will show up on the 23cm port with 49 dB attenuation, while 23cm signals get attenuated at least 60 dB to the 1.3m port.

70cm to 23cm isolation:

430-to-1300MHz Isolation

This trace is a little more exiting, with 70cm signals showing up on the 23cm at 48 dB attenuation. The other way around, 23cm signals to the 70cm port get attenuated only 39 dB (which is less-than-specs).

2m to 23cm isolation:

No pretty screenshot for this one, I forgot to take one! It was a pretty boring straight line, with basically all frequencies showing at 60 dB attenuation. So no problems here!

What does isolation mean?

In more practical terms, what is the effect of the isolation attenuation between ports? I have run some numbers, based on the 25 Watt that comes out of the FT-736R on 2m – 1.3m – 70cm, and the 10 Watt from the 23cm module. These are the results of power showing up on “the other” port:

2m on 23cm = 25 microWatt
23cm on 2m = 10 microWatt
2m on 1.3m = 13 mW (!)
1.3m on 2m = 0.3 mW
2m on 70cm = 25 microWatt
70cm on 2m = 25 microWatt
70cm on 1.3m = 1.6 mW
1.3m on 70cm = 79 microWatt
70cm on 23cm = 0.4 mW
23cm on 70cm = 1.3 mW
1.3m on 23cm = 0.3 mW
23cm on 1.3m = 10 microWatt

The only one in that list that has me a bit worried is the 2m energy on transmitting that shows up on the 220 MHz antenna port. All of the band modules for the Yaesu have a narrow-band filter on their input stages, so it should still be OK. Damage power levels for the inputs are likely quite a bit higher than this.

What does pass-through loss mean?

We can do a similar thing based on the pass-through losses. How much of our precious output power actually makes it into the coax that feeds the antenna?

For this I used the Marconi 2955B to measure actual output of the transmitter modules, first without any ‘plexers, then with the du- and tri-plexer attached in the feed line. These are the results:

2m direct = 25.9 Watt
2m via du-tri-plexer = 22.4 Watt
Loss is -0.63 dB

220 MHz direct = 25.6 Watt
1.3m via du-tri-plexer = 22.6 Watt
Loss is -0.54 dB

70cm direct = 25.3 Watt
70cm via du-tri-plexer = 20.5 Watt
Loss is -0.91 dB

I did not measure 23cm because the Marconi tops out at 1 GHz. Based on the measured pass-through I would expect these numbers:

23cm direct = 10 Watt
23cm via du-tri-plexer = 8.6 Watt
Loss is -0.85 dB

In theory these should be the same loss numbers as measured in the pass-through category before. They are close, but slightly different. I am guessing that the measurement tolerance of the Marconi is to blame (it seems to have about +/- 0.2W granularity at these power levels).

These losses are small enough that I cannot tell the difference (I tried) between ‘direct’ or ‘via the di-tri-plexer’. The S-meter shows the same signal, they sound the same, even the noisy ones. It’s just too small a difference to notice. In an absolute border-line case it may make a difference, but I have lots of other sources of loss that would need fixing first before this becomes relevant (such as a better antenna).

Let’s get nekkid!

No good story is complete without an in-the-nude pic, so here it is:

Comet CFX-324 Inside

These are the guts of the Comet CFX-324 triplexer. It is actually surprisingly simple, though I suspect finding just the right values is not quite as straight forward. What it shows is a conventional high-pass filter at the bottom (the 390-500 MHz output), with 3 inductors and 3 sets of capacitors (split as two-per-location), making a 7th order high-pass filter.

The middle section starts out with a low-pass filter, that also feeds the low-frequency section, and it then branches off to a high-pass and band-pass section. From there it hits the 200-320 MHz port.

The low-frequency (1.3-150 MHz) section adds a 7th order low-pass filter to the mix before it hits the port.

I didn’t take the Diamond duplexer apart. The inside will likely be a number of L’s and C’s etched right onto the PCB. At those high frequencies small changes have a profound effect, and I would rather not mess with it.

And that is the tale of two ‘plexers!
Hope you enjoyed it.

A Merry Christmas and Happy New Year to all of you!


Find this useful? Share it with your friends!