23cm high-power hybrid combiner

23cm/1296 MHz high power 90 degree hybrid power combiner (by DJ9BV/SK)

30 years ago I built three copies of Rainers, DJ9BV (SK), 90° 23 cm hybrid combiner. The design was originally published in the DUBUS magazine (in 1994). Later on I got some modified dimensions for the coupler from Rainer, these seemed to work really well back then. Fast forward 30 years. I now have the need for some 23cm hybrid couplers again (for my new dish project). The ones I had back then are long gone (in the hands of another ham).

So I decided to try and redo the design from back them, this time using modern CAD and my usual "sub contractor" for milled aluminium and brass (same as I used for my 70cm hybrid combiners).


(I also make a low power version 90° hybrid combiner)


The original article in the excellent DUBUS magazine is copyrighted material, I got a written permission from Joachim ("Joe") Kraft CT1HZE to use and replicate the material on this page!

According to Joachim, I'm the only person (at all) that has permission to copy and publish the original article!


Original article from DUBUS magazine as a scanned PDF file (Thanks to Joachim/DUBUS!)

A small hybrid for low power (input of a dual PA setup).


Some theory behind the 90° hybrid in Pozar's Microwave Engineering (2012) book (page 363 and forward).


Drawing the combiner up in CAD from the dimensions was pretty straightforward and took less than 2 hours. The original design that I built had N connectors on all four ports, this time I wanted to use 7/16 connectors for higher power handling. The choice of connectors was easy, I designed around the same Radiall connectors I used in the 70cm combiner (Radiall order code R185403547).


After the design was done, I sent the design (STEP) files for manufacturing (2 x aluminium parts and 1 x stripline). The surface treatment on the two aluminium parts will be "bead blasted" and the brass stripline "mirror polished".


All production data are available here for download, using these you can have your own combiner produced.

I can also supply ready made and tested units, or complete kits (assembled by you), see below.


Some of the parts below can be ordered thru Scandcut.com in Sweden at very good prices. Check their website for what materials/thickness they currently support.


Production files

  1. Top aluminium part, STEP file
  2. Thread explanation for top part, PNG file
  3. Bottom aluminium part, STEP file
  4. Thread explanation for bottom part, PNG file
  5. Stripline (2 mm brass), STEP file
  6. Optional "jig" part for aligning stripline (see below), 4 needed, STEP file


Screws needed:

  1. 16 pcs DIN 912 A4 Insex M3x10mm for 7/16 connectors 
  2. 16 pcs DIN 912 A4 Insex M3x25mm for top/bottom aluminium parts
  3. 4 pcs DIN 912 A4 Insex M4x8mm for attaching bottom part to chassis (if needed)


Connectors needed:

  1. 4 pcs Radiall R185403547


Before assembly, remember to clean all threaded holes if needed. Sometimes you need to drive a tap into the holes to clear any chips still inside the holes from manufacturing. 


I can produce a limited number of the combiner for anyone interested (or you can have your own made from the data linked above).


Completed combiners is sold for €410,- each plus shipping (approximately 1.5 Kg).

Should you be interested in assembling and testing it yourself, the price is €370,- plus shipping (all parts, screws, 7/16 connectors are in the kit, only thing not included is the "Jig")


My original DJ9BV combiners from 30 years ago (1995). Was used for combining multiple 2C39 water cooled cavities on 23 cm for EME, first on my 5 meter dish and later on my 8 meter dish.

Connecting hybrids in a dual PA setup

Using my "QRP" 90° hybrid as an "input splitter" and the "QRO" 90° hybrid as an "output combiner", the connections must be done as shown on the figure below. It is VERY important to use the ports as shown below to get the correct phase. It also VERY important that the two cables on the input side and the two cables on the output sides each are the same length (however, the two input cables does not need to have a specific length as compared to the output cables, only both input cables must be of the exact same length (phase) and the two output cables muse be of the same length (and phase). Using the calculator here it is possible to see the effect of phase differences.


I highly recommend to test the setup using only low power and check that everything are connected properly before applying full power!

The two 50 Ohm terminating resistors must be able to handle sufficient power, as a rule of thumb, they should be able to handle 50% of the power from one amplifier as an absolute minimum. In the case of a malfunction on one PA module, 50% of the power from the still functioning module will have to be dissipated by the termination resistor. The output terminating resistor will also see approximately 50% of the power reflected back from the antenna.


In my 23cm dual PA, I use a 1500W attenuator on the output hybrid and a 10W on the input hybrid. Both these are monitored by one of my Dual RF Head-Analog modules that are connected to a REPAM module.

CAD model and design

I designed the combiner in Autodesk Inventor. The internal dimensions of the enclosure and the stripline was taken from the original drawings in DUBUS magazine (including updated drawings of the stripline from Rainer DJ9BV (SK)).


The connectors used in the original design was N connectors. I wanted to use 7/16 connectors instead. I chose the same as in my 70 cm combiner from Radiall (order code R185403547). These connectors are not cheap, but quality looks superb.

The center pin on these connectors ends up in a 7 mm diameter "cup" on the back with a 3 mm hole. In order to make it easy to "interface" the stripline to these, I made small protrusions on the stripline that would fit into the "cups" and be soldered in place. I also made the stripline 2 mm thick instead of the original 1 mm, the housing was modified for the added thickness, basically to maintain the defined 9.5 mm air between housing top/bottom and the stripline.


The walls of the milled aluminium box are 5 mm thick, this enables me to use M3 screws for holding the two parts of the housing together. However, this raises a small problem. The 7 mm diameter "cup" of the 7/16 connectors needs a clearance of 16 mm around it to match 50 ohm, As the "cup" does not extend all the way thru the 5 mm thick walls, there is a couple of mm from the end of the cup to the stripline where the impedance is not 50 ohm. 


Testing will show if this a problem (my best guess it is not too important). Should it be a problem, an easy fix is to mount a "collar" of 7 mm in diameter and 2 mm long, around the "pin" from the stripline just where it enters the "cup" of the 7/16 connector, effectively extending the "cup" all the way thru the 16 mm opening in the wall.



Jig for mounting the stripline

I made a small "jig" that will help with aligning the stripline, both in "X" direction and the height above the bottom. The stripline is designed so there should be 9.5 mm air above and below the stripline. Using four jigs, you can place the stripline accurately (note, the jig is NOT supplied in the "assemble yourself kit" !


The two jigs that are placed at the same wall as the connectors will make sure the stripline has the same distance to the wall on both the left and right side. All four jigs will also make sure the stripline is placed exactly 9.5 mm above the "floor" of the housing.


The file for the jig is available at the top of this page together with the rest of the design.

The first prototype from production

Parts for the prototype arrived. Everything was as expected, quality as usual very good from the "milling house" I use.


The housing, top and bottom, are both made of 6061 Aluminium, surface was "bead blasted" (don't use anodized!). The stripline are done in brass with "mirror polished" surface. The pictures below is exactly as the parts looked from production, the only thing I did to the brass, was to remove the thin oil the stripline was stored in (in a plastic bag).


Stripline is soldered to the 7/16 connectors (using a high wattage soldering iron and some flux). The 7/16 connectors are mounted using M3 stainless screws (DIN 912 A4, Insex, M3 x 10mm), top and bottom are assembled using M3 stainless screws (DIN 912 A4, Insex, M3 x 25mm).



Measurements on first prototype

Testing imbalance, Isolation and phase difference on the prototype combiner (all at 1296 MHz)

Return loss on all ports better than 24 dB.


All measurements are done on my R&S ZNLE3 VNA, calibrated with ZN-Z150 automatic calibrator (both devices are in active calibration).


Amplitude imbalance:

P1 to P2 -3.02 dB

P1 to P3 -3.03 dB

 

Isolation:

P1 to P4 -26.3 dB

P2 to P3 -27.9 dB

 

Phase difference:

P1 to P2/P3 <3 degrees


Measurements on production version

I had some combiners produced for fellow hams, I measured the same parameters as shown above for the prototype.

Return loss better than 24 dB on all four ports .


Insertion loss less than 0.02 dB (almost impossible to measure).


Amplitude imbalance:

P1 to P2: -3.02 dB

P1 to P3: -3.00 dB


Isolation:

P1 to P4: -29.4 dB

P2 to P3: -25.2 dB


Phase difference:

P2 to P3: 5.7 degrees


Below is a test setup with the hybrid and one of the QRP 90° hybrids. The setup shows excellent phase and amplitude match, the insertion loss is also very low, in practice it is dictated by the loss of the QRP splitter, the QRO 90° hybrid on this page adds practically nothing with regards to insertion loss!

Calculate phase difference based on output power and power on isolation port

I made a small Python script that can calculate the phase difference on the two ports of a 90° splitter/combiner setup. You need to enter the power measured on the output port and the power measured on the isolation port. You also give it the frequency (in MHz) and the velocity factor of the coax cables used.


The program will then tell you the phase difference between the two signals that corresponds to the power measured on the isolation port as well as the length of cable this corresponds to depending on the entered velocity factor.


REMEMBER! There are many things that can influence this, the program is just meant as a simple tool that can visualize the phase difference!

As an example, if there is not a perfect load on the output port of a combiner, typically 50% of the reflected power will be present on the isolation port (with the last 50% divided with 25% to each of the two input ports), this will "mask" the result, giving you the impression that there is a phase error between the two PA modules, when in reality some/all of the power measured on the isolation port is due to mismatch on the output port!


As a fun exercise, I asked Grok (the AI on the X platform) about the isolation port and mismatch:

(read with caution, but I believe the text to be pretty correct)

Grok on isolation port power in a Hybrid combiner