23cm 600 W Power amplifer

As I'm waiting for the mechanics to be produced for my coming 4.8 meter dish for 23cm EME, I started looking into building a suitable power amplifier for the system.

Again to "get the most" out of the amplifier and at the same time, save some space in the lab/shack, I decided from the beginning that the PA should be mounted near/at the dish. This will allow me to use the shortest possible coax between the PA and the feedhorn, and also remove the need for flexible "around the rotor" high power/low loss cable to be used. The PA will be remotely controlled and monitored using my REPAM and Dual RF Head modules.

I used the same principle in my 500W 70 cm power amplifier used in my current 70cm EME system.

Kit assembly instructions from Jim W6PQL

Schematic of the PA module from Jim W6PQL

Test procedure from Jim W6PQL

Datasheet for the MRF13750HR5 device from NXP.

The NXP device can be ordered from Jim W6PQL or from f.ex Mouser.

I found the LDMOS at Arrow for 50% of the usual price! Link to Arrow

Soldering the LDMOS, application note from NXP.

So far, the system consists of the following sub modules, some designed by myself, and some purchased:

Cooling system with two fans. I designed a "generic" system that can be used in other PA systems.
Overcurrent detection and on/off switch. Described and documented here also.
Dual RF Head (logarithmic amplifier) that samples the fwd/ref signals from a directional coupler.
Remote PA Monitor (REPAM), allows control and data collection/security of a remote PA.
W6PQL 600W module (I selected the "kit form" of the modules, "assemble yourself")
Power supply Flatpack2 HE 3KW, connected using my interface board.
Various stuff for mounting, a relay for remote power/on off, a 12VDC PSU (for bias etc.)

If you don't want to source the mechanics shown here yourself, I can deliver the parts for you.

A complete set for one 600W W6PQL PA module consists of:

The box (200x120x60mm), with no surface treatment ("as milled").
Cover for the box, with no surface treatment ("as milled").
Copper heat spreader for the PA (surface is mirror polished)

The parts are as shown below under "enclosing box".

All parts threaded and ready for installation!

Please note, this is ONLY the CNC milled parts, no connectors, fans, cables, PCB etc!

Price for item 1) to 3) above is Euro 295,- (2 sets for Euro 275,- each)

Email me for details, info on the My CV/Contact page.

Design files

Here you can find the design files/production files for the mechanical parts of the system. Will be uploaded once they are checked for fit etc.

Copper heat spreader (10 mm thick copper, mirror polished), STEP file
- Thread instructions for copper spreader, PNG file
Enclosing milled aluminium box, STEP file
- Thread instructions for box, PNG file
Lid for enclosing aluminium box, STEP file
Brass "strip" (2 needed) for in- and output coax, STEP file
Drill template for copper spreader and box (for drilling holes in heatsink for both), STL file (for 3D printing), also as a STEP file

Apart from the CNC milled parts and the actual PA PCB with components, I used the following:

1 pcs 60x60mm fan for the cover, link to Mouser
1 pcs 60x60mm fan guard, link to Mouser
6 pcs ferrite cores (1 for Bias, 1 for temp sensor and 2 x 2 for drain power supply), link to Mouser
1 pcs N female bulkhead connector for 0.250 inch semi rigid cable, link to Mouser
1 pcs N female bulkhead connector for RG-402/0.141 inch semi rigid cable, link to Mouser
2 pcs feedthrough capacitors/Pi filter, link to Mouser, alternative link to Mouser
32 pcs M3x10 mm screws
8 pcs M4x10mm screws, 8 pcs M4 lock nuts
2 pcs M4 feedthrough capacitors, link to SV1AFN
RG-402 semi rigid cable for input, link to SV1AFN
RG-401 semi rigid cable for output (this is just one possibility), link to Mouser

I also used a DS18B20-PAR and a "Terminal Ring" for the temperature sensor, see details below.

This all depends on how you will monitor the temperature etc of the PA module, I use my REPAM module for such things (Fan control, temp monitor, current monitor, SWR etc).

PA/Heatsink/Fan assembly

The completed assembly, PA module in milled aluminium box and heatsink with fans will look like this.

This is the exact same heatsink/fan assembly as I use in my 500W 70cm PA module.

Below is a dual module solution, also shown is my 23cm high power 90 deg. hybrid.

Two different ways of mounting the modules are shown.

The side-by-side assembly below shows some of my other projects:

Enclosing box

I decided to design a milled aluminium "box" that would contain the PA module, connectors for in- and output and a few feedthrough capacitors. Jim (W6PQL) suggested that it was a good idea to cool the output components on the PA module as they can get warm at full output. I designed the lid for the milled aluminium box so a 60x60 mm 12V fan could be mounted.

I also designed a copper heat spreader for the module and had that manufactured at the same time as the milled box and lid. Shown on the screenshots below are also two small brass strips that will be used to solder the coax cables to, these will then be secured to the PCB/heat spreader with two M3 screws.

The extra M3 screw in the "LDMOS trench" is for a temperature sensor (1-wire DS18B20 sensor that connects to the REPAM module). This threaded hole can ONLY be used for a sensor if you don't use the PTFE PCB Clamp from Jim W6PQL, otherwise there is no room for a sensor at this position! I decided to let the threaded hole stay in the design as there is a number of people who makes the own "PTFE PCB Clamp", they might be able to use this M3 hole for a sensor. In my case, I mounted the 1-wire sensor in one of the 4 holes around the LDMOS, see pictures below.

The data linked above for the parts are for version 2 of the box. The first one, as shown on the screenshot/pictures below, the box had a height of 40mm. I could see a slight influence on the power out and gain when putting the cover on the box (efficiency seemed to stay the same, only gain was lowered a tiny bit). To minimize this effect, I made the box 60mm high instead of the original 40mm.

Drilling template

Designed a drilling template that will help place all the holes correctly in the heatsink when it is being drilled. This one I just 3D printed as it is a "use once and throw away" type of device.

Mechanical parts received

Received the mechanical parts for the PA module, copper heat spreaders and enclosing boxes, back from milling.

Connectors and heatsink preparation

Finished in- and output coaxes for the two modules. Also used the drilling template to mark the holes needed in the heatsinks. Tried some simple mockups, everything seems to fit as planned.

Cable on input is RG-402 semi rigid, on output it is RG-401 semi rigid.

Connector used on input: Amphenol 082-6099-RFX

Connector used on output: Amphenol 082-6163

3D printed the first of two ducts for the two of four fans. The design for the duct is here.

Soldering of LDMOS devices

Soldering LDMOS is always exciting! You have to be EXTREMELY CAREFUL when handling stuff like this as it is VERY HOT (200+ degC)!!!

The soldering process follows the instructions from Jim W6PQL outlined in his video. This method has proven to work fine over the years in countless PA constructions.

I shot a video of the process.

Initial tests

The first few things to be done, is setting the idle (Idq) current between 1.6 and 1.8A according to Jim W6PQL. There is a small trimmer on the board that adjusts this. It is VERY (VERY!) important to set this to the lowest possible bias voltage BEFORE the 50VDC are turned on! If the bias voltage is too high, you risk that the LDMOS will draw excessive current once the 50VDC are turned on!!

Remember: "Measure twice, solder once" !

The bias trimmer was set to the lowest possible bias voltage, in my case it was around 1.4V. At that setting, the LDMOS did not draw any current at all when 50VDC was turned on.

Also important is the first tests MUST be done with a current limited power supply, set maximum current to around 3A.

If you don't have a current limited power supply, don't think a normal fuse will help you, it will not be able to react fast enough!

After the idle current have been set between 1.6 and 1.8A, you need to check the return loss (SWR) on the input of the amplifier. Connect a suitable dummyload (does not need to be high power for this first test) on the output, and use a VNA to measure the return loss on the input.

On the input side of the amplifier, four small "pads" are located next to the input connection pad on the PCB board. By connecting one or more of these pads to the input pad, you can move the point of best return loss down in frequency (more pads connected means lower frequency for best return loss). The Idq also affects the return loss.

On my modules, I needed to connect one of the four pads to get the best return loss around 1296/1300 MHz at an Idq of 1.6A. The return loss changes a little bit depending on if the cover is on or off on the aluminium housing around the PA module (putting the cover on, lowers the best RL a bit in frequency (due to the added capacity probably)).

Below is a plot of the return loss for both the completed modules.

Test setup for initial tests:

Module #001 return loss, sweep from 1290 to 1310 MHz:

Module #002 return loss, sweep from 1290 to 1310 MHz:

Module #001 input match using one of four pads:

Module #001 final view after adjusting pads on input and output:

Notice that on the pictures shown above, the PTFE PCB clamp that holds down the PCB around the LDMOS are NOT yet mounted. It should be mounted before doing any long transmissions, see Jim W6PQL description.

Below are pictures of both modules with the PTFE PCB clamps installed and also the 1-wire temperature sensor (DS18B20 based) that will connect to the REPAM module for temperature monitoring/alarm and fan control.

The sensors consists of a DS18B20-PAR sensor and a Terminal Ring. The DS18B20-PAR are glued inside the Terminal Ring using AG TermoGlue. A two wire cable are soldered to the terminals of the DS18B20-PAR, heat-shrink are mounted on the finished sensor. The cable are then wrapped around a ferrite core three times, connection to the outside is thru a 1 nF feedthrough capacitor.

The 50VDC power inputs and the 12V bias input also uses the same ferrite core.

NOTE: While testing the dual module setup (at around 1.1KW) I noticed once in a while there was a communication failure with the temperature sensors in the modules. I relocated the DS187B20 sensors to the screw that sits right under the bias ferrite. This helped and so far I have not seen any problems even at full power.

Power output

Next up was testing the RF performance of both modules.

I connected a 50V/60A power supply to each module at a time. Played around with the pads at the output connection as Jim W6PQL describes, it seems that I need two pads connected for max output/efficiency.

I did some measurements on both module #001 and #002, they behaved very much like the same with regards to power out, gain and drain current. Below is some screenshots of the measurements.

For this test, I was limited in drive power (max around 8.2W) from my IC-9700. Later I will try with more drive power (using one of my small 23cm "driver" PA modules.

Module #001:

Module #002:

Comparing module #001 and #002 power output:

Comparing module #001 and #002 drain current (Idd):

First test of dual PA

As I'm building the two modules described here in an outdoor enclosure for my new 4.8 meter EME dish, I did some preliminary testing of the setup. Below are some data form the PAMonitor application that interacts with the REPAM module in the PA system. More details about this setup on the 4.8 meter EME dish page.

Maximum power input to the input splitter was around 17W in the test below.

Carsten Groen's (OZ9AAR) personal place