KX3 Heat Sinks for WSJT and other Digital Modes
The popular Elecraft KX3 is a very flexible, small footprint HF+6m SDR (Software Defined Radio) that can be used without a computer.  However, if one wishes to use the radio for digital modes such as WSJT, some interface between the computer and the radio must be used to isolate audio between them and provide for other computer communication functions.  This is outlined elsewhere.  The purpose of this site is to discuss my experiences with use of the KX3 on the full duty digital mode of JT65A. 

When I first received my KX3 in the summer of 2012, I ran JT65A  for an hour (3w output, 10.0 VDC during transmit), and the OSC temperature peaked at 42C and finally cooled to 39C just before the next transmit period started.  Since my primary interest is to use the KX3 on JT65A mode for 6m EME, good stability essential.  Initially, the drift during transmit and receive periods was too  extreme for reliable JT65A operation on 6m under weak signal conditions.  Elecraft came up with a great temperature compensation routine which became part of the KX3 firmware, and it made the frequency very stable during full duty cycle digital mode operation on HF. 

However, I still noticed a very pronounced frequency jump as things began to warm up at the start of each transmit sequence up on 6m, plus a slow steady drift during the rest of the cycle.   I thought that if the OSC could be tied to a larger mass, it might be less prone to change frequency so readily.  I also wanted to make sure that the KX3 could function well even in higher ambient temperatures than my very cool basement ham shack, since I have tropical DXpediton locations in mind for it.

I found what appeared to be a manageable aluminum heat sink on EBAY that would be large enough for me to put across the top of the KX3 (to cool the PA) and still provide enough material for a heat sink on the bottom case of the KX3 for the OSC.  The particular piece that I chose was offered by seller "goldpart" and described as "100x220x18mm Aluminum Heat Sink for Electronics Computer Electric equipment H168", weighing 472 grams.  It cost $19.63 and was quickly shipped from China.

I cut off a 7.5" long x 1.7" wide piece to mount on top of the KX3 in place of the standard flat heat sink.  I still was able to cut a piece 6" long x 1.45" wide to fit on the rear/bottom of the KX3 to act as a heat sink for the OSC.  The heat sinks were mounted with additional 4-40 x 3/8" long screws with split ring lockwashers and nuts on the inside and two washers and a lockwasher on the surface of the heat sink.   This made the tip of the screw flush with the nut surface on the inside of the KX3.  The only exception was the 4-40 x 3/4" long screw used in the top to pass through the hole in the KXFL3 board.  Silicone "Type Z9" heat sink compound was used between the case and both heat sinks.


Locations for the screws were chosen so they would avoid interfering with any components inside the KX3.  The two existing holes in the rear and four in the top were used, with additional screws added (4 more in the top and 3 in the rear).  AJ7LL was kind enough to donate his time and milling machine to remove sections of heat sink fins so I could fit the screws and washers into the mounting holes. The installed heat sinks are shown.  Note that the heat sink on the rear still permits the KX3 to be supported by the feet.

 The location of the OSC on the RF board was kindly provided by Elecraft engineering.  In order to couple the OSC to the rear case and heat sink, product called "Gap Pad®" from Bergquist Company was used.  Gap Pad® is a thermally conductive while electrically insulating material that is pliable and somewhat compressible, depending on which particular product is used.  For this project, I chose their Gap Pad®  5000S35 material, which had the highest thermal conductivity.  The pads come in small sheets with a paper backing on one side and a clear backing on the other.  Both sides are somewhat tacky, and are meant to adhere to a case or circuit board.  However, the side with the clear film backing is somewhat stickier.  

The photo below shows the rear cover with the heat sinks installed.  You can see where located the additional screws.  In order to couple the heat sink to the PA transistors, heat sink compound was applied to the case before re-installing the RF board.  

As you can see from the photo above,  piece of .100" thick Gap Pad® was installed on the inside of the case after thoroughly cleaning it with steel wool and degreaser. The stickiest side was placed against the case.  A 1.5"x.875" piece of .100" thick Gap Pad® was placed as shown over the OSC, but positioned so it did not cover any components that were higher off the RF board than the OSC.  When the RF board was re-installed, the gap pads were the proper thickness to connect and bridge the thermal gap between the OSC and the case, with easy compression of the Gap Pad® materials. 

The temperature calibration routine was done again on the KX3 with the heat sinks installed.   Then it was set up as shown in the photo below, transmitting JT65A into a dummy load.  During transmit sequences, the KX3 display showed 10.0 VDC at 1.7A.  After two hours of JT65A operation (in my 10C basement ham shack) at the KX3's three watt output setting, the OSC temperature was peaking at 35C after a few seconds, and rapidly cooling down to 33C (within 5 seconds after ending the transmit sequence).  The PA temperature peaked at 41C and also cooled down to 33C.  Note that the maximum temperatures were much lower than those experienced without the heat sinks, and the coolest temperature was very quickly reached as soon as the transmitting sequence ended.   In order to successfully use the KX3 Temperature Compensation routine, I needed to add heat to the KX3 to get the OSC up to an "ambient" temperature of 35 C for the initial calibration.

The trace of the transmitted signal (JT65A sending "73") still seemed to have a  slight "hook" of a few hertz a during the first few seconds of each JT65A transmitting sequence as the OSC temperature quickly ramps up to 35C.  I suspect there may be a way to correct for this hook when the temperature suddenly increases, by re-running the temperature compensation routine again more slowly and carefully.  However, I am not sure it is necessary to go any further at this point, since the KX3 appears to be functioning in a very stable manner on 6m. 

Note that the settings shown on the WSJT SpecJT waterfall below are not typical normal operating conditions, but were set to help attenuate the KX3's tranmit signal, in order to be able to see the most detail.

I have not yet verified the currently configured KX3 performance through actual on-the-air 6m EME tests, but expect that it will also be able to receive the JT65A signals without drifting since the OSC so quickly returns to its resting cool temperature, before the other station's transmit sequence begins.
 The use of the Elecraft Temperature Compensation routine that was developed and included in the KX3 firmware, plus the addition of heat sinks, appears to provide adequately stable frequency during JT65A operation on 6m.  Many thanks to the patient and creative folks at Elecraft for their ingenuity with the firmware, and also for their suggestions regarding the magic Bergquist products. 

It is not known whether the stock KX3 with only the newly revised Temperature Compensation firmware would by itself exhibit similar performance.  However, the addition of the heat sinks in combination with the Temperature Compensation routine does appear to have made a substantial improvement compared to the initial attempts with the firmware routine alone.

Lance Collister, W7GJ
Revised 16 November, 2012