Testing “generic” MOSFETS for RF PA use

Earl, 4Z4TJ / VA6TJ

Low cost plastic MOSFET transistors can be used to make effective low power HF PAs.  I have an amplifier I made with IRF510’s that effectively brings a Softrock transceiver output up to legal QRP (5 W out) with a 15 volt power supply.  With that experience in mind, I started looking for plastic MOSFETs that were actually designed as RF amplifiers, and not as components in a switching power supply.

A while ago I bought some supposed RF MOSFETS on eBay from a supplier in China.  I did not take sufficient care in checking out the history of the parts that are marked MS1307 from International Rectifier.  A bit of Googling around showed that IR never made a transistor designated MS1307.

I decided that the transistors needed to be tested under actual RF amplification conditions, and not just with a multi-meter.  Therefore I build up a simple test circuit,  based on a 40m CW xmtr by VA3IUL, where I could plug in a MOSFET and test it as a 40m amplifier.   The transistor socket was taken from the wiring harness of a discarded ATX computer power supply.  The source of Q2 (tied to the emitter of Q1) needs to be grounded to obtain output – this is where the CW key would be attached, or in my case I soldered in a micro switch.

Capturebig

Capture2

Capture3big

0.5 W out into 50 ohm load through LPF (5 v/division;  10 V P-P).  Supply voltage = ~12 V

 

Initial results can be seen from the photos on the previous page.  The transistors do amplify in this circuit and I can get a nice waveform if I adjust the gate bias (R5) voltage to give a maximum 1 W output into 50 ohms.  Adjusting the bias to exceed this power level causes the scope trace to turn ugly showing that bias has shifted the transistor off the linear portion of its load-line.   What needs to be done to determine the operating limits is to vary the input level to the MOSFET gate while maintaining the bias so the transistor is operating in its range and at different supply voltages.  This is a good reason to buy a nice signal generator  J

With the simple setup shown in the schematic at a supply voltage of 13.6V, PA voltage gain as measured with the scope probe, was determined to be approximately 15 db.  Vin = 4 V P-P and Vout into 50 ohms = ~25 V P-P.

After this test, I thought about improving the impedance match into the MOSFET gate and buffering the output of the crystal oscillator.  I adapted a section of a circuit from one of my Softrock transmitters and came up with this:

This circuit did not do much to improve the waveform going into the gate of the MOSFET but it did permit higher output power with a nice sine wave after the LPF.  I could easily obtain  1.5 W out with a 13.6 V supply and a MOSFET gate bias voltage of 4.0-4.1 V (measured with open source).

 

Capture4big

 

 

I also tried reducing the bias on the oscillator transistor Q1 base by lowering the value of the base to ground resistor to about 2/3 the original value.  This did not have any noticeable effect.

 

Capture5big

 

Breadboard with the added buffer amplifier and 9:1 input transformer to the MOSFET gate

 

I do not consider this circuit to be a useable transmitter.  It’s a test jig for performing  a rudimentary check on RF MOSFETs or picking a matched pair.   As a transmitter the circuit is not appropriate: the leads are too long; there is no thermal protection in the bias, etc.  For these reasons I am not going to invest effort in testing at a higher frequency.    However, I did listen to the 7.3 MHz output (into a dummy load) on the station transceiver and, in fact, the CW sounded just fine, no key clicks, shifting notes or other QSD.

After a suitable number of cups of coffee the next morning, I realized that L1, R6 and C2 are to zero beat (pull) the crystal frequency when operating CW and are not needed for a test jig.  If you remove them, it may be necessary to increase the value of C1 .

 

 

 

Planned improvements:

  1.  Add an ammeter to be able to determine DC efficiency.
  2. Regulate the bias voltage supply.
  3. Compare RF amp results using different switching power supply transistors
  4. Determine the effect of increasing the supply voltage to 24 V.

 

I recommend checking out VA3IUL’s website http://www.qsl.net/va3iul/ where there are many circuits and ideas for projects.  Thanks to Iulian for his assistance with this little project.

 

Earl, 4Z4TJ / VA6TJ