A 136kHz Experimental 3kW Class D Amplifier 
This is a view of my 136kHz kit. The unit under the morse key is the transmitter/receive pre-amp, and the unit to the left with the scope probe attached is the experimental 3kW Class D amplifier. To the right of the multimeter is a bank of psu smoothing capacitors and behind them are four 1KVA toroidal transformers.
The transmitter is based upon a 7490 decade divider driven at 1.36MHz from the Marconi 2945 test set on the bench above which gives a 1Hz frequency setting resolution at 136kHz. Other 74 series devices in the transmitter are used for keying and buffering and produce the drive for a 50 watt push-pull MOSFET output stage.
The large 3dB 50 watt attenuator on top of the 136kHz transmitter is to avoid overdriving the amplifier.
The amplifier powered up showing 7.5A of output line current on an 8A FSD thermocouple meter.
The amplifier uses a total of 8 - IRFP450 MOSFETS in a push-pull arrangement with a 100V supply. An output line current of about 7.5 Amps into 50 ohms can be obtained which represents about 2.8kW. The output winding of the toroid is tapped so that line currents 3.4A, 4.9A, 6A, 6.6A and 7.5A may be obtained. No filtering is used on the output - tests are only into a dummy load, and if used "on air" with a suitable LPF, the usable output would probably be in the order of a couple of kW.
With the gates of the MOSFETs biased at zero volts, drive requirements for the amplifier is 25 watts, which produces a peak gate drive voltage of just over 10 volts. Drive is applied to the gates via a small ferrite toroidal transformer, and using zero bias, a slight "dead zone" is obtained between the time one side switches off and the other side switches on (as happens in commercial switched mode PSUs). This significantly reduces the switching voltage "spike" seen on the MOSFET drains.
I may be tempting fate by saying this - but so far I have not "lost" any MOSFETs in the amplifier.
By accident I once left the amplifier running while I took a phone call and forgot about it. About two hours later I suddenly remembered. I had closed to door to stop the family pet cat from jumping on the workbench and becoming barbequed, and upon opening the door, the room was like a Swedish Sauna from the heat given off by the dummy load! Fortunately my Bird load is rated at 2.5kW but it was very hot.
The IRFP450 MOSFETs are directly mounted on two large pieces of copper bar - each 5" x 8" x 3/8" thick - one for each side of the push-pull circuit. Even at the full power "extended" test these were still comfortably warm to touch.
Note - this amplifier is not suitable for linear operation. Because of the relatively small heatsinks it is essential to apply sufficient drive to ensure that the MOSFETs operate in a switching mode. A low level of drive could cause the devices to operate in a linear region which would almost certainly cause the devices to over dissipate and destroy themselves.
For further details of my 136kHz Trans-Atlantic tests, see my 136kHz pages.
Return to the Amplifiers Page
Page last updated 8 Jan 2002