Two problems plague power amps installed on military aircraft
By Contributing Writer William J. Garner, Engineering Consultant
A while ago I was employed by a small firm as an engineering consultant to work on power amplifiers that were to go onto military aircraft. A certain quantity of power amplifies were designed and delivered to the Air Force to be installed on a jet fighters and the application was to apply about 20 watts in the UHF band to an antenna.
Since many electronic devices are used on aircrafts, this unit was selected to operate by turning on or off the power supply voltage or Vcc to save power. After awhile they were returned back to the firm with the output bipolar power transistor inoperative. I was brought in to determine the failure and fix the problem.
Removing the bottom cover, I saw that the output power device was a bipolar silicon stud mount transistor with 4 leads fanning out and soldered to the PC board. Powering up the unit with the required 35 volts DC, I measured the collector voltage to be 35 volts and the emitter voltage to be zero volts, which means no collector current.
Obtaining a schematic from the design engineer, I could see no obvious reason for the unit to fail. Next I decided to remove the ceramic cover on the device to see if any visible damage was evident on the transistor surface. Using a slip-joint plier, I popped off the cover, and with the aid of a stereo-microscope I noticed that the surface of the emitter and base fingers were black which indicated an over-voltage.
Next, I looked at the schematic and the device ratings and saw that the device Vbe was rated at 30 volts; while when operating in its Class A mode, the Vbe was 25 volts. So one would think this device would be in the safe operating mode with the Vbe rating of 30 volts.
Then I looked at the schematic and noticed that the bypass capacitor value on the Vcc supply line was .01 ufd, while the the bypass capacitor on the base resistor divider was 1.0 ufd. Using Pspice and a triggered scope, I determined that when the Vcc of 35 volts was applied, the voltage on the collector was 35 volts, since the voltage on the base does not reach the required level because of the 1.0 ufd capacitor taking longer to charge. That meant that for some few milliseconds, the Vbe across the device was 35 volts, hence the breakdown event occurred.
Fixing the problem was easy–simply reduce the base bypass capacitor to 0.01 ufd, and increase the Vcc bypass capacitor to 0.1 ufd.
A different power amplifier was also returned because of a gradual failure of the output stage power delivery. This unit also used stud mount bipolar silicon device with four fanned-out leads, but the device was mounted such that the leads were below the top surface of the PC board and than appeared to be bend up on over to the top of the PC board then soldered.
Again I popped-off the ceramic cover and noticed the many wire bond leads from both emitters were either broken or stretched almost flat. Some appeared to be vaporized with few remaining to carry the required IC current.
I ask the production engineer how this unit was assembled, and he told me that the components and the transistors were all soldered on the PC board at the same time, then the board with the transistors was installed on the chassis, and the transistors were then torqued down using the nuts supplied.
What manufacturing did not realize as that the location where the transistor was to be installed had the wrong vertical distance for mounting. Hence, when torqued down, the device went so far below the top board surface that the four leads stretched away from the device and caused the internal wire bonds to either break or stretch so flat as to degrade the current distribution. I called the device manufacturer who told me that if just one wire bond gets broken, the current distribution would gradually degrade and finally open-circuit.
The fix for this was easy: Make a copper washer to bring up the device leads level with the PC top layer, and install the devices AFTER the PC board is installed, then solder the leads to the PC board pads.
Contributing Writer William J Garner is an engineering consultant.
Tweet This
4 
