First-Hand:A Co-op Student Before Graduation
Contributed by: Dean J Chapman, IEEE Life Senior Member
While attending Rensselaer Polytechnic Institute in the late 1950s (BSEE '61), I had the good fortune to be hired as a co-op student by General Electric. This was an excellent program and provided many benefits to the would-be engineer as well as being an excellent recruiting tool for GE. Incidentally, GE's "Advanced Courses in Engineering" or A-B-C Course for young graduates was also an outstanding program that provided mutual benefits to both the employee and employer. I was fortunate to have taken part in that program as well.
In reading some of the "Tales from the Vault" stories, I am reminded of some of my experiences while a co-op with GE. My first assignment was with GE's Light Military Electronics department in Utica. This was the era of the B-70 project, and GE was subcontracted to IBM in Owego, New York for some bench-test equipment for a large coherent radar system that was slated for the ill-fated B-70. My principal responsibility while there was to keep track of the interconnections between the myriad instruments and devices that were mounted in rows of bolted-together 19-in racks and their central power supplies. There were several racks of Lambda dc power supplies whose outputs needed to be distributed to the various custom-built instruments. Each box had a different military style multiple pin connector, and those had to be cataloged and ordered. So much for the romance of designing exotic circuits.
One experience stands out in my mind. The prime equipment apparently had large traveling wave tube (TWT) amplifiers with electromagnetic solenoid focusing coils (I never did see any of the prime equipment). We had to provide a current-regulated source of 28 V dc at up to 35 amps for bench-testing the TWTs. One of the engineers had designed a transistorized regulator using the famous 2N173 power transistor. There was a bank of seven of them mounted to a copper heat sink through which cold water ran. It was my job to set up and test these circuits. I was forced to relocate to the lab's sink so that I could hook up the water hoses. The fact that I was dealing with currents and voltages that would pass for an electric welder was brought home to me one day when I dropped the probe of my trusty Simpson 260 VOM, and it brushed against the heat sink (at collector potential) and the aluminum chassis (at ground potential). There was a momentary Pffft, and I saw a drop of something splatter on the floor. I thought at first it was solder until I noticed the missing chunk of aluminum chassis and realized the drop was molten aluminum.
My remaining two assignments were with the GE broadcast transmitter group in Syracuse. In addition to designing and manufacturing commercial radio and television transmitters, this group had two very interesting and challenging contracts. The first was called "Project Heat" and involved the design and construction of several 250 KW power oscillators that were going to be used to heat the skin of military aircraft that were undergoing tests. The heating would help to simulate supersonic flight. These oscillators were required to tune continuously from about 200 kHz to about 2 MHz. They were pulsed on and off at a low audio rate, and the duty cycle was varied to control the heating power. They would have made great jammers for the AM broadcast band but were to be used in a heavily shielded building at Wright-Patterson AFB. I never heard how the project turned out or how effective the shielding might have been. My job was to build a scale model of the oscillator to test tank coil geometries for the sliding tuner mechanism. I used 1/4-in copper tubing for the tank coil and parallel 6J6 triodes to simulate the scaled-down characteristics of the triode that would be used in the final model. It was a fun project and tested our ingenuity.
When I came back for my next assignment, I worked on another special order project, a group of six 250-KW short wave transmitters for Voice of America. This design was a 250-KW conventional AM transmitter with plate modulation that required a 125-KW audio amplifier. The modulator circuitry was designed with direct coupling to reduce phase distortion to a minimum. The transmitter was required to operate at any frequency between 3-30 MHz. In addition, a frequency change had to be accomplished within a short time. This meant that the settings of all the tuned circuits, from the exciter up through the final stage, had to be recorded in a table for each operating frequency.
When the transmitter was shut down, the operator changed all the settings according to the table and turned the transmitter back on. It was supposed to come up to within 10% of full output and then live tweaking could take place after the transmitter was operating. This sounds easy except that the final amplifier contained two power triodes about the size of small golf bags sitting in a tub of boiling water with a tank coil about 24 inches in diameter wound from of one and one quarter inch copper tubing and several large tunable Jennings vacuum capacitors, each about the size of a gallon of milk. The capacitors were ganged together with a chain drive to tune them. To be able to reproduce settings, we counted the rotation of the gears with a microswitch riding on the teeth of the gears and connected to an electromechanical up-down counter. We spent more time trying to eliminate false and dropped counts in this scheme than we did designing the final amplifier. A flexible shaft coupled to a clock-type set of pointers would have worked better, but as electrical engineers, we were too stubborn to take a mechanical approach.
Of the many interesting aspects of this monster transmitter, the most interesting was the need to measure the power output. There were no commercially available dummy loads for 250-KW transmitters. The GE engineers devised a 4 x 4 ft stainless steel tank into which they put several resistors comprised of zigzag strips of heavy duty cage screening connected on one end to the tank wall and in the center to a cylindrical copper structure with a spiral slot and shorting bars for tuning. The ubiquitous Jennings capacitors completed the matching circuit since the line impedance from the transmitter was 50 ohms, and the impedance of the screen resistors was very low. Once we showed that we could obtain an acceptable VSWR with this design, the next job was to be able to calorimetrically measure the power dissipated in the resistors. The resistors operated under flowing water, so we measured the input and output temperature and the flow rate of the water flowing through the box. A little slide rule work produced the power dissipated. To prevent loss of heat to the air, we covered the box and wrapped the whole thing in fiberglass insulation. The steam would rise from the box when the transmitter was running. We resisted the temptation to steam clams in it but it probably would have worked well. Ultimately, we were able to show that the transmitter final stage was operating at an acceptable efficiency. The input to the final amplifier was 12 KV at 25 amps. The grid drive was 8 amps. What a great Ham transmitter this would have made. Incidentally, the power supply used strings of solid-state diodes, as did GE's standard 50-KW broadcast transmitter at that time.
When I left GE to go back to school, the next phase of the project was just beginning. As I said, the output of the final stage was a coaxial line comprised of a large copper pipe center conductor and a square outer “shield” of sheet metal that resembled an air duct. As I remember, the outer conductor was about 12 to 14 inches on a side. The typical short wave antenna arrays used for overseas broadcast were balanced 300 ohm geometries such as Rhombics. The GE engineers designed a huge broadband Balun about 75 feet long that contained an exponentially tapered set of conductors inside a large sheet metal outer conductor. As they adjusted it to provide a good match over the decade of bandwidth, it was necessary to tweak the spacing of the conductors. I was working in an adjacent bay to where this was taking place and it sounded like an automobile body shop as they hammered and tapped on the lines to move their spacings. The “clam steamer” dummy load design was modified to a balanced configuration for final test of the Balun under full power. I never got to see that one in operation.
As I mentioned previously, the venerable mercury vapor rectifier tubes used in large transmitters up to this point had been replaced in this design by long strings of silicone diodes, each mounted on a square copper heat sink about 4 inches on a side. These in turn were strung together on an insulated rod and mounted in the power supply cubicle. In order to ensure that the inverse peak voltage was uniformly distributed over the series strings, each diode was bridged by a resistor which resulted in a long voltage divider. As I remember, the diodes of that time were limited to PIV ratings of several hundred volts. The diode stacks were arranged in a 3 phase bridge configuration and were fed from utility-type transformers located in a cage behind the transmitter.
A fast acting circuit breaker comprised of a set of vacuum switches driven by thyratrons and large capacitors was triggered by sensitive relays that sensed over-current conditions and protected the delicate solid state diodes from damage. This scheme was put to use many times in the early testing of the transmitter when arcs occurred in the final amplifier structure. Because of the large size of the tuned structures necessitated by the massive currents and voltages present, unloaded parasitic oscillations would occur at much higher frequencies on occasion and, since they were unloaded, the voltages would become quite large. The result would be an arc to the supporting structure and tripping of the power supply would ensue. We would then go into the transmitter enclosure armed with a grounding stick and look for fresh burn marks, denoting the location of the parasitic oscillation. We would then have to try to figure out how to dampen it.
One particularly violent arc occurred in the transformer cage when one of the cables leading to the rectifier cabinet got too close to a transformer tank and arced over. We had become accustomed to snaps, cracks, and hisses from the transmitter proper but a gun shot-like bang from the transformer cage was a new and startling experience. The transformer cage was fed from a three phase primary (4160 volt) entrance. There was a large manual disconnect switch with a mechanical lock on the lever. There was only one key that operated that lock and it was the same key that opened the cage. Thus, it was not possible to turn on the switch when the door to the cage was unlocked. Many other safety interlocks were incorporated including switches on all the doors to the cubicles. The grounding sticks present in each cubicle had to be in their holders with the doors closed before the control circuit could be energized to start the transmitter. Leaving a grounding stick hanging on a bus bar had caused loss of several banks of diodes at one point and this modification was added forthwith.
There were many other anecdotes related to the massive size of this transmitter. The six transmitters were slated for installation at the VOA facility in North Carolina. After this article had been published in the Life Members Newsletter (April 2007), I received a letter from Bruce Hunter of Mill Valley, CA, who managed that facility for many years. I subsequently met him and he showed my pictures of the installation. He had many good words to say about the quality of those transmitters and it was gratifying to hear that. The transmitters were being built in 1960 and some of them are still apparently operating.
I have had many varied experiences in my career as an engineer but none were quite as exciting and unique as those that I experienced as a co-op before I had even graduated.
<rating comment="false"> Well Written? 1 (No) 2 3 4 5 (Yes) </rating> <rating comment="false"> Informative? 1 (No) 2 3 4 5 (Yes) </rating> <rating comment="false"> Accurate? 1 (No) 2 3 4 5 (Yes) </rating>