|
Messrs. Brattain, Shockley,
and Bardeen, are credited with inventing the first working transistor per their
announcement in December of 1947. This article provides what was for many people
an introduction to the operational specifics of transistors. Although its crude
point contact (aka "cat's whisker") emitter and collector arrangement against the doped
germanium base element was still in a configuration that did not yet represent a
significant challenge to the domain of vacuum tubes, people of vision recognized
its superior potential. As with most technologies, progress occurred quickly and
within a year the first commercial transistor was on the market. C.E. Atkins, a
frequent author in Radio & Television News magazine, related this story
of his own attempts to develop a transistor and offers the opinion that transistor
might indeed someday replace tubes as amplifiers, oscillators, and frequency converters.
A Crystal That Amplifies
 Details of experiments with a double-contact
germanium type crystal having transconductance.
By C. E. Atkins
Tung-Sol Lamp Works, Inc.
Bell Telephone Laboratories' recent announcement of a new electronic component
- the "Transistor" has started many "back of the shop" experiments. Like any other
new design, complete technical details on the product are withheld for a time. In
view of this we believe it to be appropriate to publish this report of an independent
experiment that was, however, suggested by the announcement of the "Transistor"
- Editor.
Rumors of a recent sensational development in crystal transducers led the writer
to explore a field he had not touched since the early 1920's. Like thousands of
other boys, he built the usual crystal receivers and tinkered endlessly with "cat-whiskers"
and coils. The very words "silicon" and' "galena" evoke nostalgia for the halcyon
days now, alas, so rapidly receding into a dim and remote era. Even in that day
the electron tube had largely supplanted the crystal detector, once the mainstay
of wireless reception. In addition to its role as a detector, amplification and
oscillation were also attempted without much success. Radio News in 1923 carried
articles on the Zincite crystal, wherein feeble oscillations were developed because
of a small negative resistance exhibited under certain rather critical conditions.
This was elaborated on in further articles in September and October, 1924. This
was referred to as the "Crystodyne Principle," a term copyrighted by Radio News.
Nothing much came of these developments and with the increasing availability
and lower cost of vacuum tubes, crystal devices of this kind were largely relegated
to the historical museum. However, during the second World War, vast strides were
made in the development of fixed crystal detectors for use in radar where at ultra-high
frequencies they were, in many instances, superior to electron tubes.
Spurred on by the fragmentary press releases and general trade gossip regarding
a three terminal crystal transducer using a germanium crystal, the writer undertook
the fabrication of such a device himself. A 1N34 crystal diode was procured and
used as the source of a ready mounted germanium crystal. By holding the 1N34 with
a pair of long-nosed pliers gripping the cathode or negative lead (which is clearly
marked) the crystal may be exposed by chopping the ceramic adjacent the metal cap
at the negative end with side-cutters. The ceramic will shatter, and there is a
brass stud inside against which the cutting force expends itself. The germanium
crystal is mounted on the stud and, with ordinary care and good fortune, it is possible
to strip the ceramic away without impairing the crystal.
Fig. 1 - (A) Voltage-current characteristics of a single
cat-whisker for two different positions (a and b) on a germanium crystal. (B) Detail
of (A) in the region of the elbow to show high conductance characteristic in the
forward direction.
The writer's attempts to utilize the cat-whisker used with the 1N34 were unsuccessful,
so this part of the assembly was discarded. Workable cat-whiskers can be made from
the heater wire of almost any 150 ma. radio tube. The tubes can be defective for
almost any cause - even heater failure, if in the right place. The average radio
shop generally has many of these at hand. The writer used type 35W4 because several
defective ones were currently available. By breaking the glass bulb and clipping
the support leads to all elements except the heater, it is possible in most cases
to slide the heater out of the cathode sleeve. The structure can be further disassembled
by gripping one of the base pins with a pair of long-nosed pliers while all the
glass around the pin is chipped away with side-cutters. One then has a base pin
and lead welded to a segment of coated heater wire. It is necessary to leave the
ceramic coating on the heater wire in order to give it sufficient rigidity. It may
be trimmed to suitable length with a small pair of sharp scissors and a short tip
(say one-sixteenth or one-thirty-second of an inch) of bare wire can be exposed
by gently and adroitly crushing the ceramic coating with a small pair of tweezers.
It is advisable to prepare several of these cat-whiskers, inasmuch as they are quite
fragile and several may be needed before a workable device is obtained.
As the photograph shows, the experimental assembly is mounted on a block of wood
approximately 5 1/2" x 3 1/2" x 3/4". The lead from the germanium crystal mounting
is saddled onto a brass nut in such a way that this assembly will turn concentrically
about a 6-32 machine screw. The machine screw, in turn, is bolted to a soldering
lug bent at right angles and soldered to a piece of No. 16 tinned bus wire. As can
be seen from the photograph, the wire is held by a Fahnestock clip mounted on the
wood block with a wood screw and spacer. This crude but effective arrangement facilitates
adjustment of the crystal transducer.
Two cat-whiskers are brought to bear on the crystal surface. Clips, extracted
from a wafer type miniature socket by cutting the bakelite away, are soldered to
short lengths of No. 16 tinned bus wire. These wires slide into binding posts mounted
on brackets by means of which electrical connections and a measure of adjustment
are obtained. Since the cat-whiskers were prepared attached to the radio tube pins,
they can be readily fitted into the socket clips. After suitable forming with tweezers
or small pliers and the exercise of considerable cautious diligence, two cat-whiskers
may be brought to bear upon the surface of the crystal. They should engage the crystal
with a light pressure and should be as close together as feasible without contacting
each other (the separation is only a few thousandths of an inch at most). A magnifying
glass is essential and considerable juggling artistry must be exercised before good
characteristics are obtained.
Fig. 1 shows a voltage current characteristic of a single cat-whisker (either
one of them). It has the shape typical of a barrier type rectifier. The curves a
and b are from data for different cat-whisker settings. Fig. 1B is a detail
of Fig. 1A in the region of the elbow in order to show with greater clarity
the high conductance characteristic in the forward direction. This data indicates
that in the forward direction the impedance is approximately 100 ohms, while in
the backward direction it is approximately 3000 ohms. These curves are typical of
many taken. In the backward direction voltage breakdown is likely to occur at above
50v, while in the forward direction burnout takes place in the neighborhood of 60
or 70 ma.
Fig. 2 - Two curves relating the transfer or control parameters
of the device.
Fig. 2 shows two curves relating the transfer or control parameters of the
device. With a back voltage of approximately -50v. on one cat-whisker, the control
effect on current to this cat-whisker of the application of small forward potentials
to the other cat-whisker is shown. Curve a, indicating a transconductance of 15,000,
was about the best the writer was able to do. Characteristics of the order shown
by curve b are more readily obtainable. As Fig. 1 indicates, small forward
potentials on the cat-whisker result in rather substantial forward current so that
unlike typical vacuum tube operation the input electrode does require appreciable
amounts of power for control purposes. However, since the output electrode potential
is much higher than the control electrode potential, a favorable ratio of controlled
power to controlling power may result even in cases where the input current is actually
greater than the output current! The writer's nomenclature designates the output
circuit as the "B" circuit, while that associated with the control cat-whisker is
called "C" circuit. This is in rough accord with ancient usage.
Fig. 3A is a schematic diagram of a test circuit for measurements of amplification
with the three terminal crystal and for obtaining the data of the foregoing characteristic
curves. Since the control electrode impedance is only about 100 ohms, a stepdown
transformer is needed to couple a signal into the device. The d.c. resistance of
the transformer secondary is only a few ohms and its effect on input circuit current
is negligible. The output load resistor RL can be shorted out while data
for the static characteristics is taken, although it is generally advisable to have
a hundred ohms or so at this point in order to prevent runaway at high current levels.
In fact, the writer's data set forth in Figs. 1 and 2 is in error to the extent
of the IR drop in a protective resistor in series with the "B" circuit.
The best conditions the writer has so far managed resulted in a change of 15
ma. Ip at -50v. Eb with a control potential change of +1v.
at 15 ma. This is a power ratio of 50 to 1. Unfortunately this condition was very
unstable. More reliable and more readily obtainable adjustments gave power ratios
of approximately 15 to 1.
As an amplifier, a load of 500 ohms seemed to give the best results even though
the curves apparently call for 1500 ohms or more. A typical favorable adjustment
of the cat-whiskers gave a maximum power output of 200 mw. at Eb -40v.,
Ib 22 ma. (B power .88 watts), a C bias of + 2.0v. and rectified Ic
of 15 ma. ("C" circuit standby power 30 mw.) a likely a.c. input power of approximately
22 mw., giving a power gain of 10 to 1. The small positive bias on the input cat-whisker
is necessary for relative linearity in output current. The waveform looked reasonably
good on an oscilloscope and the response was constant from 20 cycles to 30,000 cycles.
Substantial voltage amplification should be possible, if the resistor RL
is replaced by a properly designed' step-up transformer.
Fig. 3 - (A). Schematic diagram of test circuit used for
measurements of amplification with the three-terminal crystal and for obtaining
data of curves of Fig. 2. (B) Circuit diagram of an oscillator using double
cat-whisker germanium crystal.
Fig. 3B is a circuit diagram of an oscillator using the double cat-whisker
germanium crystal. A 15v. r.m.s. signal was developed across either of the tank
circuits at frequencies from 20 to 100 kc. depending upon the L-C ratio. Eb
was -35v., Ib 4 ma., Ic 1.5 ma. and Ec, probably
around -1.5v. due to the d.c. resistance of the transformer winding. These figures
are considerably different than the amplifier case. The oscillatory condition apparently
has a marked influence on the behavior of the crystal.
The writer has not as yet succeeded in operating the double contact crystal at
high frequencies but expects to do so at a later date.
Why does the device function as it does? An exact answer is doubtless couched
in terms of mathematical physics. Any purely verbal description is likely to be
a generalized approximation at best. Barrier rectifier characteristics stem from
the fact that when dissimilar metals are extremely close together without actually
touching (100 atomic diameters or so) electrons from the more electropositive metal,
i.e., the one with the lowest work function, spill over into the other metal. That
is to say, they actually jump the barrier. Enough of them will do this and equilibrium
is established when a back-potential equal to the difference in work functions of
the two metals is reached. If a positive external potential is applied to absorb
the electrons across the barrier the process continues and a current can flow in
the "forward" direction. A potential in the reverse direction, however, results
in but little current since electrons merely pile up at the barrier unless, of course,
the potential difference is great enough for breakdown. In the case of the germanium
crystal and the tungsten contact a suitable barrier is supplied, as it were, by
the atomic structure of the germanium crystal because it is a certain type of semi-conductor.
As Fig. 1 indicates, electrons pass through the junction more readily in one
direction than the other. If a back voltage is applied to such a germanium-tungsten
junction and a second tungsten cat-whisker is positioned very close to the first
one with a small forward potential applied to it the equilibrium conditions are
upset and it is possible for electrons to flow from the first cat-whisker in spite
of the backward potential.
What good is it? Prophesy is always hazardous, but it would seem that devices
of this kind might certainly supplant vacuum tubes as oscillators, amplifiers, and
frequency converters in many applications. While the writer's working model is admittedly
crude - one could not sneeze in the same room without substantially altering its
characteristics - one can readily see how the necessary mechanical and electrical
refinements might be worked out as in all likelihood is already the case. New components
and new circuit techniques must of course be developed to use them in radio, television,
and industrial electronics.
Posted February 1, 2023
(updated from original
post on 10/26/2016)
|
