The DRTE Computer (page 2)
The P-N-P-N Flip-Flop
The cornerstone of the DRTE computer was the P-N-P-N bistable switch invented by Norman Moody. The first published reference to the use of a P-N-P-N trigger circuit in computer design occurred in 1956 .In that year, Moody delivered a lecture on "A P-N-P-N Bistable Element Suitable For Digital Computers" at the Canadian I.R.E. convention. The bistable switch, flip-flop as it is commonly called today, was the fundamental building block of the computer's binary arithmetic and internal memory. This P-N-P-N trigger circuit was built around a complementary pair of N-P-N and P-N-P transistors. The basic feedback loop in the circuit is given in Fig.1.
This was a major departure from the symmetrical Eccles-Jordan type flip-flop designs that were so commonly used in those days. This produced a flip-flop, see Fig. 2, with the following characteristics
< 0.2 microseconds
|Output Fall Time when driving 5 similar stages||
|Output Rise Time when driving 5 similar stages||
|Output Load Current||
> 30 milliamperes
< 20 ohms
< 1.5 microseconds
According to Florida the above P-N-P-N trigger characteristics gave "the designer great freedom, since at least five similar stages can be driven while still retaining good fall times, and at least thirty one-milliamp AND gates can be controlled by a single trigger circuit" (Florida,1959:2). (Table 1.)
In 1919, in the journal "Radio Review" W.H. Eccles and F.W. Jordan first revealed their trigger circuit. This trigger circuit was the first electronic digital circuit known to have been published (. Like a multivibrator, the Eccles-Jordan flip-flop is a two-stage amplifier with its output coupled to its input. Unlike a multivibrator, the two stages are directly coupled instead of being RC-coupled. The flip-flop consisted of two triode valves with each anode connected through a load resistor to a positive supply potential. "Each anode was also connected through a voltage divider to the grid of the opposite tube so that when one tube was cut off the other tube was maintained conducting, and the conducting condition of the other tube held the first tube in the cut-off condition" (Richards,1967:3) (see Fig. 3). The circuit in Fig. 3 exhibits a symmetry in its configuration. The two stages are identical.This flip-flop has only two states (first tube on and second tube off, or vice versa). Once the circuit has assumed one state, it stays that way until an external trigger pulse initiates a change over. A transistor equivalent of the circuit shown in Fig. 3 is shown in Fig. 4. The circuit in Fig. 4 exploits the transistor as merely a solid state valve. The triode is replaced by an npn junction transistor. The solid state version of the Eccles-Jordan flip-flop exhibits the same symmetry. )
As was mentioned earlier the problem with viewing transistor circuits as extensions of valve circuits is that while all valve circuit configurations have transistor equivalents not all transistor circuit configurations have valve equivalents. While there is only one kind of triode there are two kinds of junction transistors: npn and pnp. It is this difference that Moody exploited in his design of the P-N-P-N flip-flop.
"I had developed a trigger circuit which at that time was unique for it featured a complementary pair of transistors. All the trigger circuits yoqu saw were made up of similar pairs. They were copies of vacuum tube ones. In my trigger circuit, for the first time, you had a pnp and npn whose states were off for a short circuit. This proved to be a very interesting basic circuit on which a great deal of work was done..." (Moody,l985:6).
"There wasn't anything like it anywhere else. The basic trigger circuit was fairly different from anybody else's. It had many properties which they're almost striving for now in CMOS circuitry. It was not sensitive; it wasn't fired by little interference pulses" (Moody,1985:15).
Moody's interest in trigger circuits goes back to the war years. While working on a radar problem Moody came up with a novel mono-stable trigger circuit. This device so impressed F.C. Williams that he asked Moody to come work with him. Moody's interest in faster and more reliable trigger circuits stemmed from his work on the digital counting circuits used in measuring the flux readings of radioactive particle emissions ()
It is interesting to note that the early work of Mauchly, Atanasoff and others on flip-flops was also based on the digital counting circuits used in Geiger counter instrumentation technology. "In the late 1930's a few scattered references began to appear on the use of flip-flops in counter circuits, where the application in most cases was the counting of pulses from a Geiger-Muller tube..." (Richards,l967:4). Mauchly recalled:
"In fact binary counters and vacuum tubes were already known to me; they existed in cosmic-ray labs (like his father's)... So I thought, see, if they could count at something like a million per second with vacuum tubes, why it's sought of silly to use these punch card machines, which can only do maybe a hundred cards a minute and don't seem to do much at that." (Shurkin,1985:91). There was one important problem associated with the P-N-P-N Bistable circuit. The circuit had a basic asymmetry between the current drawn in the "0" state and the current drawn in the "1" state. Recalling the circuits design Moody concluded:
"In hindsight one can see that it was wrong. A row of zeros meant no current was being drawn and a row of ones meant the maximum current was being drawn. So the current from the supply lines would vary randomly with the type of number existing and that was a very difficult problem. It is much better to have a current that never varies whatever happens because the transistors in those days drew quite a lot of current. We weren't into microelectronics; we were thinking milliampere currents and not microamperes" (Moody,1985:6).