|
|
Oral History – Hannon Yourke
Beginning Work on Current Steering Transistor
Circuits
|
|
The initial motivation for this
work was to come up with non-saturating techniques for doing logic. What was in place at the time was all
based on saturated circuits, which take an awful long time to come out of
saturation. So, I worked on this
idea and came up with what eventually became known as emitter coupled logic
(ECL). The most important parameters for these circuits was switching speed
and power dissipation. Of course,
the devices were quire different than now.
They had very high collector resistance and, to keep them out of
saturation and in a region of high bandwidth, you had to work with higher
voltages. I think that bipolar
transistors (as we had then) are not used at all now. One of the types we
used was the Philco surface barrier transistor, which was about the best
device you could get your hands on at that time. These were very uniform – I guess they were etched to
achieve very specific characteristics.
The one problem with these was that the devices would fail if the
well specified collector breakdown voltage was even slightly and
transiently exceeded – otherwise, these were very reliable.
Using
Different Transistor Types
There was also a
semiconductor group at IBM, which came up with the drift transistor. Internally, this transistor was used for
our work and also this was the type which was delivered in our machines -
there was a commitment (to use these transistors) that was made for the
Stretch program, and my
|
|
|
Oral History – Hannon Yourke
|
|
invention (the current
steering circuit) was used in the Stretch computer. This program actually came out much
later than the first machines to use those circuits. The first machine to use the circuits
was the IBM 7090. In parallel with
the Stretch program they started development of a lower performance
machines, also using current steering circuits. Stretch, I believe, came out in 1961, and the 7090 in
1959.
Current
Steering Circuit
has Broad Applicability
It was a very general
technique. The patent, as I wrote
it, really covered all types of possible alternatives. Even though the Stretch machine was
built with a combination of PNP and NPN transistors (so you didn’t have to
voltage translate), my patent included a voltage translation. The paper I wrote (“Millimicrosecond
Transistor Current Switching Circuits”) in 1957 for the IRE shows an
assortment of circuits, including triggers, exclusive-or circuits, line
drivers, and others. In fact, every
circuit in the Stretch circuit manual was designed using this technique.
So, you can see that the patent was very basic and also very important.
As far as mainframe high speed
and high performance logic, this circuit remained the basis until the time
I left the company - I retired in 1985.
Go
To Yourke Oral History, Page 3
|
|