Whatever happened to IIL?
Mark Robinson
(Integrated Injection Logic I think, usually written I-squared-L)
during my undergraduate days.
From what I remember, the gates were very simple, so you could fit
loads of them on a chip, and you could program the speed/power
tradeoff using a resistor connected to one of the pins.
OK, so what happened to this wonderful logic family (I never used
it)?
Cheers
Mark
[ I have posted to sci.electronics for knowledgeable input, but
follow-ups are to alt.folklore.computers to try and keep noise
out of the sci groups ]
--
______________________________________________________________________
Multimedia Information Systems Lab | I'm a slug and I'm alright
Dept. Of Electrical Engineering | That's 'cos I'm hermaphrodite
University of Manchester | It bothers me not that I have no
Dover St. Manchester. M13 9PL | friends
England | 'Cos I have fun with both my ends
Gary Gendel
>During an idle moment I remembered being taught about IIL devices
>(Integrated Injection Logic I think, usually written I-squared-L)
>during my undergraduate days.
>
>From what I remember, the gates were very simple, so you could fit
>loads of them on a chip, and you could program the speed/power
>tradeoff using a resistor connected to one of the pins.
>
>OK, so what happened to this wonderful logic family (I never used
>it)?
>
>Cheers
>Mark
Ah... IIL, brings back memories. I cut my teeth on designing ICs with IIL.
What you say is true, but there were some other advantages....
1) Being a bipolar process, true bipolar transistors could be placed
with the IIL logic, and sized to meet your performance requirements.
2) Since they were regular structures lined up like skyscrapers
along the injection rail, they could not only be densely packed,
but we could layout an incredible amount of custom logic in a
very short time. We developed a symbolic layout approach using
gridded paper, that allowed dozens of configurations (of, for
example a JK flip-flop) in a matter of minutes, allowing the best
configuration for speed and density to be developed.
The disadvantages..
1) Having a single input, multiple output device was a strange
animal to get used to.
2) It was relatively power hungry.
I feel the main reason that it got pushed aside was the advent of automation tools
that favored standard cell methodology. IIL inherently was a custom application,
it required that collectors were added for each fanout of the gate, you
needed a different cell depending on the number of fanouts for that cell. In
addition, the output delay depended upon the collector's relative position to the
injector and base. Logic simulators weren't designed to deal with this easily,
making design by simulation very difficult, forcing us to breadboard instead.
With today's biCMOS process, many of the unique advantages of IIL have been
duplicated.
Just my 2 cents worth.
---
Gary Gendel
Vice President: Current consulting assignment:
Genashor Corp Mentor Graphics Corporation
9 Piney Woods Drive 15 Independence Boulevard
Belle Mead, NJ 08502 Warren, NJ 07059
phone: (908) 281-0164 phone: (908) 604-0883
fax: (908) 281-9607 email: ga...@warren.mentorg.com
Andy Rabagliati
>During an idle moment I remembered being taught about IIL devices
>(Integrated Injection Logic I think, usually written I-squared-L)
>during my undergraduate days.
>
>it)?
>
Lets see - Sinclairs Black Watch (first cheap LED digital watch)
Any other Sinclair stuff ?
I2L was made (developed?) at Plessey, Caswell. We had a design project
at Brunel University for an Associative Parallel Processor using this
technology. Had nothing but trouble with it.
It had an injection layer - a constant current supply to all devices. It
was said that that you could run the chip just by shining a light on it
- the photoelectric current did the same job.
It was a single input, multiple output logic technology, using wire-AND,
and you thought "current", not "voltage".
The logic gate was multiple-collector transistor, with a pullup resistor
on the base. If any of the collectors connected to its base pulled it
down, the output collectors turned off. Thus it has the basic
(necessary) property of inversion. And horrible speed/fanout problems.
Need it to go faster ? Crank up the supply current.
Less fattening ? Crank it down.
The first chips we got back had an output breakdown voltage of about 1
volt (!) and we had to put common-base transistor amplifiers on the
outputs as level translators.
Cheers, Andy.
--
What Godzilla said to God when his name an...@wizzy.com
wasn't found in the Book of Life. (American Music Club) W.Z.I. Consulting
Bruce Cheney
TI now has I2L using GaAs. They call it Common Emitter
HBT, but it is essentially I2L.
People didn't like I2L because it required either a low
voltage power supply or a resistor from a 5 V power supply.
Those were the days when everything was 5 V TTL stuff.
Nowadays, CMOS is preferred, at least at low speeds, because
of the low power dissipation.
Bruce Cheney
David Thomas
>During an idle moment I remembered being taught about IIL devices
>(Integrated Injection Logic I think, usually written I-squared-L)
>during my undergraduate days.
>From what I remember, the gates were very simple, so you could fit
>loads of them on a chip, and you could program the speed/power
>tradeoff using a resistor connected to one of the pins.
>OK, so what happened to this wonderful logic family (I never used it)?
CMOS killed I2L. At just about any point on the speed/power curve,
CMOS turned out to be superior. The I2L gates suffered from charge
storage due to operating in saturation, so in 5um technology the
best gate delays that could be obtained were in the 10-15ns range
(12 years ago). It was not practical to put Schottky clamps on the
transistors to keep them out of saturation, because they had multiple
collectors--you would have to put a diode on each collector.
I could go on and on. I did a lot of work with I2L many years ago,
including a logic synthesis program that synthesized inverter logic.
I'd be interested in talking to other I2L fans...
--
David Thomas Texas Instruments (da...@wotangate.sc.ti.com) (713)-274-2347