An Engineer in Wonderland – Inventing the 555 and the 741
I was trawling the archive of Electronics Weekly, and came across an article from November 16th 1996.
It describes the invention of two famous analogue chips – the 741 and the 555.
Just in case you are interested, as it doesn’t seem to be available from the website, I have included the full text below.
Like the electronic equivalent of the Mini and 2CV cars the 741 and 555 are two of the most famous chips ever made. Both the cars and the chips are over 25 years old and although there are newer faster models all are still in production.
The 741 is an operational amplifier and was the bed rock of analogue signal processing for over a decade. The 555 is a timer that is so versatile it almost defies definition.
The original 741 and 555 have spawned many copies and variants from many semiconductor companies but have you ever wondered who first designed them?
Well…The 741 started life at Fairchild Semiconductors as the uA741 in 1967.
It was designed by Dave Fullagar who had emigrated to the US from England in 1965 after two years working for Ferranti in Edinburgh.
The 555 was designed a few years later by Hans Camenzind who had been working for Signetics (Now Philips).
Since the 1970′s both Fullagar and Camenzind have continued to work in the electronics industry.
Fullagar co-founded Maxim in 1983 as VP of Engineering and is still there although not directly connected with design work. He said: “I’m semi-retired now which at Maxim means a 40 hour week!”
Camenzind started his own company Interdesign after a couple more contracts one of which was for the Intersil 8038 waveform generator.
He sold Interdesign to Ferranti in 1978 has operated as Array Design ever since. He said: “I design primarily custom ICs and a complement of standard ICs for Zetex Oldham UK.
The 555 was my tenth design and I am now up to number 135.”
Fullagar on the birth of the 741
Marketing’s original idea was to make another version of the uA709 op-amp but with every parameter improved by a factor of two. I started work on this but mid-way through had an idea for another design.
The 709 was not that easy to work with as it suffered from latch-up was difficult to stabilise and had other quirks.
The idea I had was for something that would be much simpler to use. The idea became the 741.
The ‘improved 709′ was passed on to someone else and eventually emerged as the uA725.
The 741 is a two stage internally compensated design.
Bob Widlar was the first to design a two stage op-amp (with the LM101) and I thought up the idea of internal compensation to avoid the instability seen in the 709.
The other new feature of the 741 was the input stage. The difficulties of this section had been addressed in the LM101 but Widlar’s solution didn’t seem quite right to me.
It was playing on my mind when suddenly on the way to a skiing trip the solution came into my mind.”
He started work on the uA741 in 1966 and it was announced in May the following year.
Fullagar didn’t stay at Fairchild for long: “Shortly after the uA741 was announced I left Fairchild to join Intersil where I designed the majority of their early analog products. Most popular of these was the ICL80079 a monolithic FET input op-amp.”
Although the 741 is Fullagar’s most famous design, the one he best remembers was for a Canon: “My favourite product was a camera circuit developed for Canon Camera in Japan. It took the log of three inputs (aperture film speed and light intensity) summed the logs took the antilog and derived a shutter speed output. The whole circuit was temperature compensated and used a total of about ten transistors.”
Camenzind on the birth of the 555
In 1967 designing repeatable integrated tuned circuits was impossible. I was interviewed by Signetics that year and proposed that they let me try to designed one using a phase-locked loop.
I found the PLL in a 1935 Proceedings of the IRE and had seen that it wouldn’t need on-chip repeatability since it can set itself in step with an external frequency.
So with Alan Grebene (later with Exar and Micro Linear) and Graham Righby we set out to design the circuit blocks for a phase-locked loop.
Righby came up with a clever voltage-controlled oscillator which could operate at high speed and Grebene engineered it.
That product the first IC PLL was the NE560. I didn’t quite like the outcome of the work so I decided to find a better VCO on my own.
I used a voltage-to-current converter with the current set by an external resistor. This current charged and discharged an external capacitor.
Control of current direction came from two comparators.
Both the current source and comparator thresholds were dependent on the supply voltage but the two effects compensated making output frequency supply-independent.
Using this principle Jack Mattis and I designed the 565 PLL and the 566 waveform generator both of which are available which the 560 isn’t!
In 1970 I took a 4-month break to write a book. Instead of returning as an employee I asked the company for a 1-year contract as a consultant to use the VCO principles to design a timer a circuit which could oscillate or run for just one cycle.
Signetics’ engineering department didn’t think much of the idea but fortunately Art Fury the marketing man did.
He had the gut feel that it would sell and broke with the the tradition of introducing a new IC at a high price and then letting the price drop. He set the price of the 555 at 75 cents right from the start.
I got the contract and completed the circuit-design portion in six months.
I had already written a report on the design when I realised that I had been incredibly stupid. I had assumed that the compensation principle would only work with the linear voltage-to-current converter and would fall apart if an exponentially charging RC time constant was used.
For some reason I decided to check that out and found that my assumption was not true at all.
An RC time constant worked exactly the same way.
This greatly simplified my design and reduced the number of pins from 14 to 8. Ironically an employee at Signetics quit his job shortly after I delivered the report and took the design to a start-up company.
They beat Signetics to the market – with the wrong design.
Their circuit lasted exactly three months.”
Other versions of the 555 have been seen since the original. Camenzind said: “In 25 years there has never been an improvement on the venerable old 555 apart from a CMOS version by Dave Bingham at Intersil. I have just finished redesigning the circuit from ground up. The new version [designed for Zetex] is much improved. It works down to 1 Volt retains the speed has better accuracy but only draws one-tenth the current.” And added wryly: “Experience helps but so do smaller device geometries and computer-aided design!”
‘The father of analogue design’
When asked what their favourite integrated circuits were both Fullagar and Camenzind named designs by Bob Widlar.
Camenzind said: “The design that has impressed me the most over the years is the National LM10 by Bob Widlar, who else. To me that chip is like a symphony with all the components playing in perfect harmony. We probably won’t see anything like that again: it took Widlar five years to design it.”
The bandgap reference also got Widlar to the top of the list of the 10 most influential analogue designs voted on at this year’s ISSCC conference.
The top five were:
1. Bandgap reference/ regulator
2 Differential pair
3 Translinear circuits
4. Current mirror/source
5. Switch capacitor circuits.
The op-amp as we know it was also a Widlar idea.
Fullagar said: “In the very early days Widlar did some designs of his own back then presented them to the industry.
One of these was the ‘differential in single-ended out’ op-amp. This was at the time when some were thinking of differential outputs on operational amplifiers.
Widlar was also first with the three terminal voltage regulator the two stage op-amp and has been credited with the invention of the current mirror.