The first circuits that could in some sense be-called 'integrated' were made up from discrete devices soldered together to form a module, in particular, those that were 'potted' in plastic to create a kind of 'meta-component' that could be used in the design and construction of a more complex circuit. Usually these did not bring the benefits of miniaturisation associated with hybrids or monolithics, but they did bring the benefits of modularity and standardisation.

In the world of thermionic valves (or vacuum tubes) there were few attempts at modularisation. One notable exception is in valve/tube digital computers of the 1950's, where the size of the installation dictated that maintainability was crucial. Manufacturers such as IBM packaged valves/tubes together with a few passive components to create modules such as the one shown, a single flip-flop made from a double-triode. The module itself plugs in using a valve/tube base, so replacement was therefore easy, an essential feature in a computer that might suffer many valve/tube failures per day. Machines such as the IBM 704 used even larger 'trays' containing eight or more valves/tubes.

Valve(tube) flip-flip module
packaged transistor circuit

No sooner had the transistor been invented, with its major advantage of reduction of circuit size, than people were thinking about simplifying circuit design and construction by using packaged circuits of various kinds. In November 1952, the Proceedings of the IRE published an article by J.A.Morton entitled "Present Status of Transistor Development" which describes a study of the "feasibility of applying transistors in the form of miniature packaged circuit functions". Seven package types were developed, each comprising a circuit card embedded in transparent plastic, with a tube/valve base at one end serving as a plug connector. They employed very early Bell Labs junction transistor types such as the M1752. The image on the left, taken from the article, shows two examples. The packages had different functions, and were connected together in different ways to build larger circuit blocks such as a binary counter and a shift register. The article suggests some improvements that would be needed to make the packages suitable for commercial use, but this technique does not seem to have been pursued further.

Also in the 1950's a notable modular construction type that was used particularly in the USA was known as "cordwood" because of its resemblance to a stack of cut logs. Normal components were soldered in parallel as closely as possible between two or more circuit cards. The benefit of this approach is high component density, an early attempt at miniaturisation. The disadvantage is that it is a maintenance nightmare. This technology lived on into the early transistor era of the 1960's, and NASA considered it for use in the Apollo Guidance Computer, but famously took the bold decision to use relatively new and untried monolithic integrated circuits, described in my page on monolithics.

cordwood transistor circuit
Philips Series 1

A gap of approximately ten years was to pass before encapsulated circuits were tried again. This time, it was Philips in the Netherlands who in about 1961 developed the "1-Series Circuit Blocks". The 2.IA1 shown on the left with its original plastic holder is a "twin inverter amplifier". On the right four FF1 blocks are shown on a contemporary circuit card marked "100 KC DECADE". A decade counter requires four flip-flops,and each FF1 Circuit Block contains one flip-flop. The 1-Series was specified for operation at 100 KHz. There were quite a number of different types of these blocks, in many different bright colours, but all with the same ten wires emerging from one long edge. The block was made by fitting a small circuit board into a plastic case that was then filled with epoxy. It is hard to open them without destroying the contents, but I have opened one, and it contained three unbranded OC47 transistors and a few resistors. OC47 was a general purpose RF germanium PNP transistor made by Philips for internal and OEM use. These circuit blocks were also sold by Philips' UK manufacturing arm, Mullard Ltd., and possibly also by Philips' German subsidiary, Valvo GmbH. An interesting feature of these circuit blocks is that they are usually doubly-identified, for example, the 2.IA1 on the left bears a second identifier B894002 and both are used to refer to it in the Philips data books. However, I have several blocks which only have this long identifier, the shorter name such as 2.IA1 being replaced by what looks like a date code for a month in 1961.

Series1 card
Philips FF12 circuit block

By 1969 Philips had created several other series of circuit blocks. The 10-series, 20-series and 40-series had the same physical dimensions as the 1-Series, but had two staggered rows of wires, totalling nineteen. My image shows an FF12 from the 10-series, which is a "bistable multivibrator with built-in trigger gates and set-reset inputs".

A correspondent, John Murrell, informs me that the Series 1 modules are known as 'Combi Elements' by London Underground, where they were used extensively and were still in use in some equipment on the Victoria Line until surprsingly reently. In his own words, "for the first 20 years or so all the automatic driving of the trains relied on Combi Elements both on the train and trackside. The LU Museum should have one of the First Generation Auto Driver Boxes full of Combi Elements".

Philips/Mullard/Valvo also produced circuit blocks in a different shape, called NORBITs, shown in the image on the right approximately life size. These were plastic cuboids in several colours with two rows of pins, eight on one side and nine on the other. In appearance they resembled huge DIL chips, except for the different pin count on each side. The Philips Pocketbook for 1969 describes a series called 60-series, containing five members. In addition, I have two scanned pages from an unknown document that refers to a '61-series'. I also possess an FF60, which is presumably a 60-series flip-flop, and an FF90, which seems to indicate a further 90-series. A Web search also reveals a few references to 'NORBIT 2' and 'NORBIT S'. A correspondent has told me that some NORBITs contain transistors from Texas Instruments. If you can clarify the situation, please

These various Philips/Mullard/Valvo circuit blocks were not a commercial success, although a few applications of them can be found by searching the Web, which brings up a few patents that employ them. They were developed at the same time as monolithic ICs were becoming commercially available, and although the circuit blocks were probably initially cheaper, IC manufacture would become highly automated and prices were to drop dramatically.

Mullard NORBIT
Ferranti LCE modules

The Ferranti company made several series of packaged circuits, which they called "Logical Circuit Elements", abbreviated to LCE. These are well documented in the book "Anglo–American Microelectronics Data 1968–69: Manufacturers A–P" edited by G.W.A. Dummer and J. Mackenzie Robertson. Parts of this treasure trove of data on early ICs can be viewed on Google Books.

The various series are:

  • The 200 series of cylindrical packages using germanium devices.
  • The 300 series of cylindrical packages, similar to the 200 series but using silicon devices.
  • The 400 series, functionally identical to the 300 series but in a cuboid package.
  • The 500 series, combining silicon planar transistors with germanium diodes.
  • The 700 series, a low-cost industrial version of the 500 series types.

Dummer and Mackenzie Robertson also describe the LCE 1200 series of thin-film hybrid circuits.

I only have the two LCE701 units shown. I would be interested to get some from the other series, especially the cylindrical ones. If you can help provide some, please







Valid XHTML 1.0 Strict