Andy:
I agree that in the early days transistors of the Ge type were more commonly made as PNP devices. As Si devices came on the scene there was a shift toward the NPN type devices.

Years later as the "IC chip" age came into being with Si wafers it was much easier to make bulk N type material. Diffusion transistors were then made by doping the base material to make P areas on the N substrate and then N was made again on top the P areas. The result was that NPN transistors were very most common on the bipolar IC chips (analog, digital or otherwise). If these early chips had a requirement for PNP structures on the same chip they were often made using a "lateral" construction technique. (i.e. the transistor was made flat to the surface of the chip as opposed to diffused down into the chip).

I have before me a copy of the GE Transistor Manual, 7th Edition, 1964; it has almost equal treatment of theory for PNP and NPN transistors. It shows some "upside-down" schematics and "right-side-up" schematics using an assortment of PNP and NPN devices. The book does discuss "logic circuits" as implemented with either PNP or NPN discrete devices!!

As IC technology matured with the introduction of bipolar logic devices they had on-board NPN transistors which had saturating outputs referenced to ground (i.e. the conventional TTL type logic). I still contend that that the use of external discrete NPN devices became the norm because it is much easier to bias and switch NPN types from a 0.4 -> 2.7V TTL signal (when a 5V supply is in use) than it is to switch a PNP device. As such many engineers simply quit using PNP type discrete devices!! Over the years I have met MANY engineers and electronics technicians that understand NPN devices but readily admit that PNP parts are a mystery or too confusing to them.

These days things COULD be different. Many CMOS logic circuits and microcontrollers are no longer crippled with non-symmetric output voltage swings. It is typical to find devices that can easily drive outputs rail-to-rail and as such it is just as easy to design in discrete PNP devices as it is discrete NPN devices. However the expreience from the 70's and 80's persists and so you still see more NPN types used in circuit designs. Manufacturers have responded to this and thus offer greater selections of NPN devices as opposed to PNP devices.

With modern manufacturing technology it is nearly certain that any argument that "NPN devices are easier to make than PNP devices" is NULL and VOID. It is routinely possible to find NPN devices and sister PNP parts that are nearly identical in specifications (albeit complementary to each other). If one type were more difficult to manufacture over the other you could certainly expect there to be a price difference between the two. It has not been my experience to find this to be the case when you are comparing apples to apples. Take for example the following extracts from two Fairchild WEB pages. The price is exactly the same to a fraction of a cent!!






Michael Karas