Yes- the TO105 power dissipation is not as good as a T092, and the die orientation is different, being at right angles to the legs, the die is stuck to the header and then bond wires tacked accross. The blob top is done after initial electrical tests. In many ways this was a plastic version of the TO5, TO18 (bc108) TO46 etc metal cans where the die is at right angles to the legs, and is a fairly logical development, and can re-use much of the same production line. The two step testing makes sense with an expensive welded top can, less so with plastic encapsulation, where the saving is negligible.


Apart from re-tooling, the  TO92 is much better thermally, and cheaper to mass produce - one of the legs, usually the centre one, is widened out into a paddle to support the chip (becoming collector on an NPN), and all 3 legs are still connected to each other as a single piece of metal with slots in it while the die is bonded, ( so both ESD and mechanical support during overmoulding is easier). Once the plastic overmoulding is applied, the metal lead frame and the  links between the legs are cut away.  It is the location of the die and bond wires that make it flat on one side (the paddle side of the die) and fatter and rounded on the other - the bulge is the side with the die and bond wires.

The next step, to extend the mounting paddle out of the plastic for cooling bigger devices brings us the TO220 and TO202 and  TO126, and dissiapation as high as the metal bodied devices.


The TO105 designs had a short reign,  of perhaps 5 years or so,  overlapping the end of the metal bodied and the beginning of the modern paddle bonded designs.

I always got the impression that the TO-105 was the plastic version of the TO-5 metal can with the same lead spacings, so transistors with both packages would be interchangeable on the same PCB.


Two other oddities are Mullard / Philips transistors with the lockfit or X09 case, and Ferranti / Zetex transistors with the e-line or X11 case.


When exactly were devices in the TO-92 package first produced?


 

 

Not sure exactly but by 1966, (the same year the metal bodied BC107/8/9 were first released, ) Motorola were announcing the 2N3904 and 2n 4123/24  devices in TO92 according to their 1966 year book. With updated data sheets these are still made by many companies today.

It maybe that other makers were a year or so  ahead of them, but it really cannot have been by much.

Actually it must have been a time of very rapid progress when you consider the germanium black painted glass bodied OC71 and similar were being announced as the latest thing scarcely ten years earlier. (the OC 50 point contact  in 1952 and the diffused junction OC71 in 1957 according to wikipedia).