Some basic specs and dates. I don't have ready access to ANSI, Jedec or PrElectron data.

Date.....Power....Max Voltage...Max current...Frequency...Manufacturer
.............mW............Vce.............Ic...............Ft

2N3704

1970.......360............30.............800.............100...........Texas

1981........360...........30.............500..............100..........Texas

1984........360...........30.............800..............100...........Texas

1993........350...........30.............800..............150...........Philips

2008.........?..............?................?..................?............Indian

2N3904

1981........500...........40..............200...............300..........Ferranti

1984........310...........40..............200...............250..........Moto

1993.........625..........65..............100...............300..........Philips

2008.........350..........65..............500...............300...........Chinese

Will the real 2N3704 and 2N3904 please stand up?

 

I hear you both.

It would help a great deal to have a meaningful revision letter behind the part number.
But that complicates things. The Euro semiconductors already use the revision letter for other arcane purposes - usually counter-intuitive. You might think that a "C' transistor would have better specs than an "A" transistor - but au contare'.

Then someone had the gall to use letters to indicate different pin configurations. This is an absurdity which someone should've been boiled in acid over.

Revision letters should only represent improvements in an existing design. If there is a change in form, fit or function, the part number must be changed.

However, if a part not only meets, but exceeds the original specifications - or another previously superceding specification - then it should be assigned a new revision letter.

I do have some expertise in this area.

 

The Euro semiconductors already use the revision letter for other arcane purposes - usually counter-intuitive. You might think that a "C' transistor would have better specs than an "A" transistor - but au contare'.

It has long been the European practice to use a letter suffix to indicate gain (beta) groups, into which transistors are sorted after manufacture.

Thus transistors which are labelled BC108A have lowest gain, BC108B next lowest and BC108C the highest. This has continued through all the subsequent descendents of this famous transistor eg the BC548A/B/C and complementary BC558A/B/C (there were no direct complements of the original BC108).

It was Texas Instruments, using the European marking standard for some reason, that introduced the letter suffix to indicate pinouts eg with the BC182 and BC212.

Incidentally the letter B stands for Silicon and the letter C stands for Triode in the BCxxx a sensible system. this is a carry over from older valve (tube or bottle) marking practice.

The same variation of specification over the years and between manufacturer occurred in the era of valves. Today valves are still manufactured in Eastern Europe and many are finding problems with slightly different specs for standard numbers from these sources since the resurgance of interest in valve audio amplifiers.

For the benefit of younger members just starting.

Components specs are registered, by their originator, with bodies such as ANSI (USA) or JEDEC (Japan) or ProElectron (Europe) against a specific number. Any subsequent manufacturer is supposed to conform to these but many don't.

A particular case in point, the original 2N3055 transistor had very low frequency response which made it poor even in audio amplifiers, but good in power supplies..
Many manufacturers round the world started (and still do) making higher frequency versions which oscillated all too readily when substituted into lightly compensated designs, Ok for the original.