PN3904 vs 2N3904

[Electro Fan]

Is a PN3904 equivalent to a2N3904 (they just have different pinout configurations), or are they 2 fundamentally different transistors?  Thx

[dannyf]

From where I stand, the difference between the two is 'P' (ascii 80) and '2' (ascii 50), or 30.

[Electro Fan]

From where I stand, the difference between the two is 'P' (ascii 80) and '2' (ascii 50), or 30.

That's cool but how about any electrical differences

[xrunner]

Is a PN3904 equivalent to a2N3904 (they just have different pinout configurations), or are they 2 fundamentally different transistors?  Thx

A cursory look at the datasheets says basically equivalent. A basic general purpose transistor with a gain of 50 - 300. Good for basic switching and amplification at lower frequencies.

[c4757p]

They should have the same pinouts too...

2N3904 is a standard part number, in a plastic package.

2N2222 is also standardized, but it's a metal can, so many manufacturers made a PN2222 which was the same thing but in plastic.

I guess someone decided to make a PN3904 along the same lines...? It's redundant.

[EEVblog]

If you are using a 2N3904 then you are using it because it's a jellybean tranny, so PN3904 should make no difference. It is designed to be equivalent.

[dannyf]

That's cool but how about any electrical differences

Practically none, unless the datasheets suggest otherwise.

It is essentially the same process licensed to different vendors.

It exists for many popular transistors: Toshiba's 1943/5200 and 1302/3281 are produced also by Fairchild and OnSemi too.

Same with some ICs.

[Electro Fan]

Ok, Thanks

Just for the record, does 2 or the P (or any other character in the first position) signify something that can be deciphered from a table somewhere?  Same question regarding the N in the second position?  (Or are the characters in those positions essentially meaningless and a 3904 is a 3904 regardless of those two prefix characters?) 

Thanks again

[westfw]

Someone posted recently (here?  Somewhere!) that they had a PNxxxx transistor that didn't have the same pinout as the 2nxxxx transistor (with xxxx being the same.)  So check that data sheet carefully before populating a PCB.

[Rupunzell]

There was a time when 2N parts were often in metal cans with glass hermetic seals intended for MIL-spec applications.

PN numbers originated when National Semiconductor introduced their low cost version of epoxy TO-92 bipolar transistors in response to the Fairchild "pop top" transistors.

These days, most bipolar transistors are derived from a limited set of die parts that can meet a broad range of specifications by selection in process. This along with die size reduction and consolidation of part numbers offered makes up most discrete device offerings today. Knowing this, be really careful if designing a discrete part into a circuit that is demanding to every letter and number of the specification as parts with the same generic part number can have significant different performance. The spec sheet is the min and manufactures will often meet their spec sheet and no more. This is also why specifications for many discrete devices have limited parameters on their data sheets.

For those who really care and really wanna know, a Tektronix 576 curve tracer is a very good friend that gives honest information about a specific part.


Bernice

[helius]

These days, most bipolar transistors are derived from a limited set of die parts that can meet a broad range of specifications by selection in process. This along with die size reduction and consolidation of part numbers offered makes up most discrete device offerings today.

This is an excellent point. Shifting fabrication technologies and economics is going to affect all discrete devices. The parameters for newer processes are different than the ones the devices were originally designed for and generally have worse analog performance (higher leakage, DIBL, etc).

[TimFox]

American "2N" numbers were intended to be generic, i.e. independent of manufacturer, with parameters registered  at JEDEC.  (It looks like the acronym JEDEC is now a word, their website no longer lists the meaning of the abbreviation.)  European and Japanese devices have different prefixes.
The meaning of the prefix is:
1N:  diode
2N:  triode (e.g., bipolar PNP, NPN, JFET, MOS triode)
3N:  more electrodes (e.g., dual-gate MOSFET)
4N:  even more electrodes (e.g., photocoupler)
"PN", "MMBT", and similar prefixes were originally manufacturer-specific, but were often copied by competitors.
The "3904" in 2N3904 is merely a number assigned when the device was registered, in order of registration. 

[Rupunzell]

Tektronix 576 curve tracer image of a genuine Motorola MPS6915 from decades ago when Motorola made them... Collector voltage run to device break down. Motorola spec for Collector/Emitter breakdown voltage is 25 volts, Motorola means this.
http://www.datasheetarchive.com/dlmain/Datasheets-21/DSA-409785.pdf




Current generic MPS6519 curves generated on the opposite socket of the Tek 576.
Not quite the same, and this is just the beginning of the differences.



Generic devices are often not the same as the original device specifications published by the manufacture.


Bernice

[T3sl4co1l]

Bern,

I don't suppose you have a selection of 2N3055s to scare the audience with, do you?

Remember, JEDEC parts are generally "meets or exceeds".  If the parameter doesn't come with a max, ASSUME IT DOESN'T HAVE ONE!  This makes for interesting reading, when a 2N2222 is replaced with something with an fT > 800MHz, or a 2N3055 with Vce(sat) of 0.5V!

Tim

[T3sl4co1l]

There was a time when 2N parts were often in metal cans with glass hermetic seals intended for MIL-spec applications.

Although that's still ongoing; you can buy* silicon carbide transistors in 2N8xxx numbers.  Apparently the 2N series has been pretty neglected since, what, the 70s or so? (The next most common and recent numbers being, among others, SCRs in the 2N6xxx's and the 2N7000/7002 MOSFET; I guess.)

*A relative term.  Pricey if you can find them, "Call us" if not.

Tim

[SeanB]

Only $50 each.......

http://za.rs-online.com/web/p/mosfet-transistors/8098995/
http://za.rs-online.com/web/p/mosfet-transistors/8098991/
 
Though only the one is available ex stock.

SiC diodes though range from cheap to not, starting at $1 each.

http://za.rs-online.com/web/p/rectifier-schottky-diodes/8099045/
http://za.rs-online.com/web/p/rectifier-schottky-diodes/7387635/

Top end is not exactly cheap though, $150 per unit........

[T3sl4co1l]

I meant like these:
http://www.digikey.com/product-detail/en/2N7637-GA/1242-1148-ND/3679828
My bad, they aren't even into the 8000s yet!

Commercial SiC diodes are much more fairly priced over here (from Cree, Infineon, Rohm..), and quite usable for many applications.  SiC transistors are pretty reasonable for high voltages, though still on the expensive side.  Although the prices down there are .

Tim

[rsjsouza]

Someone posted recently (here?  Somewhere!) that they had a PNxxxx transistor that didn't have the same pinout as the 2nxxxx transistor (with xxxx being the same.)  So check that data sheet carefully before populating a PCB.

westfw, 2N2222 and PN2222 is one of them

[T3sl4co1l]

P2N2222*

[Rupunzell]

Yes we have plenty of 2N3055, decades of various brands and can styles including the later ones in TO247 plastic package.
Run them hard enough in a curve tracer and the beta change due to die heating is visible on the displays curves. Pulsed testing tends to reduce the stress significantly.

Consider how many text book assignments were written with 2N2222, 2N2219, 2N2907, 2N3904, 2N3906 and others. These part numbers continue on to this day in designs due to those home work and text book assignments. Those who out grew these parts in designs move on knowing more than just those class activities.
 
One has not lived unit trying to test microwave devices in a curve tracer. Part of the challenge is keeping them from oscillating and mis-behaving.


Bernice

Bernice,

I don't suppose you have a selection of 2N3055s to scare the audience with, do you?

Remember, JEDEC parts are generally "meets or exceeds".  If the parameter doesn't come with a max, ASSUME IT DOESN'T HAVE ONE!  This makes for interesting reading, when a 2N2222 is replaced with something with an fT > 800MHz, or a 2N3055 with Vce(sat) of 0.5V!

Tim