NPN vs PNP switching frequency

[made2hack]

Hello all,

Take the 2N5551 (NPN) and the 2N5401 (PNP) from Fairchild. Both 0.6A 625mW 150V jelly bean transistors. However, the 2N5551 has a max freq. of 100Mhz and the 2N5401 has a max freq. of 400Mhz.

Why the difference? And will it matter when I am using both in a circuit? And when will it matter? I mean 100-400Mhz seems high speed for most of my needs at the moment, but I don't know in the future. When will it matter that they be the same frequency? Analog? Digital? Both?

[edavid]

Look again, they both have a minimum f_T of 100MHz.  They are probably quite similar.

I know that doesn't really answer your question...

[retrolefty]

Hello all,

Take the 2N5551 (NPN) and the 2N5401 (PNP) from Fairchild. Both 0.6A 625mW 150V jelly bean transistors. However, the 2N5551 has a max freq. of 100Mhz and the 2N5401 has a max freq. of 400Mhz.

Why the difference? And will it matter when I am using both in a circuit? And when will it matter? I mean 100-400Mhz seems high speed for most of my needs at the moment, but I don't know in the future. When will it matter that they be the same frequency? Analog? Digital? Both?

 Typically one selects the transistor for a specific circuit rather the adapt the circuit for a specific transistor you have on hand. The datasheet is your best guide as to if a device is suitable for the voltage/current/frequency used in a specific circuit. Also there are complementary transistor npn/pnp pairs that often used like the popular 2N3904/2N3906 often used in circuits requiring opposite polarity transistors with compatible ratings.

[orolo]

Hi. The ft figure is not the switching frequency, but the transition frequency, that is, the theoretical frequency at which you reach unity small signal current gain. Switching frequency will be much slower, since it must take into account storage, delay and rise times, among other things, like some huge Ccb feedback while switching (these are not specified in the datasheets, which look rather sketchy). The fairchild datasheets are quite different for these transistors, but, as edavid said, the minimum ft is the same for both transistors, while the max is only quoted for the pnp part. Looking up the NXP datasheets for the 5041 and 5551, I see that they are both between 100 and 300 Mhz. (Hmm, the NXP datasheet for the 5401 has the noise figure measured in pf instead of dB  . ) Anyway, getting your switching frequency well into the tens of megahertzs with a discrete design is quite a feat; I very much doubt switching at 400Mhz is even possible (without ECL).

Anyway, to me the 5551 and 5401 look well matched, ft included, and they should not behave very differently in a circuit. They are high voltage transistors; for normal use I would suggest other jellybeans, like 2222/2907 or 3904/3907. If I'm not wrong, lower voltage parts switch faster than high voltage rated transistors. Really fast switching transistors, like the 2369, have a low breakdown voltage. I'm not an expert, but there seems to be a compromise there.

[T3sl4co1l]

It's reasonable to make a digital switching circuit (of the conventional, common emitter topology) at up to fT/10 or so.  That's ~30MHz for 2N3904s or 4401s and the like.  You may have to run at quite high currents (and dissipations, at least in the loads; which may have to be current sources/sinks, rather than plain old resistors), in order to get the effective load impedance down, to drive the capacitances fast enough.

It's possible to get real power gain above fT (because fT only counts current gain, not voltage gain), but it's highly unlikely to get anything of use above maybe 2-3 * fT.

The physical mechanism that determines fT is how much resistance the base layer has (a wider, thinner base layer, with less perimeter connecting to the metal terminal, simply can't be charged and discharged quickly), and recombination time of the semiconductor (which is improved with stronger emitter doping, gold doping, etc.).  Most BJTs are designed so that both effects cut in about the same place.

You can tell an RF transistor from some other datasheet parameters.  Veb is lower (due to the higher B-E doping).  Die photographs of them show much smaller, and more numerous, "fingers" (so, more perimeter to connect to; less base resistance).

Tim

[made2hack]

Very interesting.

Thanks guys,