Help understanding differences in transistors, especially there uses in RF

[rwgast_lowlevellogicdesin]

Ok so transistors are all different, npn, pnp, darlington, power, different gain, faster switching etc..... but my question is how do you pick the right transistor for the right application for instance I have some c2086 transistor classified as RF power amplifier for the HF band, are these for RX or TX??? Im hoping RX but I dont know jack from looking at the data sheet

http://www.radiomods.co.nz/transistors/2sc2086.pdf

I also have a bag of 2n3858a specified for AM an LF use...

http://pdf.datasheetcatalog.com/datasheets2/15/158090_1.pdf

And then I have some NE317 RF power output equivalent transistors.....

I have also noticed I have a lot of small signal general purpose epitaxial transistors, I read up on them on wikipedia but I dont see a clear case of when to choose them over a 2n3904....

I see all kinds of fenral purpose transistors used in different appliciations some I understand the choice either due to to gain or current handeling, but what makes one choose a a particular general NPN and what the heck makes an RF transistor different?

Im sorry if this question sounds newbish but I usually use a 2n2222/2n3904 and its pnp counter parts or I bump up to power transistors. Most of the time Im using them in a general fashion or for open loop gain. Im just trying to wrap my head around what makes an RF transistor special and how to tell if its TX or RX, and what is the point of the 1,000,000 genral purpose types of npn's and pnp's out there.

[KL27x]

Im just trying to wrap my head around what makes an RF transistor special

Frequency. This has implications in signal integrity, efficiency of power transmission, and in the case of FETs, switching current.

Transistors can only dissipate so much heat. While they are completely off, they essentially dissipate no heat. While they are completely on, they dissipate only a little bit of heat. The maximum sustained output rating is calculated when the transistor is fully on.

Between fully on and fully off, the transistor is dissipating varying amount of heat.

Wherever the resistance of the transistor is perfectly equal the the output impedance of the power supply, the maximum amount of heat is going to be dissipated. This is going to be way, way, way higher than when the transistor is fully on. And each and every time you switch the transistor, it is going to pass through this point.

RF is in the megahertz. To produce a radio freqency, the transistor must switch a million times a second. Compare that to an audio frequency of only thousands of times per second. To produce a radio frequency, the transistor is going to spend roughly 100x+ more time in the transition between on and off.  Hence, RF transistors are designated as such based on how fast they can switch between on and off. And vice versa. These are referred to as the rise and fall times. The faster they can achieve this transition, the more efficient they will be, and the sharper the signal is going to be. Also, in the case of FETs, the gate capacitance is another issue. An "RF" FET will typically have small gate capacitance designated by Q-something or other, I think. When switching so frequently, the current needed to switch the FET can be substantial.

Theoretically, if you use a 10 amp power FET at RF frequency, and it might only be capable of half an amp. It might produce something looking like a slightly sinusoidal flat line that is set somewhere between the rails and only moves up/down a small fraction of the rail voltage. And the current being dumped into switching it, through a very efficient RF low impedance FET driver, might theoretically be dozens of amps, in itself. So, it would be useful only as an electric heating element.

[rfeecs]

Im just trying to wrap my head around what makes an RF transistor special

Frequency.

Transistors can only dissipate so much heat. While they are completely off, they essentially dissipate no heat. While they are completely on, they dissipate only a little bit of heat. The maximum sustained output rating is calculated when the transistor is fully on.

Between fully on and fully off, the transistor is dissipating varying amount of heat.

Wherever the resistance of the transistor is perfectly equal the the output impedance of the power supply, the maximum amount of heat is going to be dissipated. This is going to be way, way, way higher than when the transistor is fully on.

RF is in the megahertz. To produce a radio freqency, the transistor must switch a million time a second. Compare that to an audio frequency of only thousands of times per second. To produce a radio frequency, the transistor is going to spend a lot more time in the transition between on and off.  Hence, RF transistors are designated as such based on how fast they can switch between on and off. And vice versa. These are referred to as the rise and fall times. The faster they can achieve this transition, the more efficient they will be, and the sharper the signal is going to be. ...

In many RF applications the transistor is on all the time (Class A).

RF transistors are simply capable of operating at higher frequencies.  The main thing that limits high frequency operation is the capacitance of the various junctions within the transistor.  Package parasitics, bondwire and lead inductances, etc. also can limit the performance.  Since the current gain or transconductance drops at higher frequencies, increasing these also increases high frequency gain.

So when looking at a datasheet, you would probably want to select a part that operates well at the bias point you have chosen (voltage, current).  Then you want a part that works well at the frequency of operation.  The datasheet may or may not tell you that.  Many RF transistors give an example performance (gain, power, noise figure) at a specific frequency, so that may give you some idea.  Another parameter that is sometimes listed is f_t, which is the frequency where the current gain drops to 0dB.  Some of the capacitances are sometimes listed on the datasheet.  These may give you an idea of how the gain is going to drop at higher frequencies and whether it will work at your operating frequency.

So for example, based on the bias point you wouldn't want to use an RF power transistor at the front end of your receiver.  You don't need to handle a large signal, so you don't need a lot of voltage or current.  You may be interested more in lower noise figure and higher gain.

[T3sl4co1l]

Differentiating characteristics of transistors:

BJTs -- the base can be made thin and wide, which causes high resistance (giving a long time constant between applying base voltage and charging junction capacitances) and high hFE.  Usually these have a low noise figure too, e.g. 2N5088, BC847C.

Junction breakdown voltage is inversely proportional to doping density (for a given doping profile), while hFE is proportional (sort of).  High voltage transistors have lower hFE, and vice versa.  hFE is higher when emitter doping is stronger than collector, so Vebo is almost always small, say 7V or less.

To get both high gain and high speed, higher doping and thicker base regions are needed.  Less base width is also a plus.  Whereas BC847 might be made with a single set of diffusions (so the base is connected to a ring of metal, where that diffusion touches the surface; and the emitter is diffused into the middle of that), RF transistors are made with many "fingers", increasing the perimeter and therefore reducing the resistance to the center of any area of the base.  So, RF transistors typically also have less hFE, Vebo and Vcbo than general purpose types.

For FETs, it's much the same: less spreading resistance and more optimal capacitance.  Specifically, in comparison to switching transistors (which are optimized for lower losses at switching frequencies, with less regard for input capacitance / matching), since linear operation is common, the pressure is towards larger die areas (more dissipation, freedom from 2nd breakdown) and smaller channel widths, leading to higher Rds(on) (hardly matters -- linear range), lower capacitances, and also the design is optimized for lower feedback capacitance (which a lot of modern switching designs finally achieve too, but this has been a constant pressure for RF parts).

As for transistors intended for specific bands, it's usually a combination of things: the power dissipation is suitable for a typical application (low power portable up to commercial/industrial power amps), the fT is some times the intended band (usually 5-20x, so that the power gain is reasonable), and the package is, well.. more or less suitable for the range.  This is tricky, because there are a number of transistors in conventional (i.e., terrible) packages, like TO-220, claimed for service at 30MHz and up.  As far as I know, they can't even be tuned for unconditional stability, and the s-parameters often show negative (i.e., generates power out the input port -- negative resistance!) values...

The good ones have suitable packages, typically with low inductance flat leads.  Transistors made for UHF+ are exclusively in this way, usually a wide flat pack with input (gate?) on one side and output (drain?) on the other.

BTW, there are also "pre-matched" transistors.  These incorporate matching/tuning circuitry for a particular load (50 ohms?) at the frequency range of interest, and aren't of much use outside that range.

Tim