Have some questions about the primary side

[KMFDM]

I have a couple questions. By looking at the specs of a switching transistor or a MOSFET how do you know how many watts it's good for? (assuming a good heatsink and airflow) Like this one: http://www.fairchildsemi.com/ds/FJ/FJP13009.pdf (I know it's old )

400V 6A, how do you calculate that out? And some have 2...or 3, how does that contribute to how much power they can handle? And how come some mix MOSFETs and transistors together? Thank you in advance

[McSteel]

In the example of FJP13009 that you linked, it's not really 400V (between Collector and Emitter) x 6A (out of Collector), unless you're pulling said 6A for less than 10 microseconds at a time, then allowing for at least one microsecond for turn-off/cooldown. In the linked PDF, you have to look at figures 5 and 6, to see how much time the transistor needs to turn off, and what it's operating area is. You may choose any value of DC voltage and DC current inside the operating area, if you respect the maximum on-time. After that passes, you need to respect the minimum off-time from figure 5, and you're golden (if you cool the transistor down properly). As usual, Watts = Volts x Amps. Having more transistors is necessary for some primary topologies, and usually doesn't mean your max power is added together - it's usually determined to be the same as the weakest transistor.

As for MOSFETs, and mixing them with standard NPNs, they're typically more capable as switchers, but also more expensive. Sometimes el-cheapo PSU will use an NPN as a 5VSB switcher, or perhaps as a driver for primary switchers... It all depends on the design.

[cypherpunks]

McSteeel said it well. It's not particularly easy. You need to know a fair bit about the circuit around the transistor as well. For example, full-bridge topologies are popular at higher powers precisely because they can get more power out of a given switching transistor that a simple flyback. But at low power, the simplicity of the flyback circuit saves enough money to pay for a beefier transistor with profit left over.

I can say that multiple paralleled transistors give you effectively that many times the power of a single transistor. There's some allowance for unequal sharing, but it's fairly small.

As for why MOSFETS... Bipolar transistors have the property that they conduct better as they heat up. This makes them unstable: if one part of the transistor gets hotter, more current will flow through it, and it'll get hotter yet, leading to that part burning out. Keeping this problem under control is what causes many of the the "safe operating area" restrictions. (There are others, but if you want a course on semiconductor physics, there are plenty on line already.)

MOSFETs are very useful in that they have the opposite reaction to heat. A hot MOSFET or part of a MOSFET will conduct slightly worse, diverting current to other parts of the transistor, which stabilizes things. This makes them better for high-power operation. But they cost more, so if a bipolar transistor will do, a few cents can be shaved that way.

ATX power supplies are actually particularly tricky to understand because the designers pull all sorts of non-obvious tricks to save money. Often, the main transformer will be designed with particular leakage inductances or interwinding capacitances that are exploited by other parts of the circuit.

[KMFDM]

Thank you very much for the replies very useful info!!!