Power per device, or per internal diode? Which si the total power one?

[Behemot]

Sorry for making new thread, but nobody answered in the FAQ yet and I definitelly need to know whether I should upgrade most of the rectifiers in shit PSUs, or just some of them to do the labeled output.

So, which one is the maximum power of rectifier in secondary? Cause according to secondary schemes (pretty much most of them if not all) how I see it, power flows through one diode only during every part of the cycle.

BUT pretty much all webs take power per device as maximum output. So which one is it actually?

[mariushm]

Re: Power per device, or per internal diode? Which si the total power one?

The current flows through two diodes at a time in a bridge rectifier.

The maximum power depends on the construction and the temperature of the rectifier - for example take this 400v 4A rectifier:

http://search.digikey.com/us/en/prod...04DI-ND/278605

So in theory it can do 120v x 4A or about 480 but if you check the datasheet :



it's only able to do that with a heatsink and only up to 100c - go above and the performance degrades. Without heatsink, it can do a bit more than 2 Amps per diode or about 250-300 watts.



So maybe just slapping a heatsink on them would be enough, rather than replacing them.

From what I heard in reality these rectifiers are usually better than the specs and don't fail so often but then again, most people using those cheap power supplies don't draw a lot of power in the first place and the fan cools them down a bit.

[Behemot]

Re: Power per device, or per internal diode? Which si the total power one?

That is primary rectifier, pretty clear (two diodes at a time but in series). I was asking more for secondary rectifiers consisting basicly of two diodes with common output.

[Th3_uN1Qu3]

Re: Power per device, or per internal diode? Which si the total power one?

I was asking more for secondary rectifiers consisting basicly of two diodes with common output.

Total device current is what determines the maximum output. Since the output of the two diodes is common, it comes logical that the currents add up. So if each internal diode is rated for 15A, then you got a 30A part. And that's exactly how they're marked as well.

The way you asked the question is confusing, because power can refer to many things, for example power dissipation. Which is sometimes a problem even though the current rating is adequate. Power is simply V*I, so all you have to do is measure the voltage drop of the diodes at the maximum current the supply is expected to provide, multiply it by current, add those figures up for all the diodes and calculate whether the heatsink is adequate, or you need beefier devices. This testing can be easily performed with a current source, which can be as simple as a 12v battery, a hefty transistor and a few resistors.

Bigger diodes also yield higher efficiency, if you lower the voltage drop you of course lower the wasted power as well.

[Behemot]

Re: Power per device, or per internal diode? Which si the total power one?

Yeah I know the other stuff. But yet how I understand this

Though the output rectifiers are usually two rectifiers in one package, in the most common topology, the two devices have different functions. One of the devices conducts during the “on” part of the square wave from the transformer, charging the inductor. The other rectifier conducts during the “off” time, providing a current path for the discharge of the inductor.

is that power flows through one diode only in any moment. Thats the reason why I asked.

From the scheme A here I can understand how the current flows through both didoes, but only in one half of the cycle. The other half they should be in backward, effectivelly throwing off half of the trasformer power?

The scheme B I don't quite get at all…

[Th3_uN1Qu3]

Re: Power per device, or per internal diode? Which si the total power one?

Scheme A is half bridge. It's your basic full wave center tapped rectifier. Remember that when the transformer secondary is center tapped, you have V+ and V- relative to ground, so one diode conducts on each half, effectively doubling the ripple frequency and making the caps' life easier.

Scheme B is two transistor forward, half wave rectification. The second diode blocks the reverse voltage spike from the inductor, it's called the "freewheeling" diode. It isn't needed in mains powered applications or in unregulated SMPS where the diode outputs direct to the capacitor, but things change when an inductor is present.

Btw, that article is pretty inaccurate - half bridge can have higher efficiency than double forward, it's just that the latter is easier to drive because the two transistors operate at the same time so there's no need for push-pull outputs on the controller, dead time, and all that stuff.

[Behemot]

Re: Power per device, or per internal diode? Which si the total power one?

So in B current flows only through the upper diode and the second just blocks spike from inductor? Cause I guess there cannot be output voltage on ground. Or there can, and than the inductor blocks it so it flows only through both diodes?

[Th3_uN1Qu3]

Re: Power per device, or per internal diode? Which si the total power one?

The load is shared about 70%/30% between the rectification diode and the freewheeling diode. The freewheeling diode isn't absolutely required, but it does make the main diode's job easier. It becomes more important in a synchronous converter where the diodes are replaced by MOSFETs, as the reverse voltage spike of the inductor will damage the high side switch (the rectifier) if the low side switch (replacing the freewheeling diode) isn't there.

[Behemot]

Re: Power per device, or per internal diode? Which si the total power one?

I see.

The reason why I am asking at all is that on some of the shitty supplies the per-device current is usually (if even) just slightly higher than rated which makes per-diode much smaller than rated. So, I change the rectifiers with lower-than-rated per-device current (fortunately usually one per PSU), but if I would have to change everything, I can fuck the job, you know what :-)

So taking in mind topology which puts most stress on jsut one diode at a time: how big percentage of rated current should handle single internal diode to make it safe for the maximum current?

[Th3_uN1Qu3]

Re: Power per device, or per internal diode? Which si the total power one?

Uh, you lost me. What exactly are you trying to do? Most cheap supplies are half bridge anyway - because that's what was popular when they copied the design.

[Behemot]

Re: Power per device, or per internal diode? Which si the total power one?

Yeah lol.

As I said, trying to determine the actual rating I need to use. If there is 25 A labeled and 30A rectifier (2x15 A), would that be enough (not countign other parts now)?

There is also second thing - most datasheets list some peak current with high frequency. I did my calc once and well, typical period on some 20+ kHz is much shorter than this time listed. It usually says it's capable of some 10times higher current in such case…