What's the deal with full-wave rectifiers?

[ben7]

Ok, so, I am hoping this isn't going to turn into a flame war, but I just have to ask:

Why do so many people hate the use of 4 diodes instead of one bridge rectifier?

I do see why, if it needs to be mounted to a heatsink, but people blabber about this even when the bridge rectifier can't be mounted to a heatsink...

[Pentium4]

Re: What's the deal with full-wave rectifiers?

Don't the 4 diodes run a little hotter since most of them are so close to each other and the board, allowing less cooling for themselves? And I think it's also when you see 4 diodes you can bet the manufacturer cut corners in other ways too

[370forlife]

Re: What's the deal with full-wave rectifiers?

Don't the 4 diodes run a little hotter since most of them are so close to each other and the board, allowing less cooling for themselves? And I think it's also when you see 4 diodes you can bet the manufacturer cut corners in other ways too

Its this mostly. If you crack open a decent quality 150W power supply you will most likely see a 4A bridge rectifier, where as cheapo-250W supplies will use the 4 diode treatment. Sure it works and they can deliver 250W but when you see those you know the manufacturer puts cost per unit before quality.

[mariushm]

Re: What's the deal with full-wave rectifiers?

Individual diodes have less surface so in theory they heat faster so you have to derate them as needed. They don't heat as much from the other diodes near them.

HOWEVER, the diodes are usually in the path of the fan, so as long as the fan works, there's cooler air flowing over their surface.

Also keep in mind 2 out of the 4 diodes are more stressed, they heat up more, so there's going to be a small variation in forward voltage drop. In practice it's not really important.

Diodes in a bridge rectifier are sharing the same larger surface so they have the potential to dissipate the temperature better and there will be less variation between the individual diodes' temperature in the package.
Heatsink can also be applied much easier which further raises the current handling capability of the bridge rectifier.

BUT, in cheap power supplies, the bridge rectifier is often without a heatsink and the bridge rectifier is horizontally right near a large round primary capacitor, so the airflow is probably less good, only the sides of the bridge rectifier may cool from the fan airflow.

There's another argument for individual diodes but doesn't really apply to power supplies: individual diodes make it easier to apply additional filtering in parallel with each diode.
For some applications (like audio for example) you sometimes want to filter those tiny drops between the moment when the diode is completely off and the moment the diode starts conducting, using RC filters.

See for example https://cdn.badcaps-static.com/pdfs/...fc2cc34d0a.pdf if you're interested.

[Pentium4]

Re: What's the deal with full-wave rectifiers?

I wouldn't mind 4 diodes so much on 230V. We need to get 230VAC over here in the states

[cheapie]

Re: What's the deal with full-wave rectifiers?

I wouldn't mind 4 diodes so much on 230V. We need to get 230VAC over here in the states

Yes, I agree. And 50Hz.

[370forlife]

Re: What's the deal with full-wave rectifiers?

I wouldn't mind 4 diodes so much on 230V. We need to get 230VAC over here in the states

Leadman beat you to that, on some of their bottom of the barrel units, two of the diodes are rated for 2A and 2 of them are rated for 1A. The 2A units do the rectifying while on 120V, while the other two rectify when running 230V.

[Pentium4]

Re: What's the deal with full-wave rectifiers?

Leadman beat you to that, on some of their bottom of the barrel units, two of the diodes are rated for 2A and 2 of them are rated for 1A. The 2A units do the rectifying while on 120V, while the other two rectify when running 230V.

Interesting. Those horrible flyback designs with insanely high ripple? I saw one recently with no cap or inductor on the 3.3V but a cap on the -12V! It had a 4A bridge though

[ben7]

Re: What's the deal with full-wave rectifiers?

Yeah, that is just terrible. I am talking mainly about the 3A ones, ex: 1N540x

This is almost sort of a rant thread, because it bothers me how quickly people get upset when they see 4 separate diodes - which are just fine for the application (wattage, etc...). Also people take for granted the fact that cheapo supplies don't use bridges that often - high quality ones do too occasionally!

I have a Dell PS-5151-1 psu here and it is only rated at 145w. Of course, it has a bridge rectifier. I also have a 90w psu made by Panasonic, which was in a printer. It has four 2A diodes. Works perfectly fine, no signs of heat stress on the circuit board...

Then, there were the fluorescent lamp ballasts. They were rated at about 90 watts max, but they used 2A diodes with a high voltage drop... not good! All of them had charring (!) around the diodes. Of course the failures were not the diodes, but the solder on the hv transformer.

[smason]

Re: What's the deal with full-wave rectifiers?

This is almost sort of a rant thread, because it bothers me how quickly people get upset when they see 4 separate diodes - which are just fine for the application (wattage, etc...).

I couldn't agree more.

[Wester547]

Re: What's the deal with full-wave rectifiers?

I don't think the type of rectification changes on the input side whether you use 4 diodes or a bridge rectifier.... there are still 4 diodes either way (unless you're referencing 3 diodes which would lower efficiency and perhaps change a thing or two)... only forward / flyback changes the kind of rectification used on the output (secondary, excepting +5VSB which I believe is always half-wave rectified since it's on the output of a flyback transformer), but that's another topic in entirety.

I guess it's the same reason why people get upset at a lack of input filtering... you could go wrong in worse areas in the design of a PSU. That being said, because diodes in a bridge rectifier are in a glass passivated package they do dissipate heat better, but then diodes do theoretically receive more airflow by comparison, so.... it could just be argued that bridge rectifiers have lower voltage drops than "4 diodes" so they are more efficient and probably do dissipate heat better overall, especially when mounted to a heatsink and especially when being overspec'd. Note that like coils and resistors, diodes do run warm if not hot (though definitely not as hot as resistors), so it helps to give them good cooling and heatsinks.

As also stated, with a 230V input, you are giving a bridge rectifier or 4 diodes double the current ability, so it may matter more on an 115V input. Mariushm also brings up a good point:

BUT, in cheap power supplies, the bridge rectifier is often without a heatsink and the bridge rectifier is horizontally right near a large round primary capacitor, so the airflow is probably less good, only the sides of the bridge rectifier may cool from the fan airflow.

This is important, especially in PSUs with 80mm fans (and this is also where PSUs with 120mm fans go wrong when they lay the bridge rectifier down or glue it to the PCB, which also impedes airflow). Angling the bridge rectifier in the path of the airflow makes a big difference in comparison to having input filtering blocking its path and having it facing the side of the PSU. Another thing to consider is that unlike secondary diode packages, bridge rectifiers often run without a heatsink so yes, they do run hot.

One last thing to consider is that bridge rectifiers are rated at very high temperatures in the datasheet (when mounted to a heatsink and/or a thermal compound of sorts) - often 100*C if not higher - I doubt they are getting that hot in a PSU, especially being so close to the fan, so they can still do good. It's the same thing with secondary diode packages - they are rated often at even higher temperatures - 100*C-150*+C, and they probably won't get that hot in a PSU so it isn't the worst thing in the whole world when you see 30A on the label for +5V, for an example, and only see a 30A rectifier on that output (probably in a TO-247 package...).... it can do its job (I guess that depends on the duty cycle as well). This also gets into the "ultrafast recovery rectifier vs. schottky" altercation but schottkys do have a significantly lower voltage drop.

I think the issue isn't that 4 diodes are used but that PSU manufactures really shouldn't defer to such methods when building high wattage PSUs. Ideally, a bridge rectifier or one on a heatsink would be used, for 300W+ PSUs at least... I say that because if all the components are up to spec but the bridge rectifier then it will short if you pull more current than what it can handle (if rectifiers AC to DC and the amount of power the PSU can provide does start there).

[lti]

Re: What's the deal with full-wave rectifiers?

People used to assume that any power supply with a bridge rectifier could output more power than one that used four diodes. Of course, there are 1A bridge rectifiers.

There was a thread here where I got yelled at for simply mentioning the kind of diodes used in a cell phone charger. The guy acted like I said it would need a 15A bridge rectifier to get the full 2.5W output.

[momaka]

Re: What's the deal with full-wave rectifiers?

I don't mind the 4-diode treatment that much, especially in low-power PSUs. If the diodes are specced well enough to handle the load with a bit of headroom, I leave them alone.

And yeah, I don't like the bashing of the 4-diode treatment either. After all, I've seen many MANY CRT TVs and monitors use those without any problem.

[eccerr0r]

Re: What's the deal with full-wave rectifiers?

I don't have any problem with 4-diode implementations of a rectifier, just make sure you use the right diode for the device in question.

Now the question is the plastic used too - wouldn't the additional plastic in the bridge add as additional thermal resistance leaving from the diode die? And also, four diodes have the surface area of four diodes...

Also, have anyone tried cutting a bridge rectifier open? I wouldn't be surprised if some of them you see four discrete diodes in them (!). I can't think of a way to make a monolithic, balanced, isolated bridge... Anyone know?

I suppose people just like seeing fewer devices on the PCB. It's marginally cheaper to manufacture a through-hole PCB with a bridge rectifier (4 holes and one device to insert) vs four diodes (8 holes and 4 devices to insert) but bridges are slightly more expensive in terms of device cost. I don't care either way, as long as it's designed to last.

(and why 50Hz? I want 400Hz AC mains Yay for smaller AC distribution transformers! Then again...skin effect... ouch )

[ben7]

Re: What's the deal with full-wave rectifiers?

I don't have any problem with 4-diode implementations of a rectifier, just make sure you use the right diode for the device in question.

Now the question is the plastic used too - wouldn't the additional plastic in the bridge add as additional thermal resistance leaving from the diode die? And also, four diodes have the surface area of four diodes...

Also, have anyone tried cutting a bridge rectifier open? I wouldn't be surprised if some of them you see four discrete diodes in them (!). I can't think of a way to make a monolithic, balanced, isolated bridge... Anyone know?

I suppose people just like seeing fewer devices on the PCB. It's marginally cheaper to manufacture a through-hole PCB with a bridge rectifier (4 holes and one device to insert) vs four diodes (8 holes and 4 devices to insert) but bridges are slightly more expensive in terms of device cost. I don't care either way, as long as it's designed to last.

(and why 50Hz? I want 400Hz AC mains Yay for smaller AC distribution transformers! Then again...skin effect... ouch )

The electric trains here, owned and operated by Amtrak, run at 25Hz

[999999999]

Re: What's the deal with full-wave rectifiers?

Now the question is the plastic used too - wouldn't the additional plastic in the bridge add as additional thermal resistance leaving from the diode die? And also, four diodes have the surface area of four diodes...

It could be a factor but it's taken into account by the current rating. They can actually take quite a bit higher pulsed current.

Also, have anyone tried cutting a bridge rectifier open? I wouldn't be surprised if some of them you see four discrete diodes in them (!). I can't think of a way to make a monolithic, balanced, isolated bridge... Anyone know?

Over a dozen years ago you would often see 4 discrete diodes wired together, in a regular plastic package, and plain epoxy sealing them in. I don't know if all that looked like this were built this way, but probably were.


Now it's more common to see them built with 4 bare dice and a thermoplastic or epoxy encasement like this, and by building this way you can get larger die in the same space:

I suppose people just like seeing fewer devices on the PCB. It's marginally cheaper to manufacture a through-hole PCB with a bridge rectifier (4 holes and one device to insert) vs four diodes (8 holes and 4 devices to insert) but bridges are slightly more expensive in terms of device cost. I don't care either way, as long as it's designed to last.

Takes up less space, far easier to add a heatsink, fewer parts to pick and place on the PCB. Cheap PSU usually have plenty of free space on the PCB in ATX form factor, while better PSU not so much.

[Pentium4]

Re: What's the deal with full-wave rectifiers?

Even in the lower end PSU's it's usually the switchers that go before the any of the diodes do, it does help a lot when they are standing up. The lowest end diodes I've seen were 1.5A each in a PSU that claimed 450W

[eccerr0r]

Re: What's the deal with full-wave rectifiers?

If the bridge has larger die in them to compensate for the thermal resistance of the larger package, that would mean that the overall cost would be higher for the silicon. Since both are rated for the same current rating, really it's only cost that separates the two.
I'm happy with whichever is designed properly, using a 1A bridge rectifier in a 300W PSU would get the same level of questions as one designed with four 1N4007's.

[999999999]

Re: What's the deal with full-wave rectifiers?

^ I didn't mean to imply they have larger dice at same current rating, actually bridges have a smaller die at same current rating when designed to be heatsunk, and rather that larger dice will fit in the space, and that would allow an even higher current rating though that rating (and derating) depends on heatsinking, airflow, ambient temp. The larger bridge package actually has a lower thermal resistance when you consider that the current rating is when used with a heatsink instead of only a case to air or leads to PCB thermal path.

You could compare something like a PBL405 constructed like the first pic in my prior post, with its avg current rating of 4A, then contrast it with the D15SB80 in the 2nd pic I posted which has a 15A avg current rating, but only if you keep the case temperature under ~ 100C, which mandates a heatsink in most applications.

PBL405 doesn't have any reasonable way to add one and that limits its average current even though it has a 150A peak current.

We aren't really comparing apples to apples though because usually when a manufacturer uses discrete diodes in a cheap PSU they are using parts spec'd for just enough current, less of a margin than that usually found when they use a (higher rated) bridge rectifier. D15SB80 is found in some ATX PSU and I don't mean 1000W units, I've pulled some out of sub-500W PSU.

Ultimately I like over-spec'd parts, so I would want the bridge rectifier over the discrete diodes. How often have you seen discrete 15A diodes in ATX PSU?

[eccerr0r]

Re: What's the deal with full-wave rectifiers?

It's up to the end consumer willing to pay for overdesigned products. As an engineer (but not a PSU engineer) I do worry about construction costs, it affects the bottom line (i.e. profit). Overdesigning with a overspecced bridge makes no sense when the switching transistors or other devices will always die first. I've yet to see a dead PSU where the input rectifier fried, so regardless of whether it's heatsinked or not, apparently most are designed well enough - 4-diodes or bridge - is easy to satisfy cooling requirements and doesn't matter which way it's done. That being said I've always hated PCB layout for 4-diode bridges and opt for a discrete bridge just to make things look neater.