How can PC PSUs use such small transformers and have a BIG output?

[TheLaw]

Hi,

I have a question that has forever been bugging me. I'm in need of a bench PSU and I thought I'd build my own with a LM350 and a 4A transformer. Of course not the most elegant or efficient solution, but it's something feasible...

But that got me thinking: How is it that if I want a small 3A output, I need a pretty large transformer, while if you look at a modern PC power supply, they manage to put out 50A+, in an enclosure roughly the same size as it would take me to achieve a lowly 3A.

A.) How does this work? I don't know much about PC PSUs (as you can tell), but I've heard mention of "forward transistors" and stuff like that. How do they get such huge outputs out of little transformers. In fact, a 12V transformer for a PC PSU is probably half the size of the transformer I would use to get a fraction of the power.

B.) Can we take any of these modern PC PSU designs and use them in say a DIY power supply like the one I am attempting? Now if I wanted more power, I could use a switching regulator like an LM2576 or something, but I would STILL need a hugeass transformer. What gives?

Not sure who would know about this, but it's always something I've been curious about and I've never figured out.

Thanks.

[mockingbird]

Re: How can PC PSUs use such small transformers and have a BIG output?

It's because they are switching PSUs and not linear.

edit: Scratch that... I see you considered that already...

[TheLaw]

Re: How can PC PSUs use such small transformers and have a BIG output?

Well if I just used a switching regulator, I'd still need a big transformer, I think. Maybe I'm wrong...but I'm fairly certain.

Maybe those transformers work at different voltages/potentials. I'm not sure.

[PCBONEZ]

Re: How can PC PSUs use such small transformers and have a BIG output?

Higher frequencies need smaller transformers for the same energy throughput.

[PCBONEZ]

Re: How can PC PSUs use such small transformers and have a BIG output?

As PSU power has gone up the switching frequency has too.

Some of the ~1000w and up units switch at closer to 200kHz than 100kHz which is why the transformers are not much different in size.

I talked to a gent that worked on an SMPS for something on a satellite.
It's switching frequency was 1 Ghz to keep it small.
.

[Gordon01]

Re: How can PC PSUs use such small transformers and have a BIG output?

I talked to a gent that worked on an SMPS for something on a satellite.
It's switching frequency was 1 Ghz to keep it small.
.

1 GHz ??? Maybe 1MHz. There are no such powerful and high-frequency transistors.

[PCBONEZ]

Re: How can PC PSUs use such small transformers and have a BIG output?

No, 1Ghz.

There are no such powerful and high-frequency transistors.

That you or I can get.
Considering this place is public I'll just say it was a high tech military application and leave it at that.
.
.

[Toasty]

Re: How can PC PSUs use such small transformers and have a BIG output?

1 GHz ??? Maybe 1MHz. There are no such powerful and high-frequency transistors.

LOL!

Exactly what do you think RADAR, microwave communications, and the satellites put out?

.

[TheLaw]

Re: How can PC PSUs use such small transformers and have a BIG output?

Higher frequencies need smaller transformers for the same energy throughput.

So if I wanted to use an Lm2576 which is capable of a 3A output..how would I go about picking a transformer?

[Evil Lurker]

Re: How can PC PSUs use such small transformers and have a BIG output?

So if I wanted to use an Lm2576 which is capable of a 3A output..how would I go about picking a transformer?

AFAIK, the process occurrs on the first full moon of the month and involves a ritual of dancing around a fire, nudity, virgins, goats, soldering irons, and a 2000 page book of transformer theory.

Actually it aint too far from the truth. Core material, air gaps, switching frequency and all sorts of other parameters need to be taken into consideration when designing a transformer.

Basically the idear is to use a core material that will quickly turn itself into an electromagnet when electricity is applied and at the same time quickly turn said magnetic potential back into electricity when the magnetic field collapses as the power is turned off. Iron is physically incapable of this feat that is why powdered ferrite is pressed and formed into a ceramic like material for the cores. Different formulas exist for different frequencies etc so and so forth. If insufficient time between on/off cycles is not given the core is said to become saturated.

[Khron]

Re: How can PC PSUs use such small transformers and have a BIG output?

One small detail, though: the LM2576 is not a part that needs / works with a transformer...

[TheLaw]

Re: How can PC PSUs use such small transformers and have a BIG output?

AFAIK, the process occurrs on the first full moon of the month and involves a ritual of dancing around a fire, nudity, virgins, goats, soldering irons, and a 2000 page book of transformer theory.

Actually it aint too far from the truth. Core material, air gaps, switching frequency and all sorts of other parameters need to be taken into consideration when designing a transformer.

Basically the idear is to use a core material that will quickly turn itself into an electromagnet when electricity is applied and at the same time quickly turn said magnetic potential back into electricity when the magnetic field collapses as the power is turned off. Iron is physically incapable of this feat that is why powdered ferrite is pressed and formed into a ceramic like material for the cores. Different formulas exist for different frequencies etc so and so forth. If insufficient time between on/off cycles is not given the core is said to become saturated.

Shucks. I didn't think it was that complicated.

Alright. But for something like an LM2576, I can use a normal transformer, right? Like a transformer I'd use with a linear supply.

It would just be more efficient? So if it's a 3A switcher, I'd need a 3A+ transformer?

Thanks!

[PCBONEZ]

Re: How can PC PSUs use such small transformers and have a BIG output?

So if I wanted to use an Lm2576 which is capable of a 3A output..how would I go about picking a transformer?

I'm not the one to ask about something like that.

Off the top of my head Th3_uN1Qu3 probably knows the most about transformers here.
I know he knows enough to 'roll his own' from scratch anyway.
https://www.badcaps.net/forum/member.php?u=21626
.

No offense if there are other transformer gurus around that I couldn't think of.
Doesn't come up that often.
.

[PCBONEZ]

Re: How can PC PSUs use such small transformers and have a BIG output?

LOL!

Exactly what do you think RADAR, microwave communications, and the satellites put out?

.

Not to mention cell phones and WiFi.
.

[Th3_uN1Qu3]

Re: How can PC PSUs use such small transformers and have a BIG output?

Alright. But for something like an LM2576, I can use a normal transformer, right? Like a transformer I'd use with a linear supply.

Yes. It'll still be big and bulky because the LM2576 is a buck regulator and you still need a mains frequency transformer to provide isolation. On the other hand, being a switching regulator, it'll dissipate a lot less power compared to a LM317 or the like, so you can save on heatsinking.

[TheLaw]

Re: How can PC PSUs use such small transformers and have a BIG output?

Yes. It'll still be big and bulky because the LM2576 is a buck regulator and you still need a mains frequency transformer to provide isolation. On the other hand, being a switching regulator, it'll dissipate a lot less power compared to a LM317 or the like, so you can save on heatsinking.

Yeah thanks.

One more question.

Switching regulators are apparently more "efficient", as they put out less heat and the power wasted isn't dependent on the difference between input and output voltage.

Ok..

But if I still need a 3A transformer whether I want a switching design or a linear design, how is it more "efficient". Like LM2576 is supposed to be like 90%+ efficient. Cool. But how am I actually saving electricity?

Thanks.

[Perineum]

Re: How can PC PSUs use such small transformers and have a BIG output?

Losses as heat, I would presume?

[Th3_uN1Qu3]

Re: How can PC PSUs use such small transformers and have a BIG output?

If you want 3A out from the LM2576 you need a transformer capable of delivering 3A... Taking efficiency into account, 3.33A to be precise. For a linear regulator you'd need more than double that.

[TheLaw]

Re: How can PC PSUs use such small transformers and have a BIG output?

If you want 3A out from the LM2576 you need a transformer capable of delivering 3A... Taking efficiency into account, 3.33A to be precise. For a linear regulator you'd need more than double that.

Really? Well that was new news to me. Thanks.

[Th3_uN1Qu3]

Re: How can PC PSUs use such small transformers and have a BIG output?

Let me explain this in more detail. A series pass linear regulator has a fixed voltage drop of the power transistor that does the regulation. For simplicity's sake let's say this is 1 volt. This is called the dropout voltage - the input voltage needs to be at least this much higher than the output voltage for the regulator to work.

At 3 amps, we would have 3 watts wasted in the regulator just to keep this 1 volt dropout voltage. Now, let's say you have a 30 volt input and you want a 25 volt output at 3 amps. Now the regulator needs to drop 5 volts, which at the same 3 amps gives a dissipation of 15 watts! As the output voltage is lowered, things start getting even worse. If you want 5 volts out at the same 3 amps, you're gonna have to drop 25 volts in the regulator... which starts to get impractical already. This is where switching regulators start to shine. Linears still have the advantage of simplicity, and they have their place when the input-output voltage difference is small and the current demand is low.

Note that i said something stupid up there: Efficiency relates to power not to current. You can use the same 3A transformer for both linear and switching regs... the difference is how much of that power is actually put to good use.