Cheap PSU: Piece of junk or semi-decent?

[TELVM]

Re: Cheap PSU: Piece of junk or semi-decent?

... Very well done, sir! ...

Thanks a lot , coming from you that really means something for me . But there is no merit on my part, barely a month ago I was (even more than now ) PSU illiterate, and would have been unable of anything without learning in spades here in badcaps.net from masters like you .

[c_hegge]

Re: Cheap PSU: Piece of junk or semi-decent?

I have considered doing this myself. The only problem is I simply lack the storage space for any more PSUs. I already have a lot of spare repaired 300-400 Watters which were decent to start with (Antecs, a Bestec, a few Hipros, a few Deltas) which I just can't seem tro get rid of. Unfortunately, all too many people think "Why get a used 300W when I can get a new 700W for $20?"

[momaka]

Re: Cheap PSU: Piece of junk or semi-decent?

Guetto-modded a bit the 'TOOQ' thing. To add PI coils for +12V, +5V & +3.3V outputs in a PCB lacking any parking slots for them, I brainfarted what I call 'the inglourious basterdaughterboard':

Haha, nice! I always wanted to do something similar for a few PSUs that I have that don't have their PCB silkscreened for coils.

Very well done, sir! Your projects inspire .

[Behemot]

Re: Cheap PSU: Piece of junk or semi-decent?

Guetto-modded a bit the 'TOOQ' thing. To add PI coils for +12V, +5V & +3.3V outputs in a PCB lacking any parking slots for them, I brainfarted what I call 'the inglourious basterdaughterboard':

I kinda thought about something liek that from time to time…I definitelly like this

[TELVM]

Re: Cheap PSU: Piece of junk or semi-decent?

Guetto-modded a bit the 'TOOQ' thing. To add PI coils for +12V, +5V & +3.3V outputs in a PCB lacking any parking slots for them, I brainfarted what I call 'the inglourious basterdaughterboard':

Scrounged the PI coils from an old deceased PSU, and soldered them bridging the positive legs of each pair of caps (2200uF Pannys & Rubys now, replacing the 1000uF 'ASSIA' & 'ChungX' crap).

=>LARGE =>LARGE =>LARGE

Cut the traces under the PCB, so that the juice has to go up from the PCB through the + leg of first cap, then pass through the PI coil, then go down through the + leg of second cap back to the PCB downstream of the trace cut.

=>LARGE =>LARGE

Also added a small daughterboard for extra input filtering components, with two Y-caps, a common mode ferrite choke, and a ferrite bead for the main wires:

A MOV across L-N downstream of the fuse:

=>LARGE =>LARGE

DIY heatsink for the bridge rectifier:

=>LARGE =>LARGE

Top view of the thing once modified to the perpetrator's delight:

A potentiometer for 'digital' control of fan speed:

=>LARGE =>LARGE

The thing under test to discover in how many pieces it explodes:

But again in a most amazingly way the thing works , with the DMM reading coherent voltages.

A pity I lack the equipment to find if ripple & noise have improved or what.


Man do I love this shit .

[Wester547]

Re: Cheap PSU: Piece of junk or semi-decent?

The different main transformer windings for positive voltages are built atop each other. In one sense it looks like the +12V winding has taps for the +5V and +3.3V, but it's actually the other way around, the +12V and +3.3V come from windings that are part of, or connected to, the +5V winding. Because the voltage from the +5V winding is regulated, the cross-regulation of the +3.3V and +12V are improved.

The +3.3V could be derived in maybe 4 ways. There could be separate taps for the +3.3V, but that complicates the transformer - two extra O/P pins and sets of wires going out and back in the transformer body. I doubt it is done that way, possibly excepting server power supplies. The 3.3V could be linear regulated from the +5V, but that would waste a lot of power (1.7V x I - 17W wasted with a 10A output!)! A buck regulator could be used, would be pretty efficient, but would be a bit complex (the regulator circuit as well as suppressing the extra noise and interaction with the main regulator - synced or not synced with the main regulator?) and take up board space.

There isn't a second toroidal coil on the secondary of those Hipros momaka speaks of, though (the ones that use DC-DC conversion for +3.3V). Most of them have what seems to be an extra tap in the main transformer as part of the DC-DC conversion for the +3.3V rail..... but usually they have two +5V rectifiers: one for the +5V rail, one for the +3.3V rail.... so even though technically speaking +3.3V would still be generated from +5V via linear regulated DC-DC conversion and Hipro's own tap (that momaka estimates to be roughly 3.6V), since they have a separate rectifier for the MOSFET to regulate down (voltage wise), doesn't that mean less squandered power?

A mag-amp seems fine to me, but as far as I understand it, using DC-DC conversion means less noise on the +3.3V rail and better group-regulation, maybe...

[PeteS in CA]

Re: Cheap PSU: Piece of junk or semi-decent?

1 is for the common mode choke for the 12V, 5V, and -12V rails ...

Actually, a common mode inductor has 2 (or more) in-phase windings that are the same number of turns of the same wire gauge on the same core. While the windings of the inductor to which you refer are on the same core and wound in-phase, the windings are of different wire gauges and number of turns. It's called a coupled inductor. It saves PCB space (I know, Duh!), keeps the inductor continuous down to lighter loads, and if the windings are properly proportioned relative to each other cross-regulation is improved.

The different main transformer windings for positive voltages are built atop each other. In one sense it looks like the +12V winding has taps for the +5V and +3.3V, but it's actually the other way around, the +12V and +3.3V come from windings that are part of, or connected to, the +5V winding. Because the voltage from the +5V winding is regulated, the cross-regulation of the +3.3V and +12V are improved.

The +3.3V could be derived in maybe 4 ways. There could be separate taps for the +3.3V, but that complicates the transformer - two extra O/P pins and sets of wires going out and back in the transformer body. I doubt it is done that way, possibly excepting server power supplies. The 3.3V could be linear regulated from the +5V, but that would waste a lot of power (1.7V x I - 17W wasted with a 10A output!)! A buck regulator could be used, would be pretty efficient, but would be a bit complex (the regulator circuit as well as suppressing the extra noise and interaction with the main regulator - synced or not synced with the main regulator?) and take up board space. A mag-amp is simpler, pretty efficient, and the board space required is reasonable, so that's the more common choice.

[goodpsusearch]

Re: Cheap PSU: Piece of junk or semi-decent?

Please feel free, it's very interesting and educative stuff .

Quoted for truth. Thanks momaka!!

[TELVM]

Re: Cheap PSU: Piece of junk or semi-decent?

... we're going off-topic, so I'll try to keep this post to a minimum...

Please feel free, it's very interesting and educative stuff .

[momaka]

Re: Cheap PSU: Piece of junk or semi-decent?

Well, I know we're going off-topic, so I'll try to keep this post to a minimum...

I think maybe 10A would be the maximum practical from a linear regulated 3.3V

I thought so too, but then I noticed a lot of the older 200-250W HiPro PSUs from OEM PCs use that and they are rated usually 14A or more. Those, however, do have a separate transformer tap for the 3.3V rail. I never measured it but I'm guessing it's probably just a bit higher than 3.5V.

Do you mean DC-DC conversion/linear regulation is less effective than the traditional 2-toroidal coil setup and schotty barrier rectifiers?

There's 2 ways to do DC-DC conversion in a PSU. One is with 2 MOSFETs in a synchronous buck DC-DC circuit, the other is with a single MOSFET in a linear regulated circuit. The linear is very inefficient but very simple to build. Buck is extremely efficient but complex and requires a buck PWM controller.
Synchronous rectification is a totally different animal, so I won't discuss it here. And yes, it's very efficient too.

[Wester547]

Re: Cheap PSU: Piece of junk or semi-decent?

I think maybe 10A would be the maximum practical from a linear regulated 3.3V, a low end motherboard would probably be OK with that (it usually only provides RAM and chipset power; both of which are usually linear regulated on the motherboard again... so almost all of the power is thrown away.)

Do you mean the amount that would be practically used by the +3.3V rail or literally the practical maximum DC-DC converted +3.3V (or linear regulated) would be good for? Some would contest that more than 15A is not drawn. You'd probably be right though if you consider the DC-DC converters on a motherboard and how many ICs essentially use even lower voltages than +3.3V because of that.

The MOSFET is used a linear regulator - efficiency would be even worse if they would take it from 12v. Why do you think that even the lowest end units use 40A parts there? They'd cheap out on those if they could. Highest i've seen so far was rated 80A.

Do you mean DC-DC conversion/linear regulation is less effective than the traditional 2-toroidal coil setup and schotty barrier rectifiers? I always thought a secondary circuit made up of MOSFETs (for DC-DC conversion, considering the limitations of group-regulated power supplies) instead of diode packages would be more effective (synchronous rectification in place of diodes).

[Behemot]

Re: Cheap PSU: Piece of junk or semi-decent?

That's how any regulated SMPS works.

I mean refference like, for example, http://pdf1.alldatasheet.com/datashe...ICS/LM123.html. The refference component, not using comparator against refference voltage

[tom66]

Re: Cheap PSU: Piece of junk or semi-decent?

Most of the time it is actually magamp regulated, not linear regulated. But it almost always uses the same transformer tap.

I think maybe 10A would be the maximum practical from a linear regulated 3.3V, a low end motherboard would probably be OK with that (it usually only provides RAM and chipset power; both of which are usually linear regulated on the motherboard again... so almost all of the power is thrown away.)

[Th3_uN1Qu3]

Re: Cheap PSU: Piece of junk or semi-decent?

I have never seen it in my life yet and only once here on pictures. Also getting it from +5 V is pure insanity, efficency is terrible. Better to get it from +12 V.

The MOSFET is used a linear regulator - efficiency would be even worse if they would take it from 12v. Why do you think that even the lowest end units use 40A parts there? They'd cheap out on those if they could. Highest i've seen so far was rated 80A.

More common is to get +5 V SB using refference from some higher voltage from small transformer.

That's how any regulated SMPS works.

[Behemot]

Re: Cheap PSU: Piece of junk or semi-decent?

I have never seen it in my life yet and only once here on pictures. Also getting it from +5 V is pure insanity, efficency is terrible. Better to get it from +12 V.

More common is to get +5 V SB using refference from some higher voltage from small transformer.

[goodpsusearch]

Re: Cheap PSU: Piece of junk or semi-decent?

No, he is right. If I get it correct he talks about using a mosfet to get 3.3V. Many cheap psus use that solution because they save some cents not having to use some parts and the result is cleaner 3.3V, but lower in A.

[Behemot]

Re: Cheap PSU: Piece of junk or semi-decent?

I gess you are out of picture. I thoguh we are talking about cheap chineese craps here. So how the hell got things like „a totally different desing to get their 3.3V rail“ here? You are way of price region we are in.

[momaka]

Re: Cheap PSU: Piece of junk or semi-decent?

This. There are PSU without this one

The PSUs that don't have it use a totally different design to get their 3.3V rail. On such PSUs, there will also be no mag amp torroid - that's because there is no mag amp circuit. Most PSUs that don't have the 3.3V rail torroid use linear regulation to get the 3.3V rail and most of these designs will actually have even cleaner 3.3V rail with much less noise.
So just because there is a torroid for the 3.3V rail doesn't mean the PSU will have cleaner output on the 3.3V rail.

[Behemot]

Re: Cheap PSU: Piece of junk or semi-decent?

1 is for the 3.3V rail

This. There are PSU without this one When it's present, it greatly improves ripple. So I personally take it as part of the filtration.

[momaka]

Re: Cheap PSU: Piece of junk or semi-decent?

Well it's not typical, but still SOME output filter in there. See the big toroid coil? I bet it sits on +3,3 V.

Which torroid are you referring to? There are 3, but they are all required for proper operation of the PSU. 1 is for the common mode choke for the 12V, 5V, and -12V rails, 1 is for the 3.3V rail, and the last small one between the 2 heat sinks is used for deriving the 3.3V rail from the 5V rail output pins on the transformer (the mag amp circuit).

The PI coils/bobbins, on the other hand, are not required but they can improve the ripple and noise filtering quite a bit.

Piece of cake then

I think I foresee more ghetto PSU modding here