the gutless, bloated, and fried power supply hall of shame

[nick122]

Re: Power supply build quality pictorial. part 2

I got this psu for free. It's a Deer (lc-b300atx). I decided to crank it open and see what's inside. Anddd oh dear... (pic 1) Of course, there's no input filtering at all, undersized heatsinks, small transformer and undersized input caps (330uf 200v. The primary switchers are, of course, 13007s. The four diode treatment (4* 1a, 100v). Now the secondary side. And wow. Look at that discoloration. The 12v rectifier are 2 3a ultra fast recovery diodes. The 3.3v and 5v rectifiers are sr1040ct (10a 40v). 3.3v pi coil is toasted. I removed the 5/12v coil because it was toasted as well. No inductors on the secondary. The secondary side is populated with g luxons and jun fus (not kung fu). None of them are bulging or leaking and all are good on esr (interesting through). The 5vsb rail has a 470uf 16v g luxon. Of course, the old, crappy 2 transistor forward. Definitely the worst psu i've ever seen.

[PeteS in CA]

Re: the gutless, bloated, and fried power supply hall of shame

Intact, that POS might have been good (using the word generously) for 200W. There is nothing in it that is adequate for 300W. It was also unsafe. Those two tan ceramic disc capacitors near the AC wire in used as Y-caps, but aren't safety agency approved. The O/P inductor probably got so hot for so long that the magnetic properties of its powdered iron core are probably badly deteriorated.

[Dannyx]

Re: Power supply build quality pictorial. part 2

I got this psu for free. It's a Deer (lc-b300atx). I decided to crank it open and see what's inside. Anddd oh dear... (pic 1) Of course, there's no input filtering at all, undersized heatsinks, small transformer and undersized input caps (330uf 200v. The primary switchers are, of course, 13007s. The four diode treatment (4* 1a, 100v). Now the secondary side. And wow. Look at that discoloration. The 12v rectifier are 2 3a ultra fast recovery diodes. The 3.3v and 5v rectifiers are sr1040ct (10a 40v). 3.3v pi coil is toasted. I removed the 5/12v coil because it was toasted as well. No inductors on the secondary. The secondary side is populated with g luxons and jun fus (not kung fu). None of them are bulging or leaking and all are good on esr (interesting through). The 5vsb rail has a 470uf 16v g luxon. Of course, the old, crappy 2 transistor forward. Definitely the worst psu i've ever seen.

I've seen a similar setup under a different brand name too - a "Premier" from what I recall. It too had those discrete rectifier diodes instead of a TO-220 package and lack of filtering...definitely rang a bell when I saw the pics - I'm still using that Premier to drive 6 meters worth of RGB LEDs around my house

[momaka]

Re: Power supply build quality pictorial. part 2

I got this psu for free. It's a Deer (lc-b300atx).

I think I just had a deja vu moment. Oh wait, I know why!

https://www.badcaps.net/forum/showpost.php?p=224302&postcount=491

The four diode treatment (4* 1a, 100v).

Actually, it looks like those diodes are RL205, which are general purpose didoes capable of 2 Amps at Ta=75°C. Even with those and at 120 V AC line input, they would still probably be the last thing to fail. I had a JNC PSU with the same diodes, and after its 5VSB committed suicide, I used those diodes along with the 13007 BJTs to fix a 12V halogen lamp PSU that had its bridge rectifier and 13007 go nuclear for whatever reason.

Now the secondary side. And wow. Look at that discoloration. The 12v rectifier are 2 3a ultra fast recovery diodes.

Haha, you got DoaB-ed (Diodes on a Bracket)
(and I just coined a new term here )

3.3v pi coil is toasted.

You know why?

Because that spot calls for a much larger toroidal inductor, not a tiny PI coil. Given that the 5V/12V coil also burned along with the FR diodes on the 12V rail, looks like this was probably in a somewhat more power-hungry system, possibly with a high-end GPU. Perhaps an Athlon XP (heavy 5V rail use) and a modern 12V-heavy AGP video card. Or maybe the other way around... a Pentium 4 CPU (12V-heavy) and a Radeon 9700/9800 video card (3.3V/5V heavy).

No inductors on the secondary.

Deer's call: Ripple suppression? We don't need no ripple suppression! Look, the PSU still works perfectly fine.

The secondary side is populated with g luxons and jun fus (not kung fu). None of them are bulging or leaking and all are good on esr (interesting through).

Lol, go figure why. I've seen Deer/L&C PSUs with much better caps, and they all bulged. In this one, the caps were clearly over-stressed... and they are fine?

The 5vsb rail has a 470uf 16v g luxon. Of course, the old, crappy 2 transistor forward.

You mean 2-transistor flyback. If it was 2-transistor forward, I would have used the 5VSB to power the whole computer instead.

Definitely the worst psu i've ever seen.

Yeah, it's definitely high on that list. Though these are some note-worthy opponents, IMO :

https://www.badcaps.net/forum/showthread.php?t=6341&highlight=echo+star

https://www.badcaps.net/forum/showthread.php?t=39222&highlight=cyberlink

[nick122]

Re: the gutless, bloated, and fried power supply hall of shame

At least it looks much better now

[Dannyx]

Re: the gutless, bloated, and fried power supply hall of shame

At least it looks much better now

Is that spray paint ? Good waste of spray paint

[Behemot]

Re: the gutless, bloated, and fried power supply hall of shame

Oh Deer…shoot it on sight lol Haven't though I'd see another diode treatment ever again. Guess they are still around…

[momaka]

Re: the gutless, bloated, and fried power supply hall of shame

At least it looks much better now

It's a museum piece now.

[PeteS in CA]

Re: the gutless, bloated, and fried power supply hall of shame

Deer's call: Ripple suppression? We don't need no ripple suppression! Look, the PSU still works perfectly fine.

I'm pretty sure what folks call "Pi coils" are mostly to suppress spike noise rather than switch frequency ripple. If that is the purpose, the nH of inductance in the output wires could suppress some of the spikes the "Pi coils'" few uH suppress.

Back in ancient times Boschert used ~1/2 inch diameter powdered iron (-26 material) toroids with 2 or 3 turns instead of several turns on a ferrite rod. That was back when they built their P/Ss in Manteca, CA.

[Wester547]

Re: the gutless, bloated, and fried power supply hall of shame

I thought the other purpose those ferrite coils (or coils with an "air" core in some cases) serve is that of a voltage divider, but for the output caps. So the cap(s) before the coil handle most of the switching frequency ripple (or the pulses from the transformer and rectifiers) and the cap(s) after handle very little (although the magamp circuit or toroidal inductor before the output caps would handle the majority of the switching frequency ripple by comparison). It seems to be the consensus that caps and inductors that form a PI filter can better suppress high frequency ripple than simply using multiple caps in parallel or only one beefy cap, but I suppose it would depend upon the rest of the design.

[momaka]

Re: the gutless, bloated, and fried power supply hall of shame

I'm pretty sure what folks call "Pi coils" are mostly to suppress spike noise rather than switch frequency ripple. If that is the purpose, the nH of inductance in the output wires could suppress some of the spikes the "Pi coils'" few uH suppress.

Well, many PSUs that don't have them (and were designed to have them but they weren't installed) usually have much higher output ripple and noise. Typically, that's something the low-end junk PSUs found in this thread often do. So those "PI coils" certainly must be doing something useful.

I thought the other purpose those ferrite coils (or coils with an "air" core in some cases) serve is that of a voltage divider, but for the output caps.

It's not really a voltage divider as the voltage on the input and output of the circuit is the same. Technical term is a CLC low-pass filter. Actually, in the case of SMPS it's LC-LC low-pass (two inductor-capacitor low-pass filters chained). But technically speaking, you are correct that the cap before the "PI coil" (the first LC low-pass) handles most of the switching frequency filtering and the cap after it (the second LC low-pass) is there just to further curb the high-frequency ripple and noise.

That said, the self-resonant frequency of a filter circuit with and without the PI coil is going to be slightly different. This has to be considered when the rest of the PSU is designed (otherwise you can get nasty ringing on the outputs and/or the PSU filter may not filter the switching frequency that well). Thus, it is possible to make a PSU output filter without PI coils and still get really good ripple and noise suppression. But the PSU just needs to be designed for it (probably both the capacitor specs and the main toroidal inductor inductance will need to be altered). Also, the compensation may need to be tweaked too. And of course, the switching frequency plays a big role too.

Now that's about as far as my knowledge stretches. As for calculating values for any of the above stuff... I haven't that much a clue.

[PeteS in CA]

Re: the gutless, bloated, and fried power supply hall of shame

Well, many PSUs that don't have them (and were designed to have them but they weren't installed) usually have much higher output ripple and noise. Typically, that's something the low-end junk PSUs found in this thread often do. So those "PI coils" certainly must be doing something useful.


It's not really a voltage divider as the voltage on the input and output of the circuit is the same. Technical term is a CLC low-pass filter. Actually, in the case of SMPS it's LC-LC low-pass (two inductor-capacitor low-pass filters chained). But technically speaking, you are correct that the cap before the "PI coil" (the first LC low-pass) handles most of the switching frequency filtering and the cap after it (the second LC low-pass) is there just to further curb the high-frequency ripple and noise.

That said, the self-resonant frequency of a filter circuit with and without the PI coil is going to be slightly different. This has to be considered when the rest of the PSU is designed (otherwise you can get nasty ringing on the outputs and/or the PSU filter may not filter the switching frequency that well). Thus, it is possible to make a PSU output filter without PI coils and still get really good ripple and noise suppression. But the PSU just needs to be designed for it (probably both the capacitor specs and the main toroidal inductor inductance will need to be altered). Also, the compensation may need to be tweaked too. And of course, the switching frequency plays a big role too.

Now that's about as far as my knowledge stretches. As for calculating values for any of the above stuff... I haven't that much a clue.

I wasn't suggesting the inductors serve no useful purpose, just suggesting what the cheap-crap PSU sellers might be hoping. I doubt such outfits have the techno-chops to tweak their designs so as to get decent R & N out of a PI-coil-less circuit (and I don't either, but then I'm not out there designing and selling PSUs).

[Wester547]

Re: the gutless, bloated, and fried power supply hall of shame

It's not really a voltage divider as the voltage on the input and output of the circuit is the same. Technical term is a CLC low-pass filter. Actually, in the case of SMPS it's LC-LC low-pass (two inductor-capacitor low-pass filters chained). But technically speaking, you are correct that the cap before the "PI coil" (the first LC low-pass) handles most of the switching frequency filtering and the cap after it (the second LC low-pass) is there just to further curb the high-frequency ripple and noise.

Yes, voltage or potential divider was the wrong term. That type of output filter more or less "divides" the ripple current the caps have to handle, though. I always thought of the "PI filter" as a capacitor-input filter.

That said, the self-resonant frequency of a filter circuit with and without the PI coil is going to be slightly different. This has to be considered when the rest of the PSU is designed (otherwise you can get nasty ringing on the outputs and/or the PSU filter may not filter the switching frequency that well). Thus, it is possible to make a PSU output filter without PI coils and still get really good ripple and noise suppression. But the PSU just needs to be designed for it (probably both the capacitor specs and the main toroidal inductor inductance will need to be altered). Also, the compensation may need to be tweaked too.

I understand that the feedback loop and the output filter is generally designed for certain grades of capacitors and deviating from that design can either result in an overdamped or underdamped output filter rather than critically damped (which you also brought up some time ago). Is it also possible that certain PSUs are designed to have PI output filters on certain rails and just multiple caps and no coil on other rails? I've noticed that tendency with Delta/Newton PSUs. +5VSB, +3.3V, and +5V have PI output filters but the +12V rail almost never does. I'm not sure if I chalk that up to cheapness or just saving space (as those PSUs are not silkscreened for a coil on the +12V output). And it's pretty rare, but sometimes Hipro does that too.

And of course, the switching frequency plays a big role too.

Higher output frequency would probably mean that you can get away with smaller inductors, but it would require higher grade capacitors (not so much bulk capacitance although that can certainly be useful for managing transients).

[Dan81]

Re: the gutless, bloated, and fried power supply hall of shame

I guess it's my turn to post something...
What I have for you: a Raidmax RX-380K (i.e. a Sun Pro KY-480ATX).

outside: https://www.badcaps.net/forum/attach...1&d=1388363334

label: https://www.badcaps.net/forum/attach...1&d=1388363334

top side (taken a few years back): https://www.badcaps.net/forum/attach...1&d=1388363334

another top side (a little more recent - you can see another cap has failed in there just from the PSU sitting in storage): https://www.badcaps.net/forum/attach...1&d=1388363334

primary side showing diodes in bridge rectifier and the bent leads on the parts attached to the heat sink: https://www.badcaps.net/forum/attach...1&d=1388363334

secondary side shot: https://www.badcaps.net/forum/attach...1&d=1388363334

botton side (solder side): https://www.badcaps.net/forum/attach...1&d=1388363334

This is probably the crappiest power supply I have around. The case is made out of paper-thin steel. The heat sinks are equally anorexic. Soldering on the bottom side is plain awful. Lots of loose solder balls came out when I first opened it. It's a miracle it didn't short out at the factory. The primary transistors have their leads so bent that they are nearly touching each other! As you can also see from the bottom-side picture, the separation between traces on the primary is very small. I'm pretty sure this thing never passed UL, CE, or any other safety marks. Then again, it's clear from the label the safety marks were forged. The wires are all 20 AWG (with 22 AWG for floppy connector) and completely unlabeled! (not a single safety agency or UL number in sight)

380 watts output??? Yeah right . It may not be very clear from the pictures, but the PCB has some serious "burn-in", and this was not caused by a stuck fan since the fan is spinning fine (although, I should note that the fan was quite dry... looks like it never saw lubrication from the factory). Speaking of which, the fan is a Te Bao Metallic Plastic model M802512M rated for 12V and 0.14A. IIRC, this PSU was powering a mediocre Athlon XP 1600+ system with a GeForce 5200FX video card and 1 HDD, so nothing too heavy.

Primary/input side:
Not much to talk about input filter. We have us two blue (but fake!) Y caps, and a sugar-cane-colored input choke . No X cap. Better than nothing, still.
Moving on... two bigger and two smaller diodes for the bridge. Too lazy to read part numbers, but I'm guessing 3A and 1.5A or 2A diodes. The two bigger diodes are always used, so this PSU should be capable of pulling 300W from the wall with these diodes (of course it won't, as you'll see from the rest of the parts list).
Also on the primary: 2x 200V 330 uF Metacon GK primary caps that read 200 uF and 196 uF on an ESR Micro V4. Primary transistors are a pair of 13007 BJTs in half-bridge configuration. Main transformer is size 33. 5VSB transformer is the standard cute small stuff you always find in all cheap PSUs. Speaking of 5VSB - it's that good old friend, the 2-transistor design . It doesn't have a critical capacitor, though.

Secondary/output side:
12V rail has a 12A 200V fast recovery rectifier. 5V and 3.3V rails share a single 20A 40V schottky . The 3.3V rail is derived from 5V rail in a linear fashion with a CEP51A3 (48A, 30V, N-channel MOSFET).
Output caps:
12V rail: 1x brown CS "LOW ESR" 16V 1000 uF cap placed after a PI coil.
5V rail: 2x JEE "LOW ESR" 10V 1000 uF caps (1x before and 1x after PI coil)
3.3V rail: 1x JEE "LOW ESR" 10V 1000 uF cap after PI coil.
5VSB: not exactly sure. I think 2x CS 16V 470 uF caps originally (1x before and 1x after PI coil), but one of them failed, causing the 5VSB to crazy with anything higher than 50 mA of load. I think I saw the auxiliary rail for the PWM controller (an AZ7500BP) go as high as 30V when this was going on. I recapped the failing CS cap with a questionable STONE 16V 470 uF cap out of a Philips TV that had other STONE caps of the same ratings failed. But it's a crap PSU, so why waste good caps on it? I just wanted to get it working again, which the STONE cap did. 5VSB is good for up to 1.5A of load now, though it does get warm with that load. Before you jump and scream at me... don't worry - I'm not going to put this PSU back in a PC ever again .

Actually, the case from this power supply is now housing the guts of a beefy HiPro HP-P2507F3P 250W PSU that a friend gave me.
Sorry for that wall of text above. Hope you liked this one

Hm....

Something is wrong here - I just got a PSU with the same PCB (7700A transformer) - however the transformer itself isn't EI-33. I looked at the core and the size - and I was shocked to find it's a EI-40.

Any idea how much can a EI-40 do? Heatsinks are quite small but are thick enough.

[PeteS in CA]

Re: the gutless, bloated, and fried power supply hall of shame

Well the transformer core size is credible for 400-500W. With heatsinks, both thickness and surface are matter. Thickness matters for the thermal resistance within the heatsink - can the heat flow from the point of origin to the farthest fins. Surface area is important where exposure to cooling air is concerned.

What of the switch devices and rectifiers?