OEM power supplies - how bad are they really? (feat HP proprietary 240W)

[manikyath]

i'll start by clearing a few things up:

1: i know nothing about the construction of power supplies or quality of components.

2: the reason i have this unit is because its broken, and had to order a replacement.

3: the unit did not damage any other components when it died, neither did it release any smoke, or odd smells.

4: i have taken the PSU apart before, more specificly to rip out the fuse to see if that was the issue - which it wasnt.

5: i'm hoping on some feedback from people smarter than me about what i found inside, see it as a crowdfunded review: i'm offering the device, and hoping smarter people add the knowledge i'm missing.

----

that cleared up: lets get started with specs.

 

input: 100-240V AC 4A 50-60Hz

i fished from the online spec sheet that its 90% efficient certified.

output: 240W max

+12Vmain 16A max

+12V cpu 16A max

-12V 0.15A max

+12Vsb 1.3A max

(this is where they reiterate total output cant be more than 240W, in english and the different flavours of asian characters)

 

something a few of you may notice:

"wheres the +5V and +3.3V, and why is it +12Vsb instead of +5Vsb?"

 

well... this thing only outputs 12 volts appareantly, and the rest of it is handled by the motherboard.

 

this brings us to the form factor: (spoilering all images for size reasons, i love you, australia)

i didnt understand the custom form factor, until i was called in for an emergency, where a mission critical machine at my parents' company was stone dead.

they gave me a spare machine to swap parts as needed, i proceeded to swap the power supplies as a test. turned out to be the right call, overall costing about a minute to get the power supply swapped, the machine ready to be back online.

ever carey holzman with his fancy screwdriver takes longer to just undo the screws on a standard power supply.

 

part of this concept also is the custom PSU connectors: (forgive me for blur, i am a techie, not a camera operator)

from left to right:

- a standard 4-pin cpu connector 2x brown cable, 2x black cable (12v 16A max)

- a proprietary 6-pin connector 2x yellow cable, 2x black cable, 1x purple(vsb) 1x blue (-12v) (12v 16A max, -12v 0.15A max, 12Vsb 1.3A max)

- a proprietary.. "control header" lets say. 6-pin, only 4 wires connected: grey, green (the one you ground to jumpstart), white and white w/ red stripe.

 

the power input is a regular ordinary power connector.

---

that brings us to the guts:

the PSU disassembles quite easily once you figure out how.

one of the first things about the inside i noticed was the fan, which isnt soldered to the board, but instead connected with a standard 4-pin PWM connector.

 

and a first look at the guts of this.. "creation"

 

disassembling further we find the underside of the pcb, where all the soldering happened: (once again, i'm too incompetent to handle a camera well)

 

another thing to note is a few components soldered straight to the power connector

from what i know of electronics: thats a pair of ceramic capacitors between the phases and ground, a resistor between the phases, and a "yellow box" carefully glued in place.

 

a thing to note is they took very good care of the ground lead, which was secured in place with a screw with one of those ridged washers.

 

further taking a look, this seems to be the side where they transform down to 12V, and rectify the voltage. (wild guess, from the bridge rectifier)

and at the other side we have some caps (everyone loves talking about the caps in PSUs, dont we?)

i cant help but notice the artfully placed globs of white glue all over the place.

 

another thing you folks may like to see in detail is the big cap in the center, which i've always wondered what they are for:

----

 

rounding up: i cant test what kind of power it outputs, because its a dead unit. but with it being a dead unit, theres something else to note:

all caps are okay, theres no obvious burn marks on the PCB, the fuse is okay. kinda makes me wonder what died.

repeating again: the PSU died inside a mission critical system, said system has been running fine for months since it had a replacement PSU, it was not damaged.

 

also should note that the company has a good douzen of these deployed, and its the first one to die. in fact, its the first hardware death they've had in a LONG time. and most of their systems are HP systems of vareous levels of proprietary and age.

----

the whole thing about different kinds of capacitors has made me think tho. if garbage like this can run for years straight in a hot enviroment, what do the crappy brands (corsair CX) and especially the chinese units do wrong that make them so "combustable"? this thing clearly doesnt have "all japanese solid state capacitors" inside, and all of them seem to be absolutely spotless.

 

i do realise this is only a 240 watt unit, and a €90 one at that...

----

so, this is where the crowdsourcing starts:

 

- how horrible are the caps?

- is it common practisce to have the fan connected via a header?

- does anyone have some insight on other components inside?

- should i bother with at least roughly reverse engineering the schematics?

- ...

 

if you have something noteworthy to add, i'll edit it in place with propper credit.

[STRMfrmXMN]

Many are made by Delta so they're actually good. Dell uses many of those. Lenovo not so much. Not sure about HP.

 

 

the whole thing about different kinds of capacitors has made me think tho. if garbage like this can run for years straight in a hot enviroment, what do the crappy brands (corsair CX) and especially the chinese units do wrong that make them so "combustable"? this thing clearly doesnt have "all japanese solid state capacitors" inside, and all of them seem to be absolutely spotless.

 

i do realise this is only a 240 watt unit, and a €90 one at that...

----

so, this is where the crowdsourcing starts:

 

- how horrible are the caps?

- is it common practisce to have the fan connected via a header?

- does anyone have some insight on other components inside?

- should i bother with at least roughly reverse engineering the schematics?

- ...

 

if you have something noteworthy to add, i'll edit it in place with propper credit.

Caps: Probably utter shit TBH.

Yes.

Caps are probably cheapened as is some soldering work but if it's 90% efficient then likely it was made to some sort of quality standard.

 

 

The Corsair CX-series has cheap capacitors among terrible soldering and generally mediocre quality of everything. It's also designed to restart itself in the event it gets too hot which, if it had a high enough heat tolerance would be OK, is pathetic on something with a 30C ambient temperature tolerance.

[Sakkura]

SamXon capacitors are mediocre, but not terrible. Better than eg. the Ltec capacitors in some of the early Corsair RM models. It's also important to note that capacitor lifespan depends hugely on the conditions - higher temperatures drastically reduce lifespan, and how much ripple they're subjected to also matters.

 

The fan header isn't too unusual.

 

To get more detailed info on the quality of the power supply you'd need clear pics of the diodes, transistors and control ICs in the unit, and get some expert opinion on it. I'm not seeing any major red flags though. Soldering quality looks good, I wouldn't be surprised if this was Delta.

[manikyath]

Many are made by Delta so they're actually good. Dell uses many of those. Lenovo not so much. Not sure about HP.

 

Caps are probably cheapened as is some soldering work but if it's 90% efficient then likely it was made to some sort of quality standard.

HP is actually considered very good quality in the buisiness world, potentially even more so than dell.

 

as for the quality standard, it has an "EPA 90% efficiency rating" as well as "TÜV Rheinland product safety" that basicly states it wont spontaniously combust.

[Mew]

HP is actually considered very good quality in the buisiness world, potentially even more so than dell.

 

as for the quality standard, it has an "EPA 90% efficiency rating" as well as "TÜV Rheinland product safety" that basicly states it wont spontaniously combust.

 

Yep, these companies screw around a little less on their business side. They couldn't care less about their consumer line. 

[iamdarkyoshi]

Message me in two hours. I can answer the questions here. At school right now

[thekeemo]

Message me in two hours. I can answer the questions here. At school right now

well its been 2 houra

[iamdarkyoshi]

- how horrible are the caps?

 

Not too bad, but they certainly arent top quality rubycon...

 

- is it common practisce to have the fan connected via a header?

 

Not too uncommon. However, using a PWM fan is something I havent seen before.

 

- does anyone have some insight on other components inside?

 

Basically, in a switchmode supply, the incoming mains goes through a filter (the bodge mounted parts on the mains connector) and then it goes into the bridge rectifier and into the main filter cap (rated at a high voltage) once the PSU has this roughly stable high voltage DC supply, this is when the magic begins. There are usually one or more controller ICs in charge of controlling mosfets, which pulse at a very high frequency into the transformer. The output of this is put into some high speed diodes to turn it back to DC, and then, more filtering, usually in the form of capacitors and inductors, which can "delay" a pulse of electricity for when the transformer is not outputting a pulse. After this, the controller IC monitors the voltage produced, and adjusts its control over the mosfet acordingly. The feedback from the output to the input has to be isolated, and this is usually done with an opto isolator. It is essentially an infared LED and receiver, providing a physical separation of the electrical signals. A similar ish concept can be seen with the VRMs in a motherboard or GPU, but they are usually multiphase, and the input and output are not isolated. They pulse in tandem, so there is less "off" time.

 

Correct me if anyone finds any wrong info here I am pretty tired...

 

- should i bother with at least roughly reverse engineering the schematics?

 

If you want. Perhaps just look at the major parts, such as the input or output, and draw a DaveCad drawing. They are really quite complex.

[comicsansms]

I got a HP that its PSU blew up. The PSU had the same factor form, but I don't think it was the same.

 

It took out the Motherboard, the HDD was fine, I haven't tested the IDE DVD drive or the CPU, so I don't know if they are okay.

 

I removed the Motherboard and I realized the PSU dying had melted the rubber standoffs and now I have standoff in my carpet!

[manikyath]

I got a HP that its PSU blew up. The PSU had the same factor form, but I don't think it was the same.

 

It took out the Motherboard, the HDD was fine, I haven't tested the IDE DVD drive or the CPU, so I don't know if they are okay.

 

I removed the Motherboard and I realized the PSU dying had melted the rubber standoffs and now I have standoff in my carpet!

the question is, what did you do to said device?

 

this thing was in a steaming hot enviroment (middle of a factory, small office, surrounded with junk blocking airflow, the factory roof works like a solar collector)

[comicsansms]

the question is, what did you do to said device?

 

this thing was in a steaming hot enviroment (middle of a factory, small office, surrounded with junk blocking airflow, the factory roof works like a solar collector)

I don't know where it was, I got it after it was broken.

I assume it was in a open plan office, but i could have been in the workshop area.

The label said "PC 1061 Broken PSU?"

 

I think it would have had good airflow, It also wasn't gunked up when I got it.