Power supply build quality pictorial. part 2

[ncovert]

Re: Power supply build quality pictorial. part 2

I just happened to notice how this thread has recently had a lot of people talking about power supplies (I will admit I have done so too ) instead of posting inside pics.
Shouldn't we just discuss our power supplies in a separate thread and then use this one more for its original intended purpose of posting pictures of the inside of PSU's? Just a little something to think about...

[Shocker]

Re: Power supply build quality pictorial. part 2

Already been done, ages ago

[tom66]

FSP Group FORTON/SOURCE FSP350-60

Got three PSUs today. I'll put them in separate posts.

FSP Group FORTON/SOURCE FSP350-60
350W rated. 3.3V @ 28A, 5.0V @ 30A, 12V @ 19.5A, 5Vsb @ 2A.

This is a rather heavy, well built unit however it only intermittently powers up. Do you want to know why? It has one bulging Jamicon, and one bulging Teapo. When it does power up, the 12V ripple is just a tad high, at 1.8Vp-p (not a typo.) Amazingly, the 12V still measures 11.7V. The 5V ripple is fine and the 5V is a tad high at 5.22V (could be due to bad caps -- cross-loading?), but still close enough. It'll usually run for about a second, then die, or sometimes it will start and run fine after playing around with PSON enough.

It uses passive PFC, which is not a bad idea (cheaper, lower component cost, generally higher efficiency, lower operating cost to customer), but the cap choices are quite poor -- Teapo, OST, Jamicon, etc.

It has a speed controlled Yate Loon fan. No idea if this is any good?

Admittedly, I'd always somewhat considered Jamicon to be above-average general purpose caps, and OK if not abused too much, but I'm starting to regret that decision.

First scope shot shows 12V on DC coupling, second shot is a closer look at 12V ripple, third shot is 5V startup (fine.)

[Behemot]

Re: Power supply build quality pictorial. part 2

It is very common. Most - but not all - of the older versions with 80mm fan had good caps even after like 5 years when I got them. These new 12cm versions usually have bad caps after like 3-4 years and they do not start with load or even not at all. They are also more sensitive to small caps parameters. I think they are just trying to push maximum of that old platform (because it haven't really changed much for 8 years, probably more) - power and efficiency wise - so than it hase tighter mantinels and is sensitive to basically anything.

After recap you still have pretty decent unit though.

[Wester547]

Re: Power supply build quality pictorial. part 2

^ I always thought power supplies with 140mm/120mm fans as opposed to 80mm fans cooled better and had larger real estate/surface area to do so. Is there something about those style of PSUs that is less efficient than that of the old in that regard? That FSP doesn't look horrible but I detest how it has conductive glue like every other FSP I've seen. I would say that Teapo, Jamicon (maybe, though I wouldn't bet on it - their parent company - Kaimi - is also Asiacon/Hermei's parent company), and OST could last a while if they aren't stressed and aren't running hot, in that application anyway. Truth be told, though, because of lackluster heatsinks and bad choice of capacitors (and the aforementioned glue), FSP power supplies are bounded to fail sooner or later if they're used as they came from the factory. Also, a temperature-controlled, Yate Loon fan would be fine if it's not a low speed fan or a sleeve bearing fan (if this is an 120mm fan).

[Behemot]

Re: Power supply build quality pictorial. part 2

Fans are temperature controlled as far as my memory goes in FSP PSU's. The thing pretty much is those old supplies were „built like tanks“ as somebody likes to use here and usually got lower power.

Now, you make them not so heavily built, use even cheaper caps (used to be OST/Teapo, now Jamicon, it's even worse crap) but on the other hand you want higher output power and efficiency from them. It has to have some affect, yes? Also remember, if they make it through warranty, it's enough, right?

Also generally FSP tends to use smaller heatsinks as the efficiency goes up. They can save material and stuff. So even with higher efficiency, their PSU's are still as loud as they have been couple years ago. They have to provide the same airflow both for the small heatsinks and crappy caps.

[tom66]

Re: Power supply build quality pictorial. part 2

But wait! There's more!

FSP Group FSP300-60THA
300W rated. 3.3V @ 20A, 5V @ 26A, 12V1 @ 8A, 12V2 @ 14A, 5Vsb @ 2A

This unit appears to work perfectly, however, it is full of Teapo and one large CapXon. The primaries are OST. In general, the build quality is quite good. It is cooled using a single 80mm fan and uses passive PFC, like the other FSP PSU.

No caps are bulging, and it does start, but I've not put it under any load yet. I don't have a sacrificial computer.

Passive PFC inductor is bolted to the removable half of the case (on the 350W it was bolted to the non-removable half), and looks very similar to the 350W model.

It also uses a fan speed controller but a newer updated version using an LM393 and lots of SMD. The controller in the 350W model used an LM358 and all through hole. It looks to be doing PWM speed control, a good solution.

5th photo is supposed to show DM311 controller.

Primaries are 680uF 200V. In the 350W model they were 820uF 200V. Note that output power in watts = ~1/2 capacitance in uF.

[Behemot]

Re: Power supply build quality pictorial. part 2

Output power equals the capacitance directly! The two caps are in series, that means their total capacitance is 340 uF.

It's typical THA, THN, GLN and stuff. They got like dozen of versions, many of them OEM, but they shared more or less the same design. Than they upgraded it with 120mm fan and the rest you already know

[tom66]

Re: Power supply build quality pictorial. part 2

Output power equals the capacitance directly! The two caps are in series, that means their total capacitance is 340 uF.

Yeah, I know. I've made the point before, but I'm trying to simplify it. The equation doesn't apply for PFC due to the different topology and it's very approximate. But it puts things like L&C to even more shame, with 350W output from fake 330uF in series (actually 220uF), so just 110uF to play with...

It's typical THA, THN, GLN and stuff. They got like dozen of versions, many of them OEM, but they shared more or less the same design. Than they upgraded it with 120mm fan and the ret you ylaready know

Interesting. Apart from the caps they're very well built units, I wouldn't hesitate to put them in a low power build of my own.

Coming up (tomorrow, likely): Dell 170W built by Delta. I see a Nichicon through the vent... this might be good.

[Behemot]

Re: Power supply build quality pictorial. part 2

Why not? Passive FC is in there just as balast to improve power factor the same way capacitor is present in big AC motors.

[tom66]

Re: Power supply build quality pictorial. part 2

Why not? Passive FC is in there just as balast to improve power factor the same way capacitor is present in big AC motors.

It works for passive PFC, what I meant is it's not applicable for active PFC.

[Behemot]

Re: Power supply build quality pictorial. part 2

U wrote just PFC

In active PFC it is better to use directly high-voltage cap. You can still use two 200V cap and I would bet it still works the same - has to - the voltage will spread accross them, resulting in total capacitance dropping to half.

[tom66]

Re: Power supply build quality pictorial. part 2

Typo/accidental omission on my part I always meant active PFC, I promise .

I have a PSU in my desktop that uses two series caps but also active PFC. It is 500W rated.

[Behemot]

Re: Power supply build quality pictorial. part 2

Are you sure they are not in paralel?

[ben7]

Re: Power supply build quality pictorial. part 2

WOW, that jamicon is VERY cooked! Look how the plastic wrapper has shrunken! I bet the control loop in that PSU is so stable, it ran fairly well with that badcap there, and the badcap just kept cooking!

[tom66]

Re: Power supply build quality pictorial. part 2

Are you sure they are not in paralel?

They're 200V, so yes, they are in series.

[tom66]

Re: Power supply build quality pictorial. part 2

WOW, that jamicon is VERY cooked! Look how the plastic wrapper has shrunken! I bet the control loop in that PSU is so stable, it ran fairly well with that badcap there, and the badcap just kept cooking!

Actually, it's an odd one with no wrapper - the label is just printed on it. In colour, no less.

[ben7]

Re: Power supply build quality pictorial. part 2

Actually, it's an odd one with no wrapper - the label is just printed on it. In colour, no less.

Why you say it has no wrapper?

I can clearly see the edge of the plastic wrapper.

If it really does have no wrapper, than, that is EXTREMELY odd!!!!!

[tom66]

Re: Power supply build quality pictorial. part 2

Why you say it has no wrapper?

I can clearly see the edge of the plastic wrapper.

If it really does have no wrapper, than, that is EXTREMELY odd!!!!!

Maybe I'm getting mixed up, you're probably right, I'll have another look tomorrow.

[ben7]

Re: Power supply build quality pictorial. part 2

Woo! 1000th post in this thread!

...

Ok, I'm finished