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

[goodpsusearch]

ATNG = Deer = L&C = Force = Premier = T&P Ans = Turbo X = Allied = Solytech = Rosewill (some)

[c_hegge]

^
ATNG and deer/solytech are two completely different companies

[ben7]

Originally Posted by Behemot

I am not sure what the functional difference between EI, ERL, EL, EE etc. cores is, if any, nobody really answered my questions some time ago when I started a thread for that. So I take it it is not as from my experience, all the transformers with tens of different labels look and work the same.

So the only thing we know for sure is the core diameter which the number represents. So the question is whether goodpsusearch measured the core and knows it is fake label, or did not and only thinks something. Sure it is also quite low so it is definitelly small, but it can still be low and real 35mm size.

To be honest, I don't think many people actually know what the differences are. I could probably do some google magic and find something.

[Dan81]

Originally Posted by goodpsusearch

ATNG = Deer = L&C = Force = Premier = T&P Ans = Turbo X = Allied = Solytech = Rosewill (some)

Actually:

ATNG - different company as Solytech
Deer=L&C
Force = Ritmo (may be something else)
Premier = either L&C or ATNG (most available here are L&C tho)
T&P ANS - clearly Deer (seen one,also there's T&P-CIC)
Turbo X - dunno (are they the ones behind Ultra X-Finity?)
Allied - high end Deers (reliable,MSI even rebranded a 500W model)
Solytech - Deer's new company,started picking up though
Rosewill - Deer for Stallion series and ATNG for anything else

[Behemot]

I think you are somewhat messing brand names and OEM suppliers/manufacturers. Some of the brands are the manufacturers own but some are most likely independent companies who just use ATNG, Deer and possibly some other OEM suppliers.

[goodpsusearch]

Originally Posted by Dan81

Actually:


Force = Ritmo (may be something else)

http://www.e-shop.gr/images/PER/BIG/PER.700161.jpg

Force = Deer

[Pentium4]

It's hard to make a list. Deer are behind so many random no name one batch PSU's from the early 2000's

[Dan81]

Originally Posted by goodpsusearch

http://www.e-shop.gr/images/PER/BIG/PER.700161.jpg

Force = Deer

Oh.

The one Google turned up was a Ritmo.

[momaka]

Originally Posted by Behemot

I am not sure what the functional difference between EI, ERL, EL, EE etc. cores is, if any, nobody really answered my questions some time ago when I started a thread for that. So I take it it is not as from my experience, all the transformers with tens of different labels look and work the same.

IIRC, The Unique here did a quick "btw" explanation of the differences between EE and EI cores - but this was in a thread many years back. (I could find it for you if you really want, but I have to take a shit in about 10-ish minutes or so ).

Originally Posted by Behemot

So the only thing we know for sure is the core diameter which the number represents. So the question is whether goodpsusearch measured the core and knows it is fake label, or did not and only thinks something. Sure it is also quite low so it is definitelly small, but it can still be low and real 35mm size.

I have never fully taken one of those "fake" 35-core transformers apart, but I can tell you that the top "beams" of the core do indeed appear thinner (I think there was a website that did a breakdown of one, but again, that was a super long time ago - more than several years when I read it). Also, you have to take something else into account. A larger transformer can have more turns of wire and/or thicker turns of wire. So essentially, even if those "fake" 35 transformers had the same core size as "real" 35-core transformers, the thinner wires inside them would simply not allow the same amount of output power (or they will run much hotter when you try to get the same amount of power out of them).
So in laymens terms: the size of the transformer does matter - especially when you are dealing with old H-bridge topologies which are all known to run in the <50 KHz switching frequencies (if even that).

Now, you could still get a lot of power out of those "fake" 35 transformers (much more than the "usual" 300W) - but only if you modify it accordingly. Again, look up some of Unique's older threads on the forum. In one or two of them, he shows how to make your own switching PSUs, and in another he has all sorts of transformer calculators and whatnot.

Originally Posted by Behemot

I think you are somewhat messing brand names and OEM suppliers/manufacturers.

+1

Solytech and ATNG are totally different PSU manufacturers (although some of their older H-bridge designs do appear very similar).

Solytech = OEM/parent company behind Deer, Allied, and L&C, although this is a name that came up later on. Originally, it was only Deer IIRC.
- L&C -> bottom of the barrel cheap PSUs from Solytech. Extremely crappy and re-branded by many many "no-name" small brands.
- Deer -> "mid-grade" budget line from Solytech.
- Allied -> "top of the line" from Solytech, but again for budget PSUs.
Solytech also makes more decent PSUs than the above that do not go under these brands.

Rosewill = IIRC, NewEgg's house brand. Made from variety of manufacturers, including Solytech and ATNG.

Premiere, Turbo X, Force, ANS, JNC, and many others = all small companies that simply buy cheap PSUs and re-brand them as their own.

[Behemot]

What I mean is the "35 size" does not mean anything about the core thicknes, material, height or winding. It just states it is 35 mm wide. That's all you get. But than you may still have tens of 35mm core variants with 50/100 % difference in power they can provide, I am not implying otherwise.

[momaka]

Well, you may as well be right from a scientific point of view...
But from my personal experience all I can tell you is this: if the PSU looks even a slight bit goofy and has a shorter "35-size" transformer, I know the label on it is probably a lie.

[mariushm]

In one of the Premier (Deer) power supplies I owned, the sticker on the transformer said ERL-35 but after I desoldered it, on the bottom where you usually don't see (between transformer and pcb) was a sticker saying ERL-33.

So just like with primary capacitors (with fake sleeves saying a capacitance one step higher than real one), often those cheap power supplies have undersized transformers, the horrible pair of diodes solder to metal strip instead of actual to-220 diode pair, cheap 1-2A through hole diodes for bridge rectifer etc...

[c_hegge]

^
I've seen that too, with the EI-33 sticker on the bottom of thr transformer, but an ERL-35 on top.

[Dan81]

Originally Posted by mariushm

In one of the Premier (Deer) power supplies I owned, the sticker on the transformer said ERL-35 but after I desoldered it, on the bottom where you usually don't see (between transformer and pcb) was a sticker saying ERL-33.

So just like with primary capacitors (with fake sleeves saying a capacitance one step higher than real one), often those cheap power supplies have undersized transformers, the horrible pair of diodes solder to metal strip instead of actual to-220 diode pair, cheap 1-2A through hole diodes for bridge rectifer etc...

The 2003-based Deers (e.g. my new ANS LC-B350ATX which I'm modding) say EI-33ASG. I've seen EI-28ASG in a 250W JNC (also Deer) though.

It's the newer 2005Z PWM chip ones that they started to label EI-33 trafos with ERL-35-2005.

Exception to above are Allieds and all Deer clones based on Allieds.Those ones have LT3500 something,and use real 35 size trafos,or real 33 size trafos with the usual EI-33ASG marking.

This is what I noticed when I was working with Deers.

[Behemot]

High-voltage capacitors are question of its own. My ESR Micro is measuring like 85 % of labeled capacity even for brand new Chemi-Con or Rubycon caps in power supplies. Constantly for two years. Just checked with CT-Micro, completelly different device, and the same result.

It may be just that most of the ESR/capacitance meters in this price segment all measure with charging to like 0.1 V. That may be too low for these 200V+ snap-in caps. So for the cheaper ones, their real capacitance may already be <90 % nominal and it measures even 15 % less and you think it is already out of spec, one step down. Would be good to confirm this with some professional high-end meter?

[momaka]

Originally Posted by Behemot

High-voltage capacitors are question of its own. My ESR Micro is measuring like 85 % of labeled capacity even for brand new Chemi-Con or Rubycon caps in power supplies. Constantly for two years. Just checked with CT-Micro, completelly different device, and the same result.

It may be just that most of the ESR/capacitance meters in this price segment all measure with charging to like 0.1 V. That may be too low for these 200V+ snap-in caps. So for the cheaper ones, their real capacitance may already be <90 % nominal and it measures even 15 % less and you think it is already out of spec, one step down. Would be good to confirm this with some professional high-end meter?

Good catch.
I checked a bunch of high voltage caps a while back and tabulated some results. Post here (see attachment at end of post):
http://www.badcaps.net/forum/showpos...5&postcount=57

However, it appears that not all of the HV caps had less that stated capacity - some were spot on spec or close. Notice it's the really shitty brands that were way off.

[Behemot]

Yeah but it is plus or minus 20 %. They may just have higher capacitance than nominal so while reading -15 %, you end about the nominal.

I've too run into some old quality caps which had sometimes even over spec…but that may very well mean they are already leaking internally as well as they are fine and their oil electrolyte just holded the high capacity for all those years. But now everything I have measured in new power supplies( which were never used) in last two years always was around 85 % nominal. All of them.

It is also possible that 400V capacitors are much more off than 200V? But it is only snap-in caps, as for leaded KXJ, they are all on nominal capacity. I usually measure in circuit but in a few occasions, I removed them and measured the same (+- like 5 uF).

[momaka]

Originally Posted by Behemot

Yeah but it is plus or minus 20 %. They may just have higher capacitance than nominal so while reading -15 %, you end about the nominal.

Right. I just wanted you to note the really crappy brands like KYS and MK on my chart - those things are like half of the stated capacity on their label. And quite high ESR too.

I wonder if this will work as a test:
put the HV caps in series with a really high resistance resistor and see how much they have charged after 5 time constants. In theory, after 5 time constants they should be charged to 90% of the supply voltage (IIRC). Of course the resistance shouldn't be way too high because then the internal leakage current of the caps could affect the measurement.

According to Digikey's RC time constant calculator:
http://www.digikey.com/en/resources/...-time-constant
...it should take about 32 seconds to charge a 680 uF cap in series with a 47 KOhm resistor. Hopefully the 47 KOhm resistance isn't too high to be affected by the internal leakage current of the cap, though. I don't know, what do you think of this test?

[Behemot]

That may do. If the caps also read high ESR than sure, they are bad, no discussion here. My problem is with new Chemi-Cons, Rubycons etc.

[Behemot]

Elanpower 1U power supply. The sticker already rings a bell though not exactly same as usually and the UL number leads nowhere.

But upon opening it is quite sure this is FSP Group origin - SPI transformers, bloated 8mm crapxons KF. Strange thing though it uses CM6800G and some Siicon Touch 16pin monitor (it is glued in silicon) while FSP usually has custom ICs.

Should be no problem fixin' with some Chemi-Cons KZH 1000/16 and KZN 1500/6.3.