Power supply build quality pictorial. part 2

[Pentium4]

Saved proprietary Macron MPT-350D

Okay, so this is another one of the last projects I did at my last company. Thing got about 39,000 running hours on it. The customer brought this computer in and wanted general diagnostics run, even though everything was "fine." I was immediately interested by the chassis ( http://www.newegg.com/Product/Produc...82E16811119071 ) and once I looked at the back of the power supply I yelled "MACRON" I noticed that it was indeed a Macron once I looked at the label, and that it wasn't quite an ATX PSU....I immediately knew that the PSU had bad caps (I even bet my coworker $1000 on it, he did not accept. DAMN! ) but the computer already had a memtest running on it. I opened the top case, and WOW, the thing was fucking hot! Just the side chassis of the PSU was 42C! I guessed that it was powering a Prescott P4 because just by glancing at the board I knew it didn't support Core 2, and the early 2005 sticker on it suggested it wasn't a Pentium D. Once the memtest was done I powered it down and took the PSU out. The thing had been off for about a total of 6 minutes, and the secondary heatsink still measured 75C! The primary was about 50C. I let it cool down and took the heatsink off the CPU. It was indeed a Prescott 3.2GHz P4, 640 model. The thermal paste was so dry and got so hot that it actually started turning brown. The unit was also powering a single 7200RPM hard drive, one DVD drive, a Radeon X1650, 4 1GB DDR modules, two TV tuner cards, a wireless G PCI card, a PCI sound card, and two 0.30A exhaust fans running at full speed. After looking at the inside of the PSU only having a 16A ultra fast, I'm utterly amazed that the thing didn't blow up! And look how cooked that resistor got. I couldn't see the value so I decided not to replace it.

When pulling the secondary heatsink to replace the 16A ultra fast with a 30A schottky, the 16A rectifier partially split down the middle in the space where the three leads go in to the silicon, and you can see how the heat discolored it. The thing is a true champ Most of the rectifier screws were becoming loose so I tightened them.

Almost all of the caps tested bad. The only caps that didn't test bad were both 2200uF 6.3V G-Luxon LZ caps on the 3.3V rail (I don't know how this is possible) and the 1uF 50V caps on the secondary. The 10uF 50V GoldLink caps on the BJT circuit read ~2.5uF and over 40Ω ESR I replaced all of the caps with Japanese parts, and because of the high heat and running hours, I replaced the 1000uF G-Luxon caps on the input with barely used 820uF Asia'X. I also redid a lot of the soldering joints that looked like they got weak from the heat they endured.

The Superred fan span okay, but they covered the bearing so that I couldn't get to it. Because of this, I didn't take the risk and just replaced it with a Power Logic ball bearing fan Because of the ripple that the hard drive endured, I talked the customer into cloning their drive to a brand new 500GB one. Everything went successful. I only had an hour to work on this thing, so those are the things I did to it. But afterwards, the thing ran so much cooler! The side chassis of the PSU was barely lukewarm, and the air coming out of the back of the PSU was just warm instead of hot. Just another example of how tough these old Macron PSUs are. I don't think the customer realized what had to happen for their precious computer to maintain proper operation, but I sure felt good after the project was done.

[ben7]

Re: Power supply build quality pictorial. part 2

^ Replace the 10 ohm resistor, it looks cooked! It might be for the snubber on that rectifier, so that might be why the rectifier started to get really hot.

Edit: and replace the green film cap near it. It has turned dark green/brown, so yeah it's not had a happy life.

[momaka]

Re: Saved proprietary Macron MPT-350D

I was immediately interested by the chassis ( http://www.newegg.com/Product/Produc...82E16811119071 ) and once I looked at the back of the power supply I yelled "MACRON"

I guess you can't hide your love for those things. Well, me neither .

It's sad that this case/PSU got so many sub-par and negative reviews on newegg. The worst are reviews that say "Oh, its a 350W generic, so it must be crap". Then they go and buy a sub-$20 "non-generic" 500W+ Logysis or the like and think it is all good. Ugh.

Imagine how many of those have been thrown away now due to crappy caps. Makes me wish I could somehow get them and fix them. They are such wonderful PSUs after a recap.

I immediately knew that the PSU had bad caps (I even bet my coworker $1000 on it, he did not accept. DAMN! )

He probably knows you'd be right and it was best not to mess with you.

I opened the top case, and WOW, the thing was fucking hot! Just the side chassis of the PSU was 42C! I guessed that it was powering a Prescott P4 because just by glancing at the board I knew it didn't support Core 2, and the early 2005 sticker on it suggested it wasn't a Pentium D.

Well, good thing it wasn't a Pentium D. That would have made it run even hotter.

The thing had been off for about a total of 6 minutes, and the secondary heatsink still measured 75C! The primary was about 50C.

Dang. And those heatsinks actually looks pretty decent too. My MPT-301 has the really skinny ones. I can't imagine what would have happened if this Macron had those.

After looking at the inside of the PSU only having a 16A ultra fast, I'm utterly amazed that the thing didn't blow up! And look how cooked that resistor got. I couldn't see the value so I decided not to replace it.

Yeah, that 16A ultrafast can take quite a beating before letting go. On my MPT-301, the Supperred fan had completely seized. Yet, whoever had it still wrote on it "Fan does not run. Power out OK"

It's not visible from the photos I posted, but I had the same exact resistor and capacitor really darkened. I think ben7 is correct, it is a 10 Ohm resistor. As long as it's still in spec, it doesn't need to be replaced. Same goes for the green cap. Macron usually uses 103j (i.e. 0.01 uF 5%) polypropylene rather than ceramic - a nice touch IMO.
I've seen the same happen in an Allied PSU with those snubber resistors. This is usually caused by the output caps failing.

The Superred fan span okay, but they covered the bearing so that I couldn't get to it. Because of this, I didn't take the risk and just replaced it with a Power Logic ball bearing fan

I hope you saved that Superred fan. The one in my MPT-301 was sealed like that as well and it was completely seized - couldn't even turn it by hand. Had to drill the plastic on the back to get it taken apart. I did, and in the process, I destroyed the plastic c-washer that keeps the fan rotor blade assembly from pulling out. But it ran fine without it, so I left it as is. That was some 5 years ago or so. It still is running to this day. Not rattling or anything either. So Superred fans are good, IMO.

Everything went successful. I only had an hour to work on this thing, so those are the things I did to it.

1 hour. ONE hour? It takes me that long to scratch my ass. Then another one to realize that I did so I can stop.
Geez man, you're a true PSU wiz.

[Behemot]

Re: Power supply build quality pictorial. part 2

I share the same for different FSP Group units. Quite good too and mostly with bloated caps only.

[Pentium4]

Re: Power supply build quality pictorial. part 2

^ Replace the 10 ohm resistor, it looks cooked! It might be for the snubber on that rectifier, so that might be why the rectifier started to get really hot.

Edit: and replace the green film cap near it. It has turned dark green/brown, so yeah it's not had a happy life.

Ah, 10Ω. Well it's too late now, that unit has got a solid 1,150 more running hours on it by now. I'm no longer at that company I did the work for, but I specifically asked my coworker if this exact unit came back in, and he said no

He probably knows you'd be right and it was best not to mess with you.

Haha, I guess that says something though, because he had only been at the company for 3 days

I guess you can't hide your love for those things. Well, me neither .

It's sad that this case/PSU got so many sub-par and negative reviews on newegg. The worst are reviews that say "Oh, its a 350W generic, so it must be crap". Then they go and buy a sub-$20 "non-generic" 500W+ Logysis or the like and think it is all good. Ugh.

Imagine how many of those have been thrown away now due to crappy caps. Makes me wish I could somehow get them and fix them. They are such wonderful PSUs after a recap.

Yeah, I love these PSU's Can't hide it...I'm also in love with proper AC plugs And I know, a lot of those newegg users have no idea what they're talking about. But...what's new? It's still a shame that Macron insulted themselves by using such bottom of the barrel caps in their well designed units. They seemed to squeeze the most out of the ancient half bridge design. I guess your only option to save some is to check fleabay often, and check your local PC recycling stores.

Well, good thing it wasn't a Pentium D. That would have made it run even hotter.

No kidding! I'm amazed at how much punishing heat 90nm CPU's can handle though. I think this board uses lead solder as well, I don't think lead free solder could have survived, the BGA probably would have failed a long time ago.

Dang. And those heatsinks actually looks pretty decent too. My MPT-301 has the really skinny ones. I can't imagine what would have happened if this Macron had those.

Very true. But what I'm thinking is that the PSU was getting abnormally hot for a number of reasons:
1.) That 16A rectifier had to be overloaded or close to it its entire life. (Reducing efficiency in the process?)
2.) The rectifiers were becoming loose on the heatsink
3.) The poor cooling on the Pentium 4 right below the fan intake of the PSU was pulling extra heat inside the PSU.
4.) Pretty sure the failing caps would create more heat?

Because afterwards, the PSU ran about as cool as your MPT-301 powering a P4 system with no video card. So either it was extra heat from the CPU, or the efficiency was becoming abysmal to create so much heat! Seriously, the before and after heat difference was amazing.

Yeah, that 16A ultrafast can take quite a beating before letting go. On my MPT-301, the Supperred fan had completely seized. Yet, whoever had it still wrote on it "Fan does not run. Power out OK"

Oh yeah, I remember that thing

It's not visible from the photos I posted, but I had the same exact resistor and capacitor really darkened. I think ben7 is correct, it is a 10 Ohm resistor. As long as it's still in spec, it doesn't need to be replaced. Same goes for the green cap. Macron usually uses 103j (i.e. 0.01 uF 5%) polypropylene rather than ceramic - a nice touch IMO.
I've seen the same happen in an Allied PSU with those snubber resistors. This is usually caused by the output caps failing.

I figured as much. I know that resistors can handle an incredible amount of heat. Something like 160C for metal film? I also figured that all this damage was caused by the overloaded rectifier and failed output caps, and that the new, almost double rated schottky + new caps would be good enough to keep it going. I've also seen this is an Allied and a Deer PSU before. So yep, failed output caps sounds accurate haha

I hope you saved that Superred fan. The one in my MPT-301 was sealed like that as well and it was completely seized - couldn't even turn it by hand. Had to drill the plastic on the back to get it taken apart. I did, and in the process, I destroyed the plastic c-washer that keeps the fan rotor blade assembly from pulling out. But it ran fine without it, so I left it as is. That was some 5 years ago or so. It still is running to this day. Not rattling or anything either. So Superred fans are good, IMO.

Yes, I normally save Superred fans if they have a seal. I already have so many spare fans, I didn't want to spend the time on this one, so I threw it away. Sorry! At least I put another Power Logic back in the field, does that make up for it?

1 hour. ONE hour? It takes me that long to scratch my ass. Then another one to realize that I did so I can stop.

Geez man, you're a true PSU wiz.

Thank you! But guys like you are the true PSU wizards, knowledge wise. I just have a speedy process and a nice soldering iron which makes things go faster. Only charged $90 to do that power supply job, they happily paid it.

I share the same for different FSP Group units. Quite good too and mostly with bloated caps only.

Which design? There are so many solid FSP units that used garbage caps

[momaka]

Re: Power supply build quality pictorial. part 2

I guess your only option to save some is to check fleabay often, and check your local PC recycling stores.

No PC recycling stores here. Rather the county does it for several months in the spring and summer at their site. Not allowed to take anything. And most people in the area here either chuck stuff in the bin or try to sell them on CL for a small fortune, lol.

But what I'm thinking is that the PSU was getting abnormally hot for a number of reasons:
1.) That 16A rectifier had to be overloaded or close to it its entire life. (Reducing efficiency in the process?)

Actually, the forward voltage drop of diodes decreases with increasing temperature, so technically efficiency improves.

I think the biggest contributors to the inefficiency were those dried up caps on the BJT drive circuit (making the BJT turn on too slow and run for longer in their linear regions) and the fact that the 12V rail was a fast recovery rectifier rather than a schottky. FRs tend to have 1V or more Vf, whereas for schottky, it is usually 0.3-0.7V - almost 2 to 3 times less losses.

I know that resistors can handle an incredible amount of heat. Something like 160C for metal film?

We had an interesting experiment back in my older college. Had to cook two resistors for a lab in various ways - one was by slowly increasing power and waiting a bit between increases to see at what power level the resistor would fail. The other was by pulsing really high amounts of power. In both cases, the resistors could handle about 2-5 times its power rating. IIRC, we did this with 1/4 Watt carbon film resistors, and most too at least 1W before they became open.

I hate killing good components, though, so that lab was not my favorite. Already knew what would happen when you abuse a resistor :P: .

Yes, I normally save Superred fans if they have a seal. I already have so many spare fans, I didn't want to spend the time on this one, so I threw it away. Sorry!

Heh no probs. It's just in my inner soul to try and save anything and everything that is salvageable.

Thank you! But guys like you are the true PSU wizards, knowledge wise.

As the saying goes: those who can't do should only teach .

[Behemot]

Re: Power supply build quality pictorial. part 2

Which design? There are so many solid FSP units that used garbage caps

Everything up to Blue Storm II / Everest / Epsilon incl. The latest started failing with other problems too but still mostly caps.

Man 90 bucks for that? I think you can buy some very nice new unit for that kind of money!

[Wester547]

Re: Saved proprietary Macron MPT-350D

Almost all of the caps tested bad. The only caps that didn't test bad were both 2200uF 6.3V G-Luxon LZ caps on the 3.3V rail (I don't know how this is possible)

Probably because G-Luxon was and is (before and after their acquisition with Teapo) inconsistent as with other Taiwanese and Chinese brands. Sometimes they fail and bulge just by sitting there for many years, other times they last several years of heat and torture as in your case (at least two of them). It may also have to do with the fact that the failed G-Luxon LZs are on the +5V rail and are literally touching the +5V/+12V toroid... and that toroid probably became quite a bit hotter than the +3.3V toroid during operation (this was a +12V heavy system from what I can surmise). Loose rectifiers can definitely impede the thermal transfer to the heatsink.

That Prescott may be leaded solder but it's still a LGA775 CPU as far as I know and the LGA775 socket is very fragile (warps easily) and its pins very easily bent and broken, and that might cause issues after that many acrid thermal cycles. The stock thermal paste surely didn't help. And that CPU only has a maximum operating temperature of almost 67*C in accordance to Intel.

And 42*C, you would consider hot? I know everyone's perception of heat and "pain tolerance" diverges, but IMO over 50*C is about the point where something becomes burning hot and uncomfortable to touch. 60*C is definitely too hot to touch and 70*C can scald your tissue in a second.

As the saying goes: those who can't do should only teach

Or those who can't do tech should only teach?

[Pentium4]

Re: Saved proprietary Macron MPT-350D

No PC recycling stores here. Rather the county does it for several months in the spring and summer at their site. Not allowed to take anything. And most people in the area here either chuck stuff in the bin or try to sell them on CL for a small fortune, lol.

That makes me sad, actually lol

Actually, the forward voltage drop of diodes decreases with increasing temperature, so technically efficiency improves.

I think the biggest contributors to the inefficiency were those dried up caps on the BJT drive circuit (making the BJT turn on too slow and run for longer in their linear regions) and the fact that the 12V rail was a fast recovery rectifier rather than a schottky. FRs tend to have 1V or more Vf, whereas for schottky, it is usually 0.3-0.7V - almost 2 to 3 times less losses.

Interesting. So the increased amperage rating, lower VF, and being tight to the heatsink made that much of a difference. Amazing!

We had an interesting experiment back in my older college. Had to cook two resistors for a lab in various ways - one was by slowly increasing power and waiting a bit between increases to see at what power level the resistor would fail. The other was by pulsing really high amounts of power. In both cases, the resistors could handle about 2-5 times its power rating. IIRC, we did this with 1/4 Watt carbon film resistors, and most too at least 1W before they became open.

I hate killing good components, though, so that lab was not my favorite. Already knew what would happen when you abuse a resistor :P: .

Yeah that'd make me cringe too, poor resistors That's amazing though that they could handle 2-5 times their rating. How long could they handle that?

As the saying goes: those who can't do should only teach

I'm sure you do plenty, you just take forever to type out the posts and upload pics I think your next dissection should be the ePOWER. No pressue. *nudge nudge*

Everything up to Blue Storm II / Everest / Epsilon incl. The latest started failing with other problems too but still mostly caps.

Man 90 bucks for that? I think you can buy some very nice new unit for that kind of money!

Yes those were good units! Yeah, but they got 5 hour turn around time. Plus, it wasn't a standard ATX PSU, seems like a rare type. Better than paying for one on eBay, waiting for it, and then probably getting one with bad caps in it. Also, fierce competition for business in downtown Seattle, so we had to make some money off our good service and special knowledge.

Probably because G-Luxon was and is (before and after their acquisition with Teapo) inconsistent as with other Taiwanese and Chinese brands. Sometimes they fail and bulge just by sitting there for many years, other times they last several years of heat and torture as in your case (at least two of them). It may also have to do with the fact that the failed G-Luxon LZs are on the +5V rail and are literally touching the +5V/+12V toroid... and that toroid probably became quite a bit hotter than the +3.3V toroid during operation (this was a +12V heavy system from what I can surmise). Loose rectifiers can definitely impede the thermal transfer to the heatsink.

I'm still surprised. The whole unit was very hot inside, and those caps are about 10 years old. And I bet that system used more 3.3V than modern ones.

That Prescott may be leaded solder but it's still a LGA775 CPU as far as I know and the LGA775 socket is very fragile (warps easily) and its pins very easily bent and broken, and that might cause issues after that many acrid thermal cycles. The stock thermal paste surely didn't help. And that CPU only has a maximum operating temperature of almost 67*C in accordance to Intel.

Intel is probably erring on the safe side. This CPU was idling at 60C and while testing got up to 80C. Imagine this thing in the past few summers...After replacing the thermal paste, it idled at 39C and peaked at 57C. That's how bad the old thermal paste was! It wasn't a stock cooler, it was significantly better, but still installed in 2005.

And 42*C, you would consider hot? I know everyone's perception of heat and "pain tolerance" diverges, but IMO over 50*C is about the point where something becomes burning hot and uncomfortable to touch. 60*C is definitely too hot to touch and 70*C can scald your tissue in a second.

No not hot to my fingers, but for the side of a PSU chassis while running, I'd say that's pretty hot!

[momaka]

Re: Saved proprietary Macron MPT-350D

Yeah that'd make me cringe too, poor resistors That's amazing though that they could handle 2-5 times their rating. How long could they handle that?

Well, what do you know, I even still have my lab report on this laptop.
Actually, the two resistors we tested (100 Ohms, 1/4 Watt, 1%) could handle about 1W of power without going out of spec for more than 5 minutes. So they may very well be capable of that rating indefinitely. However, at that power level, they did discolor quite a bit. Beyond 1W (1.15 Watts, actually), they started increasing in resistance, went out of spec, and burned out.

I'm sure you do plenty, you just take forever to type out the posts and upload pics I think your next dissection should be the ePOWER. No pressue. *nudge nudge*

Yeah, that's because I want to make my posts as perfect as possible (and they are never perfect, damit ). But I do have the pictures taken, at least.... Actually, even the write-up with all of the parts, I just checked. Don't know why I haven't posted it yet. Let's aim by the end of the week. I promise this time.

Intel is probably erring on the safe side. This CPU was idling at 60C and while testing got up to 80C. Imagine this thing in the past few summers...After replacing the thermal paste, it idled at 39C and peaked at 57C. That's how bad the old thermal paste was! It wasn't a stock cooler, it was significantly better, but still installed in 2005.

Wow, that thermal compound really was gone. It's actually not that hard to keep those Pentium 4 Prescotts somewhat cool. In general, they will run warm, but nowhere near 60C idle. 39C idle is a lot closer to what I get mine to run at (usually 42C or less, depending on room temperature, of course).

[datoska]

Re: Power supply build quality pictorial. part 2

umm, sorry but i have one fast question...
DPS-800 LB - power supply from hp workstation.
will fit in atx case ? if some holes must be created for that, its not a big problem...
i'm also interested, how much 6 pin graphics card power connectors does it have ?
and what means those 6 rails ? i'm kinda confused...

i know there is cpu line 1 and 2
but others :
12vD
12vB
12vM
12vG

(btw sorry for my english...)
and btw, i'm gonna buy it for 26.5$, its good deal for me....

[Compgeke]

Re: Power supply build quality pictorial. part 2

That's the XW8600 PSU I believe.

It won't fit an ATX board at all, it's proprietary to the workstations. I suppose with some modifications you can hack it up to a proper ATX connector and butcher the memory connector for the 12V CPU but it's probably more effort than it's worth.

It also has two 6-pin GPU power.

[c_hegge]

Re: Power supply build quality pictorial. part 2

^
I'm pretty sure it's the same as the DPS-750CB I used to use. It is mostly standard ATX, but it's a bit taller, so it won't physically fit into the PSU compartments of most cases (with the exception of some very large cases, like the Antec 1200). I think that the ATX connector pinout is standard, though. As for how many connectors it has and the rail distribution, I don't know. I haven't seen an 800LB before.

[c_hegge]

Re: Power supply build quality pictorial. part 2

Delta DPS-250AB.

https://www.badcaps.net/forum/attach...1&d=1432204397
Looks like a pretty typical TFX Form Factor unit. It came from a Dell Inspiron 530S.

https://www.badcaps.net/forum/attach...1&d=1432204397
Another shot of the Fan grille. Delta, you're doing it wrong! Wire grilles only, please!

https://www.badcaps.net/forum/attach...1&d=1432204397
The Label. It looks like we have 17A of 12V capacity. Not much, but enough for a PC with onboard graphics, I suppose.

https://www.badcaps.net/forum/attach...1&d=1432204397
Naturally, a Delta PSU uses a Delta fan. Hopefully I shouldn't have to replace it any time soon.

https://www.badcaps.net/forum/attach...1&d=1432204397
The Input filtering consists of three coils and two of everything else

https://www.badcaps.net/forum/attach...1&d=1432204397
The Internals. Not too bad, except for...

https://www.badcaps.net/forum/attach...1&d=1432204397
CapXon GL? On the 5vsb? On a cramped TFX unit? Are you serious, Delta?!

https://www.badcaps.net/forum/attach...1&d=1432204397
The rest of the caps are LTEC and Taicon. At least Delta usually get away with those pretty well. I'm still gonna recap it, though, with the exception of one 16V 1500uF Taicon which I don't have a replacement on hand for.

[Behemot]

Re: Power supply build quality pictorial. part 2

It is cheaper to cut into metal plate than buy wire grill

You can use Samxon RS 3300/16, no? There seems to be room enough for taller cap.

[c_hegge]

Re: Power supply build quality pictorial. part 2

I could, but I'd prefer not to. I'd much prefer to keep the capacitance the same.

[Agent24]

Re: Power supply build quality pictorial. part 2

I don't like Delta. They sneak a CapXon in somewhere too often for my liking.

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

I don't like Delta. They sneak a CapXon in somewhere too often for my liking.

Nothing wrong with that...


At least they are better than JunFu/AsiaX and the likes...

(But Delta also uses AsiaX)...

[Wester547]

Re: Power supply build quality pictorial. part 2

Ah, the power of the CapXon capacitors... so strong, they can't help but pop!

No, I would not rank CapXon above AsiaX, JunFu, or Fuhjyyu. They are just as bad. I don't recollect Delta ever using AsiaX, though. Any idea what the primary and secondary silicon is in the supply? And looks like the -12V linear regulator is without a heatsink so definitely a bit cheaper by contrast to some other ones. Not that it matters, -12V isn't used much at all in modern systems. And I would preemptively lube the Delta fan anyway because I've seen several ASB0812HH fans from older Delta and Newton supplies seize or come close because of a lack of lubricant.

[Agent24]

Re: Power supply build quality pictorial. part 2

Nothing wrong with that...

Nothing wrong with CapXon? Could have fooled me.
Are you saying all those bulging\leaking CapXon I've pulled and replaced were OK?