HP-D3057F3H recap

[Wester547]

Re: HP-D3057F3H recap

I also noticed Lite-on use yellow glue which I don't think is much better.

Add delta and enermax to that list.

I wonder how long Delta have been using it? I'm hoping no Newton Power has that brown glue. At least Hipro don't use that.

[c_hegge]

Re: HP-D3057F3H recap

Bestec is not the only offender to use that glue, though. I found some LiteOns of the same era to have it as well. I wonder when manufacturers will learn NOT to use it.

Add delta and enermax to that list.

[momaka]

Re: HP-D3057F3H recap

But as I've stated, the fan in mine is starting to lose its lubricant so it's starting to die, and because my PSU does not have screws for the fan, I'm not sure how to remove it.

I couldn't remove mine either. But since it was working fine I left it alone. If yours is dying, though, just rip those rubber mounts off and fix the fan since that's more important. You can use screws afterwards to hold the fan, just like other PSU do.

Bestec PSUs are good save by the brown glue Bestec likes to add and their choice of capacitors, and save for lack of OPP

OPP doesn't worry me, but the brown glue does even more than the capacitors itself since it can go conductive and make things go *bang*. That ATX-1956D of mine was a nightmare to clean since it's one of those compact ATX PSUs that's a little shorted. I couldn't clean all of the brown glue and there still lots left on the logic board. The PSU runs warm too, so that glue will likely cook faster (and thus become conductive quicker).

Bestec is not the only offender to use that glue, though. I found some LiteOns of the same era to have it as well. I wonder when manufacturers will learn NOT to use it.

Maybe recapped, they have better voltage regulation, unless you already recapped the ones you reference.

The ATX-1956D still has the original PCE-TUR caps. Regulation is more or less okay, but it cross-loads very easily.
The ATX-250 12Z is half-recapped (that is, each major output rail has 1 new cap and 1 old cap, except the 12V rail which has all new caps). Just like the ATX-1956D, regulation isn't bad but it cross-loads very very easily.

There's nothing really wrong with half-bridge. It's just that it's not as efficient as other topologies.

Exactly. Not efficient but it can work very well if well designed.

[goodpsusearch]

Re: HP-D3057F3H recap

Macron Power MPT-350P

https://www.badcaps.net/forum/showthread.php?t=12348

They look decent to me.

[c_hegge]

Re: HP-D3057F3H recap

^
There's nothing really wrong with half-bridge. It's just that it's not as efficient as other topologies.

[Wester547]

Re: HP-D3057F3H recap

I know. I put mine (HiPro HP-P2507F3P) in both 5V and 12V heavy systems, and the voltages are almost always the same. It just won't cross-load! Even Delta/Newton, LiteOn, and Astec can't beat it. Bestec - no chance.

I'm surprised Newton Power PSUs have worse voltage regulation given how well designed they are otherwise. But I do concede that the Newton Powers I've seen have better fans (usually better than ADDA) and choice of capacitors (the worst I've seen in my Newtons is Ltec - the worst I've seen in a Hipro is well, Arkon) Hipros do of that era, so them's the breaks I suppose. However, assuming these tests are correct, this Hipro does not seem to have such great voltage regulation (unless it has something to do with only having 9A on the +12V rail and being rated at 300W).

Even my HiPro has general purpose AsiaX caps - and that thing barely puts out any ripple (at least based on some reviews I read a while back... I think one was a JonnyGuru).

I had a HP-P2507F3P as well in a Dell Dimension 8200, though revision H01, whereas if I'm not mistaken (as shown in the PSU build thread) yours is H01A. But as I've stated, the fan in mine is starting to lose its lubricant so it's starting to die, and because my PSU does not have screws for the fan, I'm not sure how to remove it. I do know it's a 80x80x20 sleeve bearing fan from ADDA, and my guess it's high speed because of that (3010RPM/27.3CFM/.128" air pressure from the datasheet if I'm not mistaken). Probably was not doing the job it could have done though because Dell, in their infinite wisdom, decided to put an extra grate on the back of their case, right in front of where the PSU exhaust intake is. That may be better for dust as Dell stresses they do not like dust in their documentation, but I don't know how that impacts the cooling of the PSU. But given that the PSU has a Passive PFC it's already running a bit warmer anyway. Anyway, my version had a mix of Teapo (on the primary, 680uF/200V at 85C, and secondary), G-luxon, and Asiacon capacitors, rather than Panasonic (560uF/200V at 85C)/Asiacon. I don't know if that's any better or worse than all Asiacon on the secondary but I do know it never gave me instability problems. I don't know how stable the voltage regulation was because I only ever used that PSU in a Dell Dimension 8200 and Dell do not give you temperature sensors, much less voltage readings (though the ones of the motherboard sensors aren't accurate anyway). And I never measured it myself. But I'm guessing it was darn stable if yours was.

I don't mind Bestec PSUs, but so far my experience isn't stellar with them. I have both a ATX-1956D and a ATX-250 12Z and they both cross-load very easily. The ATX-1956D is only rated for 12 A on the 12V rail and it was obviously built for a 5V system. Currently, I have it in a Dell GX270 motherboard powering a 2.66 GHz P4 Northwood, and the 12V rail dips down to 11.85V. Puts out quite a bit of heat too. As a test, I tried putting a decent 5V load from a Radeon 9700 video card and the 12V rail got up to or over 11.90 V. I then tried it in my 5V-based 750 MHz AMD Duron test system (I call it the "PSU tester" ) and the 5V and 12V rail were doing much better.

The ATX-250 12Z is the other way around - put in in my "PSU tester" with a single hard drive and the 12V rail jumps to 12.5 V while the 5V rail is only a little high at 5.15 V! Then I put that PSU in a dead motherboard with a 1.7 GHz P4 Willamate (which draws more juice than the 2.66 GHz Northwood) and the 12V rail behaves itself.

Bestec PSUs are good save by the brown glue Bestec likes to add and their choice of capacitors, and save for lack of OPP (OTP would be nice too but you don't see that in too many PSUs.... preferably, a PSU should INSTANTLY shut down when it detects the fan is no longer running, if it does, but I don't know of any PSU with that ability). Maybe recapped, they have better voltage regulation, unless you already recapped the ones you reference.

It's actually a very good PSU. Just an older half-bridge design, that's all. I used mine as a test PSU for a year for testing various systems. It does fine in both 5V and 12V heavy systems and voltage regulation is very good - almost as good as my HiPro. Last month, I was even testing my friend's quad-core i7 2600k with it (a 95W TDP CPU). The 12V rail sat at 11.99 to 12.01V the whole time. He didn't believe me such old PSU would run his computer. Of course we removed his video card for the tests (it was a high-end nVidia so I don't think my MPT-301 would have liked that). The 5V rail was doing fine too - it went no higher than 5.15V. Given that the i7 above had build-in Intel graphics, you could easily build a decent system with it (granted it's not used for gaming).
I also had that PSU powering a 107 W AMD Athlon FX-57 CPU. 12V rail was almost exactly the same: 12.00 V to 12.05V under load/idle.

Looks are deceiving, then - I suppose even if it looks desolate (though it may not be missing any components) that the quality of the components matter even more than the amount of them. I wonder if half bridge topology is that bad of a thing, considering I always see PSU reviewers scorn PSUs for having that instead of full bridge, single transistor forward, or whatever else.

[momaka]

Re: HP-D3057F3H recap

But I'm sure they'd at least fail LESS often if they were low ESR caps.

Sure .
That's exactly why CapXon GF, GL, and KF fail a lot LESS often that KM, right
(I wish I got a penny for every failed CapXon I've seen - I'd be filthy rich).
CapXon aside, I have seen many "low-ESR" caps fail in both computer and LCD PSUs. Cheap crap caps are just unreliable and unpredictable at best - sometimes they bulge, sometimes they don't (even in same PSU models). The only predictable thing about them is that if you expose them to heat, they are guaranteed to bulge sometime in their future.

Sure, go ahead and try it. But if your computer blows up, don't say I didn't warn you.

I don't need to. The cheap Chinese PSU manufacturers have been doing that already for quite a while.
The bottom line is, decent Chinese PSUs with general purpose caps will actually be in spec in terms of output ripple and noise. It's the build quality of these caps that makes them eventually fail. Take for example my Bestecs above - both have general purpose caps but have worked *fine* for many years. Even my HiPro has general purpose AsiaX caps - and that thing barely puts out any ripple (at least based on some reviews I read a while back... I think one was a JonnyGuru).

Heck, even CRT monitor PSUs are switch mode, yet if you open one, you'd see they are littered with general caps everywhere. But they don't fail too often do they?

It's only the very cheap PSUs that skip coils and put undersized general purpose caps that you will see fail regularly and predictably.

"A little" is a bit of an understatement...
Chemi-con KME 3300uF 6.3V 12.5 x 20: 131mΩ
Nichicon PW 3300uF 6.3V 12.5 x 20: 38mΩ
Now tell me how high the ripple will go...

That's just numbers there. Put those caps in a real world application (i.e. a PSU) and there won't be that much of a difference, especially PC PSUs. Reason for that is because the large toroidal coils in the PSU also do a big part of the filtering. Now if this was a linear or flyback PSU (such as those in LCD monitors and 5VSB circuits), this would be a different matter (note that those don't have large toroidal coils). Such PSUs do benefit from low ESR caps and in fact may produce too much ripple with general purpose caps.

A modern GP series like Chemi-con KMG would be smaller - in this case, 10 x 20.

Yes, sometimes that is the case too. This is also good though because you can't always find good low ESR caps to fit. Take for example 16V 2200 uF (or higher) caps in 10 mm cans - nearly unobtanium from the good manufacturers except for a few series (think Samxon RS). In those cases, you would be better off with a 16V 2200 uF general purpose cap from a reputable manufacturer rather than using a polymer with half the capacitance. By better off, I mean the repair with the general purpose cap is almost guaranteed to run without problems, whereas the polymer with half capacitance may or may not work, depending on how picky the PSU is.

Anything else you need to know?

What's your social security number?

[Shocker]

Re: HP-D3057F3H recap

Well, yeah, that's because they use shitty caps to begin with.

Yeah, they OBVIOUSLY do. But I'm sure they'd at least fail LESS often if they were low ESR caps. Read below for the reasons.

What I meant is use a general purpose cap from a *reputable* manufacturer, then see if they go bad. I'm pretty sure they won't.

Sure, go ahead and try it. But if your computer blows up, don't say I didn't warn you.

The higher ESR will cause them to run a little hotter

"A little" is a bit of an understatement...

Chemi-con KME 3300uF 6.3V 12.5 x 20: 131mΩ
Nichicon PW 3300uF 6.3V 12.5 x 20: 38mΩ

Now tell me how high the ripple will go...

but general purpose caps also have larger cans so they can dissipate more heat too - so if they are of a good brand, they will handle it.

In the example I just mentioned the capacitors are the same physical size. And FYI, Chemi-con KME is discontinued, while Nichicon PW is still available. A modern GP series like Chemi-con KMG would be smaller - in this case, 10 x 20.

If they changed the circuit board, I never noticed it. If I remember right they even had the same legend "ATX9912P, 0103P MACRON POWER." printed near the front edge.

Anything else you need to know?

[momaka]

Re: HP-D3057F3H recap

Not in my experience. I have several Macron MPT-401s (with clear 120mm fans) from old computers. Macron have a habit of using general purpose caps and guess what - 100% of them (that I checked) have bad caps.

Well, yeah, that's because they use shitty caps to begin with. What I meant is use a general purpose cap from a *reputable* manufacturer, then see if they go bad. I'm pretty sure they won't. The higher ESR will cause them to run a little hotter, but general purpose caps also have larger cans so they can dissipate more heat too - so if they are of a good brand, they will handle it.

Here's the MPT-301, which is almost the same

Internal pics would help though. Maybe they changed the design of the 5VSB? It would be nice if they did. I have an MPT-301, so I can tell you they use a 2-transistor self-oscillating design with feedback. Not the best really. Fortunately, the design does *not* have a critical capacitor on the primary. Also, the rectifier for the 5VSB is IIRC either a 6 A or 10 A in a TO-220 package mounted to the secondary heat sink. And I think the primary BJTs are a pair of 13009s.

Kinda looks alright, except for those pathetic heat sinks.

That shot doesn't show it, but their base is actually very skinny - around 2 mm. They run hot in mine, but not the hottest I've touched. Certainly better than the heat sinks of any LCD monitor.

I think you'd be hard-pressed to find a PSU brand with better voltage regulation and stability than Hipro.

I know. I put mine (HiPro HP-P2507F3P) in both 5V and 12V heavy systems, and the voltages are almost always the same. It just won't cross-load! Even Delta/Newton, LiteOn, and Astec can't beat it. Bestec - no chance.

I don't mind Bestec PSUs, but so far my experience isn't stellar with them. I have both a ATX-1956D and a ATX-250 12Z and they both cross-load very easily. The ATX-1956D is only rated for 12 A on the 12V rail and it was obviously built for a 5V system. Currently, I have it in a Dell GX270 motherboard powering a 2.66 GHz P4 Northwood, and the 12V rail dips down to 11.85V. Puts out quite a bit of heat too. As a test, I tried putting a decent 5V load from a Radeon 9700 video card and the 12V rail got up to or over 11.90 V. I then tried it in my 5V-based 750 MHz AMD Duron test system (I call it the "PSU tester" ) and the 5V and 12V rail were doing much better.

The ATX-250 12Z is the other way around - put in in my "PSU tester" with a single hard drive and the 12V rail jumps to 12.5 V while the 5V rail is only a little high at 5.15 V! Then I put that PSU in a dead motherboard with a 1.7 GHz P4 Willamate (which draws more juice than the 2.66 GHz Northwood) and the 12V rail behaves itself.

IMO, the power supply Shocker posted looks gutless, though there is obviously worse out there.

It's actually a very good PSU. Just an older half-bridge design, that's all. I used mine as a test PSU for a year for testing various systems. It does fine in both 5V and 12V heavy systems and voltage regulation is very good - almost as good as my HiPro. Last month, I was even testing my friend's quad-core i7 2600k with it (a 95W TDP CPU). The 12V rail sat at 11.99 to 12.01V the whole time. He didn't believe me such old PSU would run his computer. Of course we removed his video card for the tests (it was a high-end nVidia so I don't think my MPT-301 would have liked that). The 5V rail was doing fine too - it went no higher than 5.15V. Given that the i7 above had build-in Intel graphics, you could easily build a decent system with it (granted it's not used for gaming).
I also had that PSU powering a 107 W AMD Athlon FX-57 CPU. 12V rail was almost exactly the same: 12.00 V to 12.05V under load/idle.

[Shocker]

Re: HP-D3057F3H recap

Do you think it would be worthy of a review???
Knock 5 points off for all the capacitor failures


Oh and by the way, that MPT-301 is not mine, it's goodpsusearch's. Here's the thread.

[Wester547]

Re: HP-D3057F3H recap

I somehow doubt that their capacity really was 1000 uF. Most cheap brands overrate by a size higher. In my experience, Elite are somewhat better than the other cheap brands, so maybe they could have been 1000 uF.

Well, it's also that Hipro overbuilds many of their PSUs despite not having the best choice of capacitors and fans. Besides those things, with a few exceptions, I think you'd be hard-pressed to find a PSU brand with better voltage regulation and stability than Hipro. They do make nice OEM PSUs. Hipro, Delta/Newton Power, and Lite-on all overbuild many of their PSUs (as seen in OEM PCs), and all for the better.

This is an internal look at a Delta 330W PSU and it too has 1000uF input capacitors (they're 200V, though, by Rubycon), though ironically the +12V rail on that PSU is only rated at 12A as per the label (which is low for a 330W, so I'm assuming it's an older unit that gives more power to the +5V/3.3V rails.. no surprise - it looks like that internal look of that PSU was posted all the way back in 2000!). It also has a Delta dual ball bearing fan, like the ones Shocker used for his MTP-401s. Nothing wrong with that - Delta is one of the best fan makes and I'd say there's no fan better than a dual ball bearing fan, period, especially for PSUs (I think fan failure is more critical in the PSU than in any other component).

IMO, the power supply Shocker posted looks gutless, though there is obviously worse out there.

[c_hegge]

Re: HP-D3057F3H recap

Kinda looks alright, except for those pathetic heat sinks.

[Shocker]

Re: HP-D3057F3H recap

Here's the MPT-301, which is almost the same:



SBL3040PT on +3.3V, 2 x SBL3040PT on +5V, and 1 x F16C20C on +12V.

The only significant difference in the 400W model is that it has 2 x F12C20C on +12V. Otherwise it's the same internally.

Output capacitors are:

• +3.3V: 2 x 2200uF 10V 10mm
• +5V: 2 x 2200uF 10V 10mm
• +12V: 1 x 3300uF 16V 12.5mm
• -5V: 1 x 220uF 16V 6.3mm
• -12V: 1 x 220uF 16V 6.3mm
• +5VSB: 2 x 2200uF 10V 10mm

Of course, there's a good chance that they will have blown their tops, as mentioned above.

[c_hegge]

Re: HP-D3057F3H recap

^
Hey, do you have an internal shot of an MPT-401? I can get them cheap, but I want to know what they're like internally

[Shocker]

Re: HP-D3057F3H recap

Almost every SMPS PSU will work even with general purpose 105C caps, as long as you keep the capacitance the same or maybe a little higher.

Not in my experience. I have several Macron MPT-401s (with clear 120mm fans) from old computers. Macron have a habit of using general purpose caps and guess what - 100% of them (that I checked) have bad caps. I recapped six (with Rubycon ZLH, Panasonic FR, Chemi-con KZE, and Panasonic ED primaries) and two of them are currently in use. I also replaced the fans...with boring black 120mm ball bearing Deltas (but who cares that they're black??? ).

I do have some other Macron models and not all of those have popped caps, but maybe they just weren't used for long. One of them is an MPT-400 with an 80mm fan. Again I recapped it (same caps that I mentioned earlier) and replaced the fan (with a ball bearing NMB - the same model I used in the Hipro that is the subject of this thread).

[momaka]

Re: HP-D3057F3H recap

I realise that the capacitance shouldn't be halved in PSUs, but 3300uF 16V polies simply don't exist, so I used the closest I could get my hands on for the experiment.

Right.
But if anyone ever runs into a dilemma with whether it's better to use a polymer with half capacitance or a regular cap with inferior ripple/ESR than the originals but same capacitance, pick the latter. Almost every SMPS PSU will work even with general purpose 105C caps, as long as you keep the capacitance the same or maybe a little higher. The result would be only somewhat higher ripple and noise on the outputs.

The only exception is 5vsb circuits and MOSFET-generated 3.3V rails - both of these are linear and benefit from very low ESR, high ripple current caps. They will work with general purpose caps too, but you'll get better performance with lower ESR caps.

1000uF primaries in a non-APFC PSU is rather an overkill for a 300W PSU

I somehow doubt that their capacity really was 1000 uF. Most cheap brands overrate by a size higher. In my experience, Elite are somewhat better than the other cheap brands, so maybe they could have been 1000 uF.

[c_hegge]

Re: HP-D3057F3H recap

I thought that sounded strange.

[Wester547]

Re: HP-D3057F3H recap

It's actually 350W, but yeah, 1000uF caps are still rather unnecessary.

Two 18A 12V rails seem about right for 350W.

ERL-37??? Never seen one of those. Can you show me a pic???

My bad, an ERL-35 in both of them, as c_hegge confirmed. ^^;

[Shocker]

Re: HP-D3057F3H recap

Were they rated at 85C or 105C if that was specified?

85°C

1000uF primaries in a non-APFC PSU is rather an overkill for a 300W PSU, though, even for a 80+% efficient and +12V heavy one, IMO....

It's actually 350W, but yeah, 1000uF caps are still rather unnecessary.

All the ATX Hipros (250W and 300W) I have contain a ERL-37 (or 35? Depending on whether the numbers before or after the "ERL" text indicate such) transformer, which I suppose makes them good for 300W in that regard.

ERL-37??? Never seen one of those. Can you show me a pic???

[c_hegge]

Re: HP-D3057F3H recap

The 250 and 300 Watters I've seen have all had ERL-35 transformers. The HP-D3057F3H (300W), will manage 450W for a few minutes at a time, but it will shut down after a few minutes. I remember Th3_un1qu3 saying somewhere that given good silicon, an ERL-35 transformer will be OK for 500W.