HP-D3057F3H recap

[Shocker]

I did this earlier today using capacitors from old motherboards. I'm glad I didn't wait longer to do it. Even so, I should have done it much earlier. You'll eventually find out why...

Original capacitors (all are 10mm except the +5VSB):

• +3.3V: LTEC LZP 2200µF 10V
• +5V: LTEC LZP 2200µF 10V
• +12V: Chemi-con KZE 2200µF 16V
• -12V: Chemi-con KZE 2200µF 16V
• +5VSB: Teapo SC 1000µF (bulged) and 470µF 10V

Replacement capacitors:

• +3.3V: Nichicon HM 3300µF 6.3V
• +5V: Nichicon HM 3300µF 6.3V
• +12V: Not replaced
• -12V: Not replaced
• +5VSB: Rubycon MBZ 1000µF 6.3V

I also replaced the primary caps (Teapo LH 680µF 250V) with Chemi-con SMQ (same value and voltage). Best to be on the safe side.

I was originally keeping it on the list of spare PSUs to be (eventually) recapped. Only today did I discover that the repair was overdue.

I went to install the fixed PSU in my PC. I pulled another (unrepaired) unit out of my PC. I waited long enough for one of the +3.3V caps in that one to go from non-bulging to leaking. (In case you're wondering, the +5VSB caps in that one weren't bulging.)

The only symptom of that crap was a faint high-frequency whine.

Morals of the story:

• Even totally busted caps don't always show obvious symptoms.
• NEVER leave crap caps alone.
• Never, EVER use an HP-D3057F3H with the original caps.

Any comments???

[c_hegge]

^
I've recapped about 20 of those in the past. Usually, the 1000uF teapo on the 5vsb fails, and the PSU won't start. They are good PSUs, though, after a recap.

[Shocker]

I would have used closer caps to the originals if I had some

But the MBZ were the only quality 1000µF 8mm I had, and while I had a third Hipro to take parts from (which was where I got the Chemi-con primaries), I chose to save the 2200µF 16V KZEs for something that actually needs 2200µF 16V 10mm caps.

Anyway, it works, and that's what matters.

[mockingbird]

I am curious how much ripple is coming out of your +5VSB line with that MBZ in there...

[c_hegge]

I doubt it would be much. I polymodded one once, and used a 6.3v 1000uF Nichicon L8 in that position. I got exactly 2.0mV from the 5vsb with 1A Load. My main concern is actually where they came from. If he scavenged them from the VRM low side of a motherboard, then they will have de-rated to 1.5V or so.

[Shocker]

I don't have an oscilloscope.

Neither the 1000µF MBZ nor the 3300µF HM were from Vcore.

[c_hegge]

They might be OK then.

[Shocker]

Actually what I'm worried about, if but slightly, is this:

http://www.badcaps.net/forum/showthread.php?t=19391

[momaka]

Originally Posted by mockingbird

I am curious how much ripple is coming out of your +5VSB line with that MBZ in there...

Should be less.
The 5VSB power supply, unlike the main power supply, doesn't have large chokes and the control loop is slightly different, so such circuits actually benefit from the lower ESR caps.

In the main power supply circuit, the output capacitor-inductor filter is tuned, so deviating from the ESR of the caps that were originally used actually increases the ripple.

[Shocker]

Take a look at the poly-mod on HWI, and note how the Hipro was the only PSU to work properly after the mod (the SmartPower worked fine on the load tester but failed in-PC, while the NeoHE went crazy). Surely if it can cope with polys then MBZ and HM can't be a problem.

[c_hegge]

Precisely what I was thinking.

[Shocker]

By the way...is that an MBZ I see there???

And I was wrong about the source of the whine. There are actually two different whine sources. The first one I think is the CPU fan motor. The second is the monitor (maybe it has bad caps, I haven't checked).

[c_hegge]

^
Yup. I had to recap it to get a baseline ripple reading, since the Teapos there had failed, and were preventing the PSU from starting

[Shocker]

The first one I think is the CPU fan motor.

No, I think that one's wrong again.

But the monitor is still a source.

[Shocker]

No wait, maybe I'm right again.

I think the fan motor whine peaks around 2000RPM.

[momaka]

Originally Posted by Shocker

Take a look at the poly-mod on HWI, and note how the Hipro was the only PSU to work properly after the mod (the SmartPower worked fine on the load tester but failed in-PC, while the NeoHE went crazy). Surely if it can cope with polys then MBZ and HM can't be a problem.

I don't mean to denounce HWI, but IMO shovenose and c_hegge got lucky with the design on that HiPro. Most PC PSUs are NOT designed to use polymers or very low ESR caps. Like I mentioned, it often introduces ringing in the DC filters, which in turn results in more ripple and noise - *most* of the time.

Using polymers with half the capacity is even worse, because the response time of the control loop (the circuit that actually regulates the voltages and keeps them stable) is severely affected and may not be able to respond properly to the varying loads from the PC anymore. This is most likely the reason why that SmartPower didn't work in their PC.

[c_hegge]

Originally Posted by momaka

IMO shovenose and c_hegge got lucky with the design on that HiPro. Most PC PSUs are NOT designed to use polymers or very low ESR caps. Like I mentioned, it often introduces ringing in the DC filters, which in turn results in more ripple and noise - *most* of the time.

I agree with you, and I do think I got lucky that time. Generally, I've never advised poly-modding power supplies, but this one just happens to be tolerant.

[Shocker]

I added a new NMB dual ball bearing fan and Panasonic ED 105°C primary capacitors. Because I want my stuff to last as long as possible.

[c_hegge]

Those SuperRed fans that they come with are actually pretty reliable. I sometimes add a drop of lubricant to the bearings just to be sure, but I've never had one fail before.

Believe it or not, I just came accross one which had the usual bad Teapo 1000uF on the 5vsb which was preventing it from starting, but the other caps were Chemi-con KZE on the 12V rail and Nichicon HM (Equivilant to Ruby MBZ/Panny FJ/Sanyo WG) on the 5V rail. It also had Chemi-con primaries. I recapped the 5vsb with a Chemi-con KY for the failed 1000uF teapo, and a panny FM for the 470uF. I used Ruby MBZ for the 3.3V rail (which had Teapos originally). If Hipro are using Nich HM or equivilent, then these PSUs must be fairly tolerant of lower ESR caps.

Originally Posted by momaka

Using polymers with half the capacity is even worse, because the response time of the control loop (the circuit that actually regulates the voltages and keeps them stable) is severely affected and may not be able to respond properly to the varying loads from the PC anymore. This is most likely the reason why that SmartPower didn't work in their PC.

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.

[Shocker]

Originally Posted by c_hegge

Those SuperRed fans that they come with are actually pretty reliable. I sometimes add a drop of lubricant to the bearings just to be sure, but I've never had one fail before.

I know. It's just that I never feel 100% confident about sleeve bearing fans, regardless of manufacturer.

I also recently got my hands on a HP-D3537F3H. The caps are:

• +3.3V: 1 x 10mm Chemi-con KZE (couldn't read value) and 1 x 1000µF 10V 8mm Teapo SC (popped; the silkscreen is for a 10mm)
• +5V: 2 x 2200µF 10V 10mm LTEC LZP (same as the older model)
• +12V: 2 x 2200µF 16V 10mm Teapo SC (same as the older model)
• -12V: 1 x 1000µF 16V 10mm and 1 x 470µF 16V 8mm Teapo SC
• +5VSB: 1 x 1000µF 10V 8mm (popped ) and 1 x 470µF 10V 8mm Teapo SC (same as the older model)

This is a more 12V heavy unit, as opposed to the HP-D3057F3H. At least according to the label.

• +3.3V 15A
• +5V 13A
• +12Va 18A
• +12Vb 18A

The combined +12V rating is 25A. Unfortunately I couldn't find a combined +3.3V and +5V rating on the label. The fan is an ADDA .

The +12V has two MBR20H100CTGs. The +5V has two SBL2060CTs. There are two rectifiers on the +3.3V but I couldn't read their markings.