Hardware Insights Forum

The +3.3V rail uses a single 20A part, SBL2060CT, not SBL3060CT, judging by the picture. Maybe the fan's lack of lubricant caused it to slow down severely which might be one reason why the 2SK2611 blew (you said the Hipro was very loud in the polymod test at 300W - this fan should have been equally loud at 300W-450W but was not...). I have never understood why LTEC last so long in Newton/Delta 250W-305W power supplies that almost look identical to the Hipro posted here but why LTEC can't make it in these units. I have opened up and compared Newton units to these and I do not see the difference. Both have equivalent toroids, equivalent fans, equivalent heatsinks, similar fan controllers, equal ventilation, and similarly spec'd schottkys and MOSFETs on the primary and secondary along with the bridge rectifiers and bulk storage (though Newton have a smaller output filter on the secondary but maybe that's because the ripple frequency may be higher)... only that Delta use the LTEC LZG series instead of LZP so maybe LZG is more reliable (LZG is rated up to 7,000 hours and LZP only up to 4,000).

Even with the 2SK2611 that blew (probably because the efficiency might have been below 70% at 450W load, at which point the chances for survival dissipates), I bet once you oil the bearings that the fan would spin faster (that would probably account for the higher exhaust temperature) and that would still easily make it a 350W PSU (once you replace the main switcher that is). Not bad. Also, hard to believe how low the ripple is on +5V, +3.3V, +12V, and -12V without linear regulation as well as +5VSB... it is odd that this unit didn't shut down. You'd think at 450W load, with that fan slowing down, that the unit would have shut down due the thermistor on the secondary heatsink detecting that the secondary heatsink had overheated...

I fixed that bit about the rectifier.

The fan speed was about the same in this unit as it was in the polymod test. It gets very noticeable (when you turn off the load tester's fans, that is), but it's more of an air moving whoosh noise than a motor whine, so it's not IMO really annoying (which is what I meant in the review).

The secondary side probably is capable of managing 450W. It did it on my last few 3057F3H's and this one has a similar secondary side. It's the primary side that was different and would have run considerably hotter, and the OTP only monitors the secondary heatsink.

I don't know why the ripple was so low, but that's what the scope told me. It normally picks it up just fine on other power supplies.

Well, you said that the previous scope was more sensitive to spikes... does that possibly make it more accurate or less? And when you say 3-4 years of moderate use, do you have any idea what that means power on hour wise (if you've ever checked the S.M.A.R.T. data of any of the hard drives that were in the machines these PSUs come from, assuming the HDDs came with the machines and weren't replacements)?

The old scope was probably less accurate. Those spikes don't normally appear on any other test results I've seen from other reviewers (and in case you're wondering, yes, I did have the two capacitors in the load tester required by ATX specs for testing as well).

I'm not too sure about exact hours, as I never checked SMART. I usually go by who owned it. If the PC was owned by an old granny, it probably won't see that much use compared to one which is used to drive something like a point of sale system at a business.

This "+5V heavy" power supply wasn't really "on the market" as it is 7 years old. Would it be really fair to take a point off reviewing a 7 year old unit just because it was +5V heavy 7 years ago, being 7 years old? Also, the discrepancy in ON resistance between the two main switchers is closer to 27%-32%, not 25% (going by their "typical" and "max" ON resistance ratings).

Totally fair IMO. It has a 24 pin ATX connector, a 12V CPU connector, and SATA connectors. It's newer than a Pentium 3, so it should be 12V heavy. I score units based on how they compare with what else is on the market today, not based on how they performed by the standards of 15 years ago.

I made the mistake of dividing the difference in RDS(on) values by the value of the Toshiba part, not the Fairchild part. The error has been fixed.

c_hegge (in the review) wrote:[The FQAF11N90C] is only rated for 7A at 25°C, but has a typical RDS(on) of 0.91Ω – nearly 32% lower. This means that, although this unit’s switcher was rated for slightly more current, it may have been running as much as 32% hotter…

First sentence: 0.91 ∕ 1.2 = 0.75833… (about 24% lower)
Second sentence: 1.2 ∕ 0.91 = 1.31868… (about 32% higher)

I do seem to recall you mentioning not being very good at maths, maybe that's the problem.

OK. Re-worded.

By the way, check this out: There's a mistake Hipro (and some other manufacturers) have made with mains wire terminations in some of their units, including this one.