Hardware Insights Forum

Nice review.

I'm surprised the power supply still worked given the discoloration caused by the hot load resistor next to that tiny Teapo.... if the PCB is really darkened, Chicony must have not chosen an appropriately spec'd resistor for the -12V rail. They have done that before in some old 200W Dell Hipros as documented by momaka a few years back on badcaps.

There was no lubricant left in the ADDA sleeve bearing fan, you say? So it must not have been spinning very fast at all until high loads? Or was it spinning about as fast as other temperature controlled and newer AD0812HS-A70GL fans you see used in PSUs?

And it looks to me like +3.3V has its own transformer tap separate from the +5V rail... do you mean that the extra transformer tap is in fact sharing the +5V rail's transformer pin judging by the underside of the PCB? Or maybe it does have its own transformer tap but is still connected to the +5V output? My reason for asking is that linear regulating +5V to +3.3V would waste tons of power.

That primary heatsink doesn't look bad to me at all... good thing that fan isn't mounted on the other end of the PSU, otherwise that passive PFC would be blocking lots of airflow to the primary side!

"It’s located on the other side of the heat sink, and, in most power supplies, is the first capacitor to fail, which is probably why Dell opted to use a Japanese part.".... not a surprise... when +5VSB is on the power supply is internally shut off otherwise which means no airflow and a much higher ambient temperature even with the very low load, especially with that diode being so close to the +5VSB capacitor. Also, you can't expect great efficiency when +3.3V is taken from +5V... I also think the FYP1010DN is freewheeling for the +12V rail. I don't know if that would make it good for 30A in forward topology? More like 26A at the most as evidenced by the review, perhaps closer to 24A-25A in warmer weather. It looks to me like they used an N-Channel Stripfet known as STP40NF03L to regulate the +3.3V rail. It also looks like Chicony's soldering quality has improved over time. Their soldering was not always this good and used to be average. I am also extremely pleased to see such low ripple (though voltage regulation could have been a tad better IMO), to be expected with linear regulated +3.3V. Any 3300uF Teapo SCs in there or was it 2200uF from what you saw (I spotted some 1000uFs in there)? And what was the bridge rectifier?

Also, was the Chemi-con capacitor a KZH? And was it 2200uF or 1000uF? ^^; Also, the power factor is somewhat low for a P-PFC unit... and judging by the datecode of those Teapo capacitors and the main and/or flyback transformer, this is a 2005 or 2006 power supply? One last thing... you don't think a wire grille would be more open for the fan?

EDIT: Also, I noticed that the fan is blowing/sucking cold air in the PSU rather than blowing hot air out... not sure if that's more or less effective. Did you catch what series the Nichicon primaries were?

Wester547 wrote:EDIT: Also, I noticed that the fan is blowing/sucking cold air in the PSU rather than blowing hot air out... not sure if that's more or less effective.

Refer back to this thread:

I wrote:...obstructing the front of the fan increases noise far more than the same obstruction behind the fan.

Of course, that's only one factor in the design of the cooling system, but it's a big one.

Also, why increase the height of the PSU from standard ATX while keeping the fan size the same???

LongRunner wrote:why increase the height of the PSU from standard ATX while keeping the fan size the same???

It's beyond me. There's room for a 92mm fan there.

Wester547 wrote:There was no lubricant left in the ADDA sleeve bearing fan, you say? So it must not have been spinning very fast at all until high loads? Or was it spinning about as fast as other temperature controlled and newer AD0812HS-A70GL fans you see used in PSUs?

I've seen that lots of times from ADDA. They usually don't give them any lubricant from the factory. They work fine at first, but don't last long

Wester547 wrote:And it looks to me like +3.3V has its own transformer tap separate from the +5V rail... do you mean that the extra transformer tap is in fact sharing the +5V rail's transformer pin judging by the underside of the PCB? Or maybe it does have its own transformer tap but is still connected to the +5V output? My reason for asking is that linear regulating +5V to +3.3V would waste tons of power.

You're right. I have fixed up that bit. The 3.3V rail does have it's own transformer pin (it was the freewheeling side that was shared ) I don't know how I made that mistake. In any case, though, it definitely uses linear conversion, which probably explains why it was so inefficient, despite being a forward design.

Wester547 wrote:That primary heatsink doesn't look bad to me at all... good thing that fan isn't mounted on the other end of the PSU, otherwise that passive PFC would be blocking lots of airflow to the primary side!

It is shorter than what most use, but I agree that it's very thick, and its fine for this PSU

Wester547 wrote:I also think the FYP1010DN is freewheeling for the +12V rail. I don't know if that would make it good for 30A in forward topology? More like 26A at the most as evidenced by the review, perhaps closer to 24A-25A in warmer weather.

I think 30A would be just fine. It has a massive heat sink, and there's no way it's gonna heat up anywhere near 150*C. I've pulled the rectifier's ratings (and more) lots of times in forward topology. In this case, I would say the primary side was beginning to max out. Since this is group regulated, the unit was probably trying to keep the 5V rail in spec, but letting the 12V rail drop (which is common when a PSU is adapted from an older design)

Wester547 wrote:It looks to me like they used an N-Channel Stripfet known as STP40NF03L to regulate the +3.3V rail. It also looks like Chicony's soldering quality has improved over time. Their soldering was not always this good and used to be average.

Interesting. I don't think I've ever seen a bad soldering job from Hipro, even in older units. They do occasionally leave component pins a bit too long, but they've always been miles better than some others (*cough* CWT *cough*)

Wester547 wrote:( Any 3300uF Teapo SCs in there or was it 2200uF from what you saw (I spotted some 1000uFs in there)? And what was the bridge rectifier?

Two 2200uF on the 12V Rail, a 4700uF and a 3300uF on the 5V rail, and three 1000uF on the 3.3V. I'm not sure what the rectifier was. The part number side was stuck on to the heat sink. Knowing Chicony/Hipro, though, probably 6A

Wester547 wrote:Also, was the Chemi-con capacitor a KZH? And was it 2200uF or 1000uF?

2200uF 10V KY Series

Wester547 wrote:Also, the power factor is somewhat low for a P-PFC unit

That's pretty normal to me. I think 230V usually causes better efficiency, but a lower power factor than 110V.

Wester547 wrote:and judging by the datecode of those Teapo capacitors and the main transformer, this is a 2005 power supply?

Maybe, but it doesn't seem to have had a lot of use. Most stuff from that era would be packed with dust and have bad caps by now.

Wester547 wrote:One last thing... you don't think a wire grille would be more open for the fan?

Yes, I agree, but honeycomb grilles are a close second. I don't think it's enough of a problem to really knock it for. At least it's not like this...

c_hegge wrote:You're right. I have fixed up that bit. The 3.3V rail does have it's own transformer pin (it was the freewheeling side that was shared ) I don't know how I made that mistake. In any case, though, it definitely uses linear conversion, which probably explains why it was so inefficient, despite being a forward design.

Forward is also somewhat inefficient because the use of half-wave rectification on the output, but you can make up for that by overspecing lots. +3.3V has its own transformer pin... but isn't the amount of power on the +3.3V rail still in part limited by the +5V rail or is it different in a linear regulated design?

Forward is normally more efficient than Half-bridge, simply because it has much lower switching losses (even if the secondary side is somewhat less efficient). Good luck finding an 80plus compliant half bridge design. I'm sure it's possible, but it's not easy to do. There's a reason that all of the decent manufacturers have moved on to forward and LLC Resonant converters.

Normally, a DC-DC converted 3.3V rail is limited by what it can steal from the 5V rail without overloading it. In this case, however, it has a separate rectifier, so the limiting factor will probably be what the transformer can handle.

It's true. The switching losses of MOSFETs are substantially lower than that of BJTs, which is probably why single switch forward gets away with using only one primary switcher on a relatively small heatsink without too much heat output (though the FET would be in a TO-3P/TO-247 package, ideally).

And well, to my knowledge, +5V and +3.3V in older, group regulated designs at least, share the same transformer output, so you can't draw some power from the +5V rail without having less available on the +3.3V rail as well (which is the reason for the combined +5V/+3.3V rating - in a sense, they are more limited by the combined rating than by the full ability of the main transformer). I was wondering if the same applied for +3.3V linear regulated circuits (since it has its own transformer tap in this case)? Also, in the Hipro, is the +12V rail missing a PI filter coil or is it just hard to see in the images (if it is missing, I would imagine those two 2200uF 16V Teapo SCs would have to be in parallel to achieve such low ripple)?

Wester547 wrote:And well, to my knowledge, +5V and +3.3V in older, group regulated designs at least, share the same transformer output, so you can't draw some power from the +5V rail without having less available on the +3.3V rail as well (which is the reason for the combined +5V/+3.3V rating - in a sense, they are more limited by the combined rating than by the full ability of the main transformer). I was wondering if the same applied for +3.3V linear regulated circuits (since it has its own transformer tap in this case)?

That is correct. Either way, the 3.3V rail does have to share a transformer output with the 5V rail, hence the combined ratings.

On this unit, I don't know. I didn't unwind the transformer, and I don't want to have to do that, but unlike most, it does appear to be a separate output (although it does still have a combined 5V + 3.3V output rating). I still have a hunch that the transformer windings would limit it more than the 30A rectifier or 40A MOSFET used, though.

Wester547 wrote:Also, in the Hipro, is the +12V rail missing a PI filter coil or is it just hard to see in the images (if it is, I would imagine those two 2200uF 16V Teapo SCs would have to be in parallel to achieve such low ripple)?

It does indeed lack a PI coil on the 12V output. The coil just next to the 12V wires is actually used on the 3.3V rail. The two caps are indeed in parallel. That's still not a lot of ripple, though, considering what I'm used to seeing from PSUs with no PI coil. It goes to show that the design of a unit can have just as much of a bearing on the ripple as the output filtering used. The large toroid coil for a 300 Watter probably helps as well.

c_hegge (in the review) wrote:...bearings...

Only ball bearings, and tapered roller bearings (which can support heavy radial and axial loads, but an individual can only support axial loads in one direction), have to be used in pairs. Other bearings (sleeve, cylindrical roller, fluid-dynamic, etc.) can be used in pairs, to support the ends of a shaft that can't be supported from the center (one bearing on each end), but a single long one can also be used if placement allows. Fans like the one in question have only one bearing.

EDIT: Added the bit about tapered roller bearings. Neither cylindrical nor tapered roller bearings are used in fans or HDDs, however.

Here's what became of this PSU.