SilverStone Upgrades ST30SF To V2.0, May Be Using A Different OEM

[Guest]

SilverStone released the second version of the ST30SF SFX unit, featuring a larger, quieter 92mm fan. The rest of the changes include different power specifications and the removal of the semi-passive mode.

SilverStone Upgrades ST30SF To V2.0, May Be Using A Different OEM : Read more

 

[Daniel Sauvageau]

I wouldn't make such a big deal out of the lack of over-power protection: as long as the output voltage and current limits work correctly, the worst thing the lack of OPP will do in an otherwise properly built PSU is cause the transformer to saturate and blow a fuse/FET.

Decent PSUs will at least sense the primary side current and tie that into the error amplifier circuit to do per-pulse current limiting, which provides enough protection to prevent the PSU from self-destructing.

BTW, I have one of these with me from a Silverstone "surprise box" I got a few weeks ago, just popped it open to have a quick look. It is a cute little critter and has a completely different internal layout from the previous version. Nichicon primary cap, Chemi-con 5VSB output caps, Teapo [strike]SC[/strike] TC just about everywhere else. Uses a Champion CM6806 combination APFC-PWM controller and a WT7527 monitoring IC.

 

[Aris_Mp]

OPP senses the primary side's current (to be more specific this happens in the APFC converter). When there is no OPP or a bad configured OPP threshold then the PSU can blow any time under higher than its nominal capacity loads when OCP is bad configured and UVP is absent. Also as I noted this PSU doesn't have UVP so it won't sense when its output rails are highly stressed and go way too low (a typical example when a PSU is about to die due to overload). Finally, in most PSUs OCP is set way too high, on all rails, in order to cope well with transient loads.

I have blown many PSUs under overload conditions, even some that on papers at least had OCP and UVP.

 

[Daniel Sauvageau]

Also as I noted this PSU doesn't have UVP

The WT7527 monitoring chip does do UVP across all three positive rails and does not require any additional external components to do so, which means the ST30SFv2 has it even though it isn't mentioned on the packaging. With the number of features integrated in switching and monitoring ICs, I think PSU manufacturers/vendors are simply getting tired of listing every single feature supported by the chips used in their products and are skipping a few since they provide negligible marketable value. BTW, UVP is actually mentioned in the ST30SFv2's white specs manual (part# G11228050) on page 05 but that one only covers 3.3V and 5V with much lower thresholds than the WT7527's minimum spec. Something funny might be going on there.

Also, the ST30SFv2 uses DC-DC converters for 3.3V and 5V, which means that 3.3V and 5V power comes out of the 12V rail's OCP budget. That's another effective way of limiting output power. Unless the PSU has been poorly designed, its components should be nowhere near within a momentary peak or short circuit from self-destruction - the ATX power specification itself requires that PSUs gracefully handle those situations by shutting down. If you manage to destroy a PSU this way, then it isn't truly ATX-compliant.

 

[Aris_Mp]

I have destroyed many PSU this way, which claimed ATX compatibility. Also having UVP is one thing, having it configured correctly is another. Finally, UVP isn't required according to ATX spec. However in any case voltages shouldn't go below the lowest limit set by the load regulation table.

According to the WT7527 spec sheet the UVP min. level at +12V is 10.3V, the typ. is 10.65 and only the max is 11.0. Even if we go with the max (11V) we are still far away from the 11.4V that the ATX spec (Revision 1.31) sets as minimum limit for this rail even under peak load (it used to allow 10% deviation in older revisions). According to my experience so far if you push hard a PSU and drop the +12V rail at 11.0, then it is a matter of minutes (or even seconds) till it blows sky high. Especially under high ambient temperatures.

OCP on the minor rails is typical and most units have it. OCP at +12V is something that not all PSUs have, or they just have it set very high in order to actually disable it. Also I don't agree that OCP at +12V is an effective way to protect the minor rails (in the unusual case that they don't have OCP), because if you have low load at +12V you can practically push way too high the minor rails and still keep the +12V rail in spec. The +12V FETs might be handle the load but this won't be necessarily the case for the minor rails' FETs.

The ATX spec (sec. 3.5) only requires, when it comes to protection features: OVP, SCP, NL, OCP and OTP. So if you go strictly with ATX you still don't have OPP and UVP. Furthermore, OPP and UVP need to configured properly to provide effective protection.

For you the lack of OPP might not be a big thing, however for me and my experience (>500 PSU evaluations so far) it is. Especially since last week I blew sky high a very expensive PSU because its OPP wasn't configured properly (was set too high). Actually I expect every PSU to be equipped with the necessary protection features. Also even if we take your worst scenario in mind, the destruction of a FET is still a major disaster for the average user. Moreover, when a component like a FET (or a bridge rectifier) dies, it will most likely take some more along with it.

In any case, I don't have anything else to comment on this matter and to be frank I don't even have the time to deal with it any more. But I do want all of my readers to be properly informed (and not misinformed).

 

[Daniel Sauvageau]

Claiming to meet a standard's specifications and actually meeting them are two completely different things. Unless the unit carries a credible certification stating that it meets a given specification or comes from a vendor with a credible history of in-house certification, then you should assume that it does not actually meet the spec until proven otherwise.

It isn't UVP's job to protect the PSU against gross overload, that's the OCP/OTP/OPP and the primary side's per-cycle current limit's job. UVP/OVP are there to shut down the PSU in case the power supply loses regulation, such as a fault in a feedback circuit causing one of the outputs to drift grossly out of spec.