Silverstone ST75ZF review

[jonnyGURU]

http://www.jonnyguru.com/PSU/ST75ZF/

You always know when I like a product when it's more than one page.

 

[troisanh]

Thanks for the review, answered some of my questions about it.

 

[[Spectre]]

The review still has that issue I pm'd you about

 

[jonnyGURU]

Going back and forth between my desktop and laptop, I think I keep uploading over my own changes.

 

[unhappy_mage]

Looks cool. I think I understand the o-scope readings, but it'd be nice if you could put labels on the axes nonetheless.

 

[jonnyGURU]

Thanks.

If you click on the Oscope thumbnail, it brings up a bigger image. The scale is in the lower right corner: .05V / 2ms.

I only really break it down if the results are disastorous.

 

[AreEss]

Yeah. If you need tight regulation, the ST75 is not for you. Nor is any other Etasis built PSU, generally speaking. They're typically 5% regulation, and they will swing around there pretty good. Especially the hotswaps.

 

[unhappy_mage]

It says .05V/div. Is that for the whole chart height, or for the height of one "block"? What about the width? Last, why only 25 KSa/sec? If the whole window is 2ms wide, and about 500px wide, that's enough room for 250 KSa. Even if one block is 2ms, that means one pixel = one sample. Oversampling and then overdrawing samples would show more of what's going on with the voltage fluctuation. If you can show what the absolute values are along the side as well as the waveform, that'd be helpful. If it's not practical to do all that, it's all good, but I'd like to see a really good PSU review site turn into a really great one

Also, maybe for the higher-quality units where the voltage regulation is pretty good there could be a higher-magnitude graph so one can compare among them? For example, on most of the graphs presented it looks like the whole graph could be expanded so that the middle .1V takes up the whole graph vertically.

 

[snowwie]

Yeah. If you need tight regulation, the ST75 is not for you. Nor is any other Etasis built PSU, generally speaking. They're typically 5% regulation, and they will swing around there pretty good. Especially the hotswaps.

Eh, what?

In his review he states that it has some of the best line regulation he's ever seen

and for the record, I believe the scope readings show ripple/noise, NOT line regulation. Line regulation is how the voltage changes at different loads. I think.

 

[454Casull]

Eh, what?

In his review he states that it has some of the best line regulation he's ever seen

and for the record, I believe the scope readings show ripple/noise, NOT line regulation. Line regulation is how the voltage changes at different loads. I think.

Line regulation is how the voltage changes depending on the input voltage IIRC. Load regulation is how the voltage changes depending on load.

 

[snowwie]

ahh

 

[AreEss]

Line regulation is how the voltage changes depending on the input voltage IIRC. Load regulation is how the voltage changes depending on load.

Correct.
Etasis power supplies generally run 5% on load and crossload, though certain hotswap models go as wide as 10%. Presuming links work (they usually don't with Etasis, sigh) let me show you some pretty numbers.
http://www.etasis.com.tw/cgi-bin/pr...5=Active&key2=EFRP2553&action=searchdbdisplay
Line regulation 1%, load regulation +-5%, +9%/-5% on -12V, cross regulation same numbers.
ST75, er, I mean ET750.
http://www.etasis.com.tw/cgi-bin/pr...d&f5=Active&key2=ET750&action=searchdbdisplay
Take a look; load and crossload are +-5% for all rails.
Now, compare those to the PCP&C 850SSI - 2% for all rails except -12V, that's a whopping 5%. The 510XE packs an impressive 1% for +3.3, +5, +12 and 5% for -5 and -12; that's better than spec requires. It's also the best regulation outside of high-end linear units. (A 350W +12V linear power supply with +1% (NOT +-, only HIGH) costs more than a 510XE.)

Now, does that mean that the ST75ZF/ET750 is a bad PSU? No. It's definitely a solid PSU; Etasis makes some excellent units, and I've never had any problems with them. What it does mean is that if you're in a situation where tight regulation matters, there's a chance you might see voltage fluctuating. Or if you have electrically noisy boards, fans, or drives on the 12V rail. It's not out of spec by any stretch of the imagination, nor is it atypical for most PSUs. It's only going to matter in situations where components are sensitive to regulation or voltage swings.
Unfortunately, I'm in the position where the stuff I put together that needs 750W, also needs 3% or better regulation under load and crossload.

 

[tonyou]

Correct.
Etasis power supplies generally run 5% on load and crossload, though certain hotswap models go as wide as 10%. Presuming links work (they usually don't with Etasis, sigh) let me show you some pretty numbers.
http://www.etasis.com.tw/cgi-bin/pr...5=Active&key2=EFRP2553&action=searchdbdisplay
Line regulation 1%, load regulation +-5%, +9%/-5% on -12V, cross regulation same numbers.
ST75, er, I mean ET750.
http://www.etasis.com.tw/cgi-bin/pr...d&f5=Active&key2=ET750&action=searchdbdisplay
Take a look; load and crossload are +-5% for all rails.
Now, compare those to the PCP&C 850SSI - 2% for all rails except -12V, that's a whopping 5%. The 510XE packs an impressive 1% for +3.3, +5, +12 and 5% for -5 and -12; that's better than spec requires. It's also the best regulation outside of high-end linear units. (A 350W +12V linear power supply with +1% (NOT +-, only HIGH) costs more than a 510XE.)

Now, does that mean that the ST75ZF/ET750 is a bad PSU? No. It's definitely a solid PSU; Etasis makes some excellent units, and I've never had any problems with them. What it does mean is that if you're in a situation where tight regulation matters, there's a chance you might see voltage fluctuating. Or if you have electrically noisy boards, fans, or drives on the 12V rail. It's not out of spec by any stretch of the imagination, nor is it atypical for most PSUs. It's only going to matter in situations where components are sensitive to regulation or voltage swings.
Unfortunately, I'm in the position where the stuff I put together that needs 750W, also needs 3% or better regulation under load and crossload.

I think using manufactuer's claimed spec to compare power supplies may not reveal the whole story. The industry standard generally requires 5% so Etasis could simply be very conservative with their regulation specifications as I have yet to see their 750W come close to 5% on the main rails (in fact, the one in jonnyGURU's review did not vary more than 2%).

 

[AreEss]

I think using manufactuer's claimed spec to compare power supplies may not reveal the whole story. The industry standard generally requires 5% so Etasis could simply be very conservative with their regulation specifications as I have yet to see their 750W come close to 5% on the main rails (in fact, the one in jonnyGURU's review did not vary more than 2%).

It doesn't always reveal the whole story, no. I'm basing my analysis based on owning and installing more than a few Etasis power supplies. Under crossload, they tend to go the full 5% swing, though how often and sharply obviously varies dependent on the power supply. Look at Test 4; under just moderate crossload, the 12V rail dramatically destabilizes.
http://www.jonnyguru.com/PSU/ST75ZF/Oscope/Test4-EPS+12V.jpg
Those wide tall spikes are indicative of significant repeating swing, which is very bad for certain components which are sensitive to it. As you can see in Test 5's results, the oscillation remains significant in comparison with the futher increased 5V load. Just four more amps, and it'll likely be swinging out to 4-5%. Especially since I have to take into account the following loads; EPS12V @ 18A, 12V2 @ 18A, 12V3 @ 18A, 12V4 @ 18A, 5V @ 22A. Under those load conditions, nearly every power supply on market falls apart. The PCP&C 510XE sticks within 2% though when pushed to the limits.

Now, is this a downside to Etasis? For extreme high end, sure. Like I said; where tight regulation matters, Etasis in general is a bad choice because of their regulation properties. But for most home systems? I certainly wouldn't hesitate to recommend the ST75ZF myself - it's a very solid PSU. Just like every Etasis.

 

[jonnyGURU]

It says .05V/div. Is that for the whole chart height, or for the height of one "block"? What about the width?

Each "block" across and vertically is representative of .05v and 2ms.

Last, why only 25 KSa/sec? If the whole window is 2ms wide, and about 500px wide, that's enough room for 250 KSa. Even if one block is 2ms, that means one pixel = one sample. Oversampling and then overdrawing samples would show more of what's going on with the voltage fluctuation. If you can show what the absolute values are along the side as well as the waveform, that'd be helpful. If it's not practical to do all that, it's all good, but I'd like to see a really good PSU review site turn into a really great one

It's practical. Just not where I started and I would like to have relative graphs for side by side comparisons.

Let's just say, there's some REALLY bad power supplies that take up a lot more room on that graph.

Also, maybe for the higher-quality units where the voltage regulation is pretty good there could be a higher-magnitude graph so one can compare among them? For example, on most of the graphs presented it looks like the whole graph could be expanded so that the middle .1V takes up the whole graph vertically.

Again, for side to side comparison. I'd like to have two graphs from two power supplies with the same scale so a direct relationship can be made.

...Look at Test 4; under just moderate crossload, the 12V rail dramatically destabilizes.
http://www.jonnyguru.com/PSU/ST75ZF/Oscope/Test4-EPS+12V.jpg
Those wide tall spikes are indicative of significant repeating swing, which is very bad for certain components which are sensitive to it. As you can see in Test 5's results, the oscillation remains significant in comparison with the futher increased 5V load. Just four more amps, and it'll likely be swinging out to 4-5%. Especially since I have to take into account the following loads; EPS12V @ 18A, 12V2 @ 18A, 12V3 @ 18A, 12V4 @ 18A, 5V @ 22A. Under those load conditions, nearly every power supply on market falls apart. The PCP&C 510XE sticks within 2% though when pushed to the limits.

True. If you look at all of the reviews with the screenshots of the O-scope, they almost ALL fall apart. Does the Turbo-Cool look that bad under those loads? Don't know. Don't have one to throw on the tester. But you can get a lot worse. In fact, the reason the OCZ GameXstream review isn't done is because Eric Ryder wanted me to send it back because I was pretty much using the whole graph. Eric thought it HAD TO BE a bad unit. Hopefully the RMA process is swift and I can redo the tests with better results.

You want to see full use of that graph? Let me grab a Powmax and throw it on the tester.

 

[Oklahoma Wolf]

Hopefully the RMA process is swift and I can redo the tests with better results.

Either that or my 600W shows up soon

 

[jonnyGURU]

Did you actually ship that?

I hope you have tracking.

 

[Oklahoma Wolf]

Yup - Canada Post takes a while, but you should see it fairly soon by now. It's been travelling for two weeks. Even sent you the tracking number (not that it'll tell you much)

 

[jonnyGURU]

I think using manufactuer's claimed spec to compare power supplies may not reveal the whole story. The industry standard generally requires 5% so Etasis could simply be very conservative with their regulation specifications....

Actually, this is very true and consistant with most of the different ratings you see on a power supply.

For example: Even when the ST75ZF showed radical fluctuation during test four, this fluctuation never exceeded .1V. That's only .8%. NOT 5%.

Hypothetically; if PC Power and Cooling's rating is 2% and that's actual, assuming that the testing methodology is similar to mine, then that would be worse than the Etasis built Silverstone.

What I've seen is manufacturers tend to be much more conservative when it comes to ratings. A manufacturer can say 5% regulation, 75% efficiency minimum and 800W at 115V input, but if the PSU actually exhibits 2% regulation in tests, 85% efficiency maximum and 1000W at 230V input, you betrter believe that the MARKETING TEAM that slaps the label on that puppy is going to go for the shinier ratings. ;-)

I've always hated the fact that PSU ratings are very subjective.

For example: Seventeam rates their ST-750EAD as a 750W based on 115V input, continuous output. MGE relabels it as a 900W based on 230V input, peak output. And Andyson 550W that's 550W with 115V input, becomes a 620W Aerocool using the same subjective reasoning. Coolmax's Green 1000W is only 900W with 115V input.

A manufacturer is going to be more conservative with their ratings than a marketing team like PC Power and Cooling, MGE or Andyson.

 

[AreEss]

True. If you look at all of the reviews with the screenshots of the O-scope, they almost ALL fall apart. Does the Turbo-Cool look that bad under those loads? Don't know. Don't have one to throw on the tester. But you can get a lot worse. In fact, the reason the OCZ GameXstream review isn't done is because Eric Ryder wanted me to send it back because I was pretty much using the whole graph. Eric thought it HAD TO BE a bad unit. Hopefully the RMA process is swift and I can redo the tests with better results.

Actually, no; that's why I love the 510XE's. PCP&C used to happily advertize them as garaunteed capable of handling any system with up to 12 hard drives. Even with that much motor load, the 12V and 5V rails just would not move. At all. In one system the addition of a choke was necessary, but that was due to noisy motors, and it still didn't exceed 2%. That said, I can garauntee you the ST75ZF in the same system, would be swinging very wide from the combination of load and noise. Something ATE's cannot simulate to this day is real world noisy components. Sure, you can load it half a million different ways, but you can't throw a dozen noisy motors into the mix. And in a lot of cases, that's where you really kill PSUs.

Like I said; I wouldn't hesitate to recommend the ST75ZF in most applications. But you can bet that it's sensitive to noisy motors and will swing under heavy loads, like everyone else. The only PSUs I've ever seen that are utterly stable under extreme loads no matter what are, surprise, the PCP&C 510's, and only the 510's.

You want to see full use of that graph? Let me grab a Powmax and throw it on the tester.

Now now, no reason to go burning down your house.

 

[Oklahoma Wolf]

I can garauntee you the ST75ZF in the same system, would be swinging very wide from the combination of load and noise.

Would be interesting to see an additional test with a dozen or so hard drives replacing the load tester on 12v3 and some of the 5v. Had I a scope, I'd go play with the ST56ZF... I'm sure I have enough hard drives lying around, new and old.

That said, I would think Zippy has a model or two capable of handling noisy loads. Of course, they cost more than an Etasis too.