Power Supplies for Today's & Tomorrows Computers

I wish I had enough cash flow to buy each PSU on the SLI list.

Seriously, if any of those companies wanted to send one to me, I'd make sure they got 'em back. Not like I need anything like that. I've got everything working just fine with a 500W and you know the ol' saying "if it aint broke don't fix it?" I tend to screw things up if I start dorking around with stuff that doesn't need dorking. ;)
 
Spectre said:
"I am not sure that word means what you think it does"

Wow. I just looked up Dorking on Urban Dictionary and it certainly has a lot of definitions.

Of coruse, if you look it up on Dictionary.com, it only has one: "A heavy-bodied domestic fowl having five toes on each foot and raised chiefly for table use."
 
jonnyGURU said:
I'm on the fence on this one.

30A is a hell of a load. A dual core and two high end PCI Express cards will probably yield about 24A at full load. Any other components that will bring that load up (drives, cards, etc.) will also bring up the 5V... except for lights and fans, of course.. but those don't put much of a load on a PSU (2A to 4A each, right?)

If I only had a dual core Pentium 4 with a pair of 7800's I could say for sure.

Maybe I should give this PSU to Kyle when I'm done with it and let him plug it into something for a while. ;)
Johnny,

Thank you for doing these test, they are by far the most detailed I have ever seen for this PSU.
Given that there are alot of other SLI Certified PSU's that don't have this problem &
companies like PC P&C, OCZ, Antec & Enermax have much stronger histories &
reputations, I thnk the safest route is NOT to recommend the Zeus at this time.

Thank you again! :)
Dave
 
jonnyGURU said:
Wow. I just looked up Dorking on Urban Dictionary and it certainly has a lot of definitions.

Of coruse, if you look it up on Dictionary.com, it only has one: "A heavy-bodied domestic fowl having five toes on each foot and raised chiefly for table use."

Hehe.....
 
davidhammock200 said:
Johnny,

Thank you for doing these test, they are by far the most detailed I have ever seen for this PSU.
Given that there are alot of other SLI Certified PSU's that don't have this problem &
companies like PC P&C, OCZ, Antec & Enermax have much stronger histories &
reputations, I thnk the safest route is NOT to recommend the Zeus at this time.

Thank you again! :)
Dave

Dave, I don't understand what you are getting at. Where is the problem for the Silverstone Zeus that jonnyGURU's tested? It held the +12V rails within spec despite him runinng one of the rails out of spec. Do you have proof of other power supplies going through the same tests (30A on +12V with 4A on +5V) doing better?
 
tonyou said:
Dave, I don't understand what you are getting at. Where is the problem for the Silverstone Zeus that jonnyGURU's tested? It held the +12V rails within spec despite him runinng one of the rails out of spec. Do you have proof of other power supplies going through the same tests (30A on +12V with 4A on +5V) doing better?
+5V min. load is 10A when +12V output is 30A to 38A
+5V min. load is 15A when +12V output is 38A to 42A
http://www.dealsonic.com/sizesttibl65.html
"despite him runinng one of the rails out of spec"

This is the problem, none of the other SLI Certified PSU's have these "extreme cross loading specs".
Why recommend this "wierd one" that does & shows major loss of +12V when the +5V amperage usage is low?

Dave
 
davidhammock200 said:
"despite him runinng one of the rails out of spec"

This is the problem, none of the other SLI Certified PSU's have these "extreme cross loading specs".
Why recommend this "wierd one" that does & shows major loss of +12V when the +5V amperage usage is low?

Dave

That is a fair argument in theory, but you don't seem to have any problems recommending OCZ PSUs despite them not being SLI certified and exhibiting worse behavior when loaded in extreme conditions.
http://firingsquad.com/hardware/2005_eternal_battle_day3/page5.asp

It is apparent OCZ chose not to disclose their cross-loading requirements openly and you did not know about it (or else you would have recommend against it?). I assume you are certain that makers of all other SLI certified PSU have disclose this information as well?

But all specifications and talk aside, is there an example of a real-world PC that will actually draw more than 30A on +12V with +5V staying below 6A? Have you actually tried the PSUs in person for the ones you recommend for and against? We want a PSU that works in what it was designed to do (power an SLI rig), right? And the track record has so far showed the Zeus to be a good one powering many SLI and non-SLI systems (Spectre, I know I have no concret statistics to post here, but trust me on this one. ;) )
 
The cross-loading requirements of the Antecs & Enermax's are are minimal & posted on their websites,
both of these SLI Certified PSU's are cheaper & have more availability. The PC P&C 510's have no problems either.

Let's stick to what we know works. ;)
 
tonyou said:
(Spectre, I know I have no concret statistics to post here, but trust me on this one. ;) )

Uh-oh I think I am starting a reputation. :p
 
Spectre said:
Actually looks like in the end it missed us by a ways. On to the next storm!
Don't say that! :eek:

They are already tracking TS#5 & it's comming our way! :(

What all got messed up?
 
davidhammock200 said:
Don't say that! :eek:

They are already tracking TS#5 & it's comming our way! :(

That is why I said it. Such an active early season.

What all got messed up?

The usual lots of trees down, powerlines, houses flooded, roofs ripped off etc. 9 inches of rain to boot!
 
Spectre said:
That is why I said it. Such an active early season.

The usual lots of trees down, powerlines, houses flooded, roofs ripped off etc. 9 inches of rain to boot!
Well the important thing is that everyone is safe. :)
 
good post a hell of a lot of work must have gone into it (i know how tedious it can sometimes be having done the other sticky) but, ARRRRHHGHHHGHGH MY EYES

Kill the colors other than red for emphasis occasionally, this is hardforums not skittles website
 
computerpro3 said:
good post a hell of a lot of work must have gone into it (i know how tedious it can sometimes be having done the other sticky) but, ARRRRHHGHHHGHGH MY EYES

Kill the colors other than red for emphasis occasionally, this is hardforums not skittles website
I will, I just copied from EOC & their light blue goes well with the colors.

Thank's
Dave
 
davidhammock200 said:
The cross-loading requirements of the Antecs & Enermax's are are minimal & posted on their websites,
both of these SLI Certified PSU's are cheaper & have more availability. The PC P&C 510's have no problems either.

Let's stick to what we know works. ;)

I am sorry, but can you please show me where the links are? I could not find any info of cross-loading requirements on either Antec or Enermax's website.

I have no doubt about recommending Antec and Enermax SLI certified PSUs either but I don't understand why you would recommend against a PSU that has already proven to work great in SLI rigs (even holding within spec despite jonnyGURU intentially running it out of spec)? I apologize for having to bring this up again, but if you tell people not to get Silverstone for cross-loading issues then why recommend OCZ that has even worse cross-loading issue and is not SLI certified?
 
That would be a bunch of links, just look at their specs &
you will see both min & max for each rail, the min for +5V is reasonably low.

The "cross loading" charts/graphics, I'll try to find.
 
For the record, I for one am not going to discourage using this, or any other, Silverstone power supply right now.

I believe they're made by Enhance (I think it's a rebadged ENS-0565) and they're a company that's been around longer than most of them, and the crossload I performed was ridiculously.. well.. crossed.

That said, the drop in voltage was pretty dramatic. Sure, it was within spec, but very sudden over a short period of time.

Of course, not realistic either. 5V loads tend to be very static. But I have to wonder if the adverse would happen. Meaning: Would the voltages have such a dramatic fluctuation is I were to maintain a steady 5V rail and then incrementally crank the 12V up towards, and over, 30A.

I guess I'll just have to do that this upcoming weekend. :D
 
jonnyGURU said:
For the record, I for one am not going to discourage using this, or any other, Silverstone power supply right now.

I believe they're made by Enhance (I think it's a rebadged ENS-0565) and they're a company that's been around longer than most of them, and the crossload I performed was ridiculously.. well.. crossed.

That said, the drop in voltage was pretty dramatic. Sure, it was within spec, but very sudden over a short period of time.

Of course, not realistic either. 5V loads tend to be very static. But I have to wonder if the adverse would happen. Meaning: Would the voltages have such a dramatic fluctuation is I were to maintain a steady 5V rail and then incrementally crank the 12V up towards, and over, 30A.

I guess I'll just have to do that this upcoming weekend. :D
I'm looking forward to your continuing efforts with this "problem child" & to your TT review as well.

Thanks,
Dave
 
davidhammock200 said:
both of these SLI Certified PSU's are cheaper & have more availability.

You also forgot that the Zeus is a "650W" EPS12V PSU, whereas the SLI certified Antec and Enermax are at 550W and 600W respectively. In the PSU industry, more wattage cost money and there is usually a big step up in cost for PSUs in 400W range to 500W and to 600W+

If you look at how much Enermax is charging people for their 660W EPS12V PSU (not SLI certified btw), you'll see the Zeus is an excellent deal at around $176 compared to Enermax at $252 (prices from Newegg).
 
tonyou said:
You also forgot that the Zeus is a "650W" EPS12V PSU, whereas the SLI certified Antec and Enermax are at 550W and 600W respectively. In the PSU industry, more wattage cost money and there is usually a big step up in cost for PSUs in 400W range to 500W and to 600W+

If you look at how much Enermax is charging people for their 660W EPS12V PSU (not SLI certified btw), you'll see the Zeus is an excellent deal at around $176 compared to Enermax at $252 (prices from Newegg).
A very good point, anyone running an EPS multi-CPU system with SLI (SSI) would almost have to be drawing more than the min +5V load requirements, so these cross loading requirements would be a non-issue for them.

Dave
 
Ok... Here we go....

All numbers are 3.3V/5V/12V1/12V2.

5/5/12/12 = 3.34/5.16/11.84/11.81
5/5/15/15 = 3.33/5.18/11.76/11.74
5/5/15/18 = 3.33/5.19/11.72/11.68

As you can see, the voltage drop is very dramatic on the 12V rail once I go up to 30A on the combined 12V rail. That said: the wattage this PSU is putting out on the third test is 445W! Picture ANY PSU you may be currently using putting out 445W. Now picture what kind of computer you would have to build to put that kind of demand on a power supply. Ok... NOW picture how this machine would be built to only have a 5A load on the 5V rail. NOT realistic, IMHO.

The testing continues with me now increasing the voltage on the 5V rail while maintaining the 33A load on the combined 12V's.....

5/7/15/18 = 3.33/5.15/11.78/11.73
5/8/15/18 = 3.33/5.13/11.80/11.76
5/9/15/18 = 3.33/5.12/11.84/11.79
5/10/15/18 = 3.33/5.12/11.84/11.80
5/11/15/18 = 3.33/5.13/11.84/11.81

As you can see, as soon as I bumped the load up 1A on the 5V from 7A to 8A, things started getting normal. And once I was at and above Silverstone's recommended 10A minimum, everything was as stable as it was when there was only a 24A load on the combined 12V rail.

So I'm perfectly happy with this power supply. SMPS power supplies ACT this way. That's what they do when they're crossloaded. I think it's unrealistic to expect a PC that's potentially going to put a 30A or better load on the 12V rail to not have at least a 10A load on the 5V rail. With a pair of dual-cores, a pair of PCI-e video cards, you're only going to be up around 24A worst case scenario. Pretty much anything else you put on the 12V rail is also going to load the 5V rail as well, easily getting you over that 10A hump.

I actually have to commend Silverstone on publishing what their crossload tolerance is. OCZ never did this. And when I emailed them about the problem I had with their power supply, they ignored me. I had to confer with Xbit lab's review to confirm I didn't have a defective power supply!
 
computerpro3 said:
good post a hell of a lot of work must have gone into it (i know how tedious it can sometimes be having done the other sticky) but, ARRRRHHGHHHGHGH MY EYES

Kill the colors other than red for emphasis occasionally, this is hardforums not skittles website
I'll be fixing the colors this weekend when I update everything. ;)

jonnyGURU said:
Ok... Here we go....

So I'm perfectly happy with this power supply. SMPS power supplies ACT this way. That's what they do when they're crossloaded. I think it's unrealistic to expect a PC that's potentially going to put a 30A or better load on the 12V rail to not have at least a 10A load on the 5V rail. With a pair of dual-cores, a pair of PCI-e video cards, you're only going to be up around 24A worst case scenario. Pretty much anything else you put on the 12V rail is also going to load the 5V rail as well, easily getting you over that 10A hump.

I actually have to commend Silverstone on publishing what their crossload tolerance is. OCZ never did this. And when I emailed them about the problem I had with their power supply, they ignored me. I had to confer with Xbit lab's review to confirm I didn't have a defective power supply!
Thank you Johnny! You have answered the big question.

This will be posted here, at EOC & DFI-Street:

Thanks to the exhaustive testing performed by "jonnyGURU",
the SilverStone Zeus ST65ZF 650W can now be fully recommended
to anyone needing a high performance PSU for their high demand EPS &/or SLI workstation, server or PC.
Thank you again,
Dave ;)

EDIT:

Posted:

http://forums.extremeoverclocking.com/showpost.php?p=1615995&postcount=234

http://forums.extremeoverclocking.com/showpost.php?p=1721731&postcount=401

http://www.dfi-street.com/forum/showthread.php?p=101777#post101777

As well as the DFI & SLI portions of the Hard PSU Guide.

Thank you again,
Dave
 
davidhammock200 said:
The cross-loading requirements of the Antecs & Enermax's are are minimal & posted on their websites,

Dave, you are not off the hook yet. I want to see those cross-loading information you said that were posted on Antec & Enermax's website for their SLI PSU! ;)
 
And for the record; I resolved my quad rail power supply testing methodology issues.

Using my new Fluke 336, I was able to determine that if I plugged two rails into one load; that the load was distributed evenly across the two rails.

For example: If I had 12V1 and 12V4 plugged into the 12V1 load on the tester, and the load on the tester's 12V1 was 30A, then the load on 12V1 and 12V4 of the power supply would be 15A each.

The difference in resistence between the load tester interface and the power source is so negligible (if existant at all) that the load is literally divided evenly between each wire lead.

I don't know how I lived without one of the clamp-type ammeters in the past. :D
 
tonyou said:
Dave, you are not off the hook yet. I want to see those cross-loading information you said that were posted on Antec & Enermax's website for their SLI PSU! ;)
Reread what I said.

I said min requirements were listed, but I did find cross loading graphics somewhere,
I'll link or post them when I see them again.

Dave ;)
 
davidhammock200 said:
I'll be off-line for a few days taking care of some family business.

See you soon,
Dave ;)

Be safe. See you soon.
 
I'm half tempted to give this Silverstone a 10. A 9.5 at the lowest.

I had this puppy up to around 670W (980W at the wall!). The lights in the house were starting to dim so I aborted.

Voltages didn't move. PSU didn't get hot. It was surreal. It actually scared the hell out of me. :D

Can't wait to write this one up. :p
 
jonnyGURU said:
I'm half tempted to give this Silverstone a 10. A 9.5 at the lowest.

I had this puppy up to around 670W (980W at the wall!). The lights in the house were starting to dim so I aborted.

Voltages didn't move. PSU didn't get hot. It was surreal. It actually scared the hell out of me. :D

Can't wait to write this one up. :p

But did the neighboors call and complain about a brown out?
 
Hmm... looks like its efficiency is below 70% at more than full load, it should be better. Btw, how many times have you ran that Silverstone out of spec? I am a little surprised it didn't shut itself down to protect itself.
 
Actually, no. It always maintained efficiency of at least 70%. When it was run under normal circumstance it was puting out 76% efficiency. The number specified above for what was measured at the wall was just off the top of my head and may not be exact.

The review is done and uploaded. I'm just waiting around for Steve over at SL Central to publish it. Hopefully he'll get it up some time this week. :rolleyes:

I didn't run the PSU out of spec much at all. The only thing I did to the PSU that was "out of spec" was the low 5V crossload (< 10A on the 5V when the 12V > 30A.)

When I did crank the PSU up beyond it's "650W rating" each individual rail was still within it's spec.

The breakdown of the load on each of the rails will be in the review. We'll just wait for that to come out and you can see all of the details of my testing methodology.
 
You didn't seem to mention who makes this PSU, is it an Etasis? I think they produce Silverstone's passively-cooled units, where Sirtec does the more traditional models.
 
Back
Top