My Dual Harper - Asus Z7S WS living review *56K warning*

[Syndil]

Sounds logical. However, the warranty is 3 years, not 2.

Are faulty VRMs the cause of all of the Z7S failures?

[jcool]

As far as I can tell, yeah.

[STEvil]

And Asus refuses to do anything about it.

:/

[Syndil]

I am investigating a theory on why the VRM may be failing. It's just a theory, but based on the info in this thread and elsewhere, the logic is at least plausible.

IIRC, there was a photo a few pages back that someone had taken of the actual VRM that had burned up. I can't find it now, naturally, but it was the VRM under Heatsink 1, nearest CPU0.

Here is a photo of the row of VRMs that live under Heatsink 1:



This photo is from a blog of an attempt to convert Heatsink 1 to home-built water cooling.

If I am remembering correctly, the rightmost VRM in the photo (under the thermal pad) is the one that burned up.



The Z7S is an SSI CEB form factor motherboard. While it is based on the ATX specification, it is not an ATX board. Specifically, the maximum size for a full ATX board (not extended or EATX) is 12" x 9.6" while SSI CEB is 12" x 10.5", and the SSI CEB specification calls for mounting points for the CPU coolers.

However, since it is based on the ATX specification, the SSI CEB Z7S can fit in many standard ATX cases with some slight modification to provide the needed mounting points for the CPU coolers. Thus, the Z7S comes with a backplate kit for mounting the CPU coolers in a case that was not designed for an SSI CEB motherboard. The Z7S does not, however, come with instructions on how to actually mount the kit. This is a picture of the kit that came with the motherboard:



It consists of a thin plastic template, two steel adhesive-backed plates, and three baggies consisting of two rubber spacers and eight standoffs a piece, with each baggy containing spacers and standoffs of a different height.

And here are the directions that probably should have come with the kit:

1) Place the template on your motherboard tray, lining up the holes in the template with the standoffs in the top right corner of the tray.

2) Remove the backing from the adhesive on one of the steel plates, and, using the template as a guide, stick it to the motherboard tray. Repeat for the second plate.

3) Find the baggy containing the standoffs that, when screwed on to the plate, match the height of the standard ATX standoffs already in your case.

4) Once you have determined which baggy to use, secure the eight standoffs to the plates, using an adjustable crescent wrench to ensure a tight fit.

5) Now remove the backing from the adhesive on the rubber spacers from the same baggy and stick them onto the metal plates, placing them within the guide lines etched into each plate. The rubber spacers prevent the motherboard from being flexed towards the motherboard tray when mounting the CPU coolers. They perform the same important function as the CEK spring clips used when mounting the board to an SSI CEB case.



Here is what it should look like. This is a picture of my old Cooler Master ATC-210, in which my Z7S was mounted for over a year:





I had to buy another of these kits from ASUS in order to mount the Z7S in its current case, a Cooler Master Cosmos. No amount of reasonable force could have removed the plates from the old case without damaging something. The adhesive is quite strong.

The theory is that perhaps the mounting of the cooler for CPU0 is flexing the board enough to pull the rightmost VRM from its heatsink, causing it to wear out faster than it should. An SSI CEB motherboard could be especially vulnerable to flexing caused by the CPU coolers, since the coolers are mounted directly to the tray beneath the motherboard, rather than the motherboard itself. Forgoing the use of the rubber spacers could exacerbate the problem.

Of those of you with failed Z7Ss, how many of you did not use the rubber spacers?

[jcool]

I use waterblocks with backplates, so no, I don't think this is due to the board flexing. I even removed the VRM sinks and checked for proper contact.. which is does make. Just low quality VRMs I'd assume.

[STEvil]

No flex at all in my installation.

[Breit]

i've just a quick question for someone who actually has a working z7s, jcool?

question is: what is everest saying on your board about the ram channel configuration?
i noticed a somewhat bad ram performance on my z7s and did a quick benchmark check with everest. surprisingly everest says nothing about quad-channel nor anything like that. it just says 'ddr2-820fb' (i'm running at 410mhz fsb) and not 'quad ddr2-820fb' as it should be. i've populated all 6 ram-slots with 1gb kingston ddr2-800 fb-dimms each. this should be a valid ram configuration for quad-channel mode on an fb-dimm system even though the latencies might not be optimal due to 2 modules on 2 of the 4 channels and the serial nature of the fb-dimm system. i also tested it with only 4 modules but same story here...
is this just a wrong reading from everest or is there something i might have missed? all the benchmarks are pretty inline with the reference systems using the same hardware but ram performance seems to suck on this board.

[jcool]

As far as I know it's a software quirk. I do not know of any software to date that can read quad channel ram interfaces. Not even CPU-Z will report it...
Doesn't show for me either, and I'm running 4 sticks in the right slots

[Breit]

maybe true, but unbelievable anyways...
it can't be that hard to detect a quad channel memory controller for a hardware analysis tool?! what really makes me wonder is that for the reference system in everest (it's an e5462 i think) it clearly says 'quad ddr2-640fb' so obviously there must be some mechanism to detect a quadchannel controller (or maybe its just a manual entry in the database... ). i'm not exaclty sure where the 640 is coming from though, it should be 667, but who knows?!

[milkmandan]

oh man with all these reports of dead z7s ws boards, i am seriously consider stripping my rig and selling the mobo and cpus.

i dont do number crunching or leave it on 24/7 like a lot of the people here, i've had this rig for what...1.2 years now, and I wouldn't want it dying on me any time soon

spent a heafty bit, wouldn't want to see it go bad.

[jcool]

Well if you sell it, you offload the problem to someone else. Your choice if you wanna do that or not...

[Jaco]

I guess you're safe if you're not crunching 24/7 .

[milkmandan]

everyone here who has been crunching 24/7 has seen these things die quickly.

i know this is farfetched question, but what kind of life expectancy should i be expecting for someone that does not crunch ?

has anyone. 'not' crunching, daily average use, have this board die yet?

--

pretty sure if i took the advantage and sold this off early, i could just get a nice i7 board and cpu for the return.

[RomanZen]

has anyone. 'not' crunching, daily average use, have this board die yet?

hi fellas, i'm following this thread since a year. helped me much when i went for dual cpu opwer in a workstation. best cooling tips! you do great!
anyway,
the board just died with burned VRM some days ago when i switched the octo:banana::banana::banana::banana::banana: back on after 2 weeks holiday.
before the pc was running 24/7. not under load 80% of the time.
i'm using it for rendering, working with 3D applications and a bit of gaming.
board is (was) a bit over 1 year old. it may rest in peace!
saved me a lot of time, now it's taking it back while getting replaced.

cheers.

[s0lid]

So it makes, 14 out of 16

[Syndil]

the board just died with burned VRM some days ago when i switched the octo:banana::banana::banana::banana::banana: back on after 2 weeks holiday.

Can you describe when exactly in the boot process it burned up? Did you get in to Windows?

[RomanZen]

not to 100%

first try
bios bootup was fine. windows 7 logo appeared while booting. pc was hung up in boot logo.

second try
i resetted the beauty, left the room for drinks, came back, only black screen. it smelt of burned electronic.
after that i pluged in the diagnosis card. the card said FF which apperas to doesn't mean fully functional .

after professional investigation i found similar problems with the vrm's and the smell did come exactly from the heatsink above them.

the pc was completely off power grid while i was on vacation. maybe that helps.

i hope it helps,

asus FF wins \o/

cheers!

[Syndil]

It could be that the heating/cooling of the VRM is performing unintended metallurgy on the components of the VRM itself. That may explain why it quickly burned up after 2 weeks of no usage.

This of course could be the fault of a VRM manufactured with poor quality materials, or it could be the fault of a poorly cooled VRM. Or both.

Roman, did you happen to notice which VRM failed? Was it the rightmost VRM under heatsink 1?

[STEvil]

*edits signature again*

You guys realize that at $550 per board (some of us paid $600+ i'm sure..) that is almost $10,000 in broken boards?

[Movieman]

Guy's:
I've been in contact with a rep from Asus over the last week.
I suggested to him that to solve this issue it would be in Asus and the guys best interests to offer all of you an upgrade to a dual socket 1366 board.
The only other option would be to pay each of you what you spent on the board and I don't see that happenning.
He has agreed to "kick it upstairs" to see how Asus mgt feels about it.
I would hope that although not a perfect solution as cpu's and memory are not transferable between socket 771 and socket 1366 it would be a solution that you guys could live with.
I'm going to start another thread for those that own dead and have lost Z7S WS boards to post in and then PM me their names and mailing info.
Let's see if we can bring this to a close and in a way that if not perfect, at least all can say, 'OK. thats reasonable".
http://www.xtremesystems.org/forums/...d.php?t=239229

[RomanZen]

sorry for the poor quality. the quickshot from the mobile should do the trick anyway. as you can see it's the second from the left.

http://picfront.de/d/FP4GwMLGw/IMG00...91119-1211.jpg

if you get news please post them here. i whould like to see a general solution for this and not having another board replaced.
for me switching to socket 1366 isn't really an option since i simply don't have another 800 to 1500€ waiting to be spent for a new cpu set.

edit: for cooling i used 6 120mm pwm fans. 3 for the case, 3 for the cpu's in a lian li b25. the 2 stock coolers from the xeon's for the ram.

cheers

[Syndil]

Thanks for the pic, RomanZen. Doesn't appear to be any rhyme or reason to it then. Just bad VRMs.

How about the DSEB-DG as a replacement? Different slots, less USB ports, and much larger (SSI EEB vs SSI CEB), but the Z7S peripherals should work fine on it. Doesn't necessarily appear to be a step down from the Z7S (as long as the previously mentioned differences don't bother you), or am I missing something?

[jcool]

Well, it can't OC. Or not like the Z7S WS can. No other board can do that....

[Syndil]

Oh, OC. I don't do that anyway, but I can understand why that would be important to members of this particular forum. I would like to be able to adjust the timing on my RAM, however, which apparently the Z7S can't do. :P

If (or when, if you want to put it that way) my Z7S fails, I will see if ASUS will send me a DSEB-DG.

[jcool]

Only Dual LGA771 to allow for manual ram timing control is the Skulltrail.