Cooler is not fixed on 4870

[nicsimi]

Hello,
I've installed a Musashi VGA cooler on my HD4870 video card and the cooler is not fixed correctly on the videocard (I've tried to show the cooler movement in the attached picture).
The cooler is cooling correctly the VGA card (comparable temperatures with what I've seen on other reviews), the single issue is with this movement of the cooler.
I've used the instructions during the installation, did I missed something?
Thanks for your support!

[hiroshige]

Hello nicsimi,

a little movement is normal because the screw-holes of the VGA Card are a bit larger than the Musashi screws - if the temperatures are fine, there is no problem.

Cheers,
Stefan - Scythe EU Hamburg

[nicsimi]

Hello Stefan,
Thank you for your support.
I will use something (maybe some rubber grommets) to avoid this movement.
Thanks and regards.

[nicsimi]

What I really forgot to say is that this is a design flaw, and I really don't like to fix by myself a design flaw for a final product released to the market (I have payed a lot of money for a product plagued with design flaws - see "product liability").
This is my first Scythe product I'm buying and for sure is the last.
I've used a lot of aftermarket vga/cpu coolers which did not suffer for such a design flaw.
In my opinion, Scythe should hire some people wich can see a product from their possible customer's "eyes", otherwise in the long run they will fail for sure.

[Shinigami]

Hi nicsimi,

Thank you very much for your suggestions and patience.

The movement of the VGA Cooler is nothing you can call bad or a flaw. It is a design aspect, because this VGA Cooler fits a lot of VGA Cards and some cards have smaller mounting holes but we want to deliver a wide compatibility range to the endusers. You generally don't need to fix anything, because there are no side effects of that. The contact to the GPU is firmly and the thermal grease is doing the usual job. Basically there is no negative aspect, because the contact stays the same, and your VGA card will not be influenced by the movement anyway.

We'll see if we can supply some rubber components for an update later on. This is our first VGA Cooler and we'll improve as much as we can there, yet technically it is already up to the task.

Thank you very much for your understanding and best regards from Scythe.

Cheers,
Andy

[nicsimi]

Hi Andy,
Thanks for your details, and sorry for my last post (I was a little bit angry).
But, I still consider the highlighted issue as a problem. I will explain why: the metal backplate with the rubber part is pressing (consistently) on the SMD components located behind the GPU (at least in my case - a non-reference Sapphire 4870).
IF the entire VGA cooler is moving like I was explained before, this movement is transferred also to the backplate and the rubber component is also moving while pressing the SMD components. This can destroy the SMD components (broken components or unsoldering the components from the pads).
That's why I strongly suggest to avoid this movement on this specific VGA design (with this backplate-rubber component solution).
Or, as an alternative, you can use a different backplate - like the one from HR-03gt for example, if you're shortening the screws.
Even if is not pushing right in the center of the GPU, I think this is not affecting considerably the cooling efficiency, but will avoid destroying an expensive VGA card.
All the best,
Nic

[Shinigami]

Hi nicsimi,

Thanks for your understanding, I appreciate that.

I know where you are coming from here, the SMD issue is something I was thinking about before but then this is the only thing that could really happen. Unsoldering would not be possible because you need a high temperature which is not possible to appear while the GPU is really cooled, even it is not cooled you should not get such high temps to have the components unsoldered. But the breaking part is something could happen with a very very low percentage, yet I agree here with you. As said before, we are working around that and hopefully next revision/production can have some fixes. This is why I really lvoe this forum, some people really care and take their time to mention problems and really thing about those which helps us solve them as well =)

Cheers,
Andy

[nicsimi]

Finally, the outputs:

Conditions:
- I've used the cooper spacer
- I've used rubber grommets (no more axial movement)
- I've used the backplate from HR-03GT
- ~ 24 degrees ambient temp.
- non-reference 4870 design by Sapphire (analogue VRM)
- inside an HAF932 case
- musashi's fans ~ @ minimum (very silent -> cannot hear it over the undervolted Nexus 12cm case fans)

Results:
- idle: ~42 degrees
- load (~10min of ATI tool load): ~62 degrees

In my opinion it is a very good achievement taking in consideration this "hot" VGA card
For VRM I couldn't read the temp. in software (Analogue VRM), but I've measure it on the backside (~66 degrees - IR thermometer).
Musashi is a good design, but with small improvements it can become one of the best VGA cooler (aspect/functionality/efficiency/silence).
----
update: VRM temp is in idle (~66degrees), under load is ~ 95degrees (backside) and after ~10min of ATI Tool the card's protection is starting (black screen).
Pls. don't forget, this is in a very well ventilated case: HAF932.
So, the final conclusion from my side: even with improvements, this cooler is not able to handle a HD4870 (even with analogue VRM). Is able to cool the GPU very well but is failling to keep the card running because is not able to cool down the VRM @ minimum speed of the fans.

[javier]

Unfortunately I have the same problem with my card, and there is a negative side effect. Yesterday my card was with the Musashi installed was showing maximum load temps of 59*C. I removed the card to day to finalize the cabling in my case, and in the process of removing the card the cooler rotated and moved a relatively large amount. The result is that my temps after reinstallation are reaching a maximum of 75*C. I've actually improved airflow to the card by moving cables out of the way, so I'm confident that this movement is what caused the change in temperature.

Facing the possibility of having to reseat my cooler and having to handle my card like a piece of fine crystal when it's installed/uninstalled is frustrating to say the least...

You can tell me and everyone else whatever you want, but this IS a design flaw. Making the cooler universally applicable has cost reliability and repeatability on a very common card. I'm left with no choice but to fabricate some kind solution to this problem. I use scythe products in EVERY PC I build (fan, fan controller, cpu cooler) and this is the first time I've been disappointed. I've come to expect a level of fit and finish that just is not present in this product...

[Vanseb]

Hi,
Here my short experience with a Musashi on a 4870.

The cooler can rotate, of course, accentuated due to the fact I have a SFF case, so the card is "vertical". The cooler can rotate on the X axis, but I think that's not a big problem.
If I look at the past, the VF700 could rotate too (not as the Musashi, I admit), not with the screw and holes, but throught the fix coverslip. But I didn't saw any report about that.
Same problem with some waterblocs, like Alphacool models for CPU. It rotates a lot due to the one screw who's used to block it.

I don't have problem with temp, even if I rotate the Musashi and block it to have an other position. I think the problem of higher temps since the Musashi rotates is with thermal grease. Maybe there's to much, or to less, and when it rotates, the "configuration" changes.