Scythe Ninja Rev.B Poor Performance

finaldeathh · Post by **finaldeathh** » Tue Feb 06, 2007 8:17 pm

I have read about the problems of the Ninja Rev B on this forum before my purchase. I went ahead and got the Ninja anyways.

System:
Gigabyte DS3 Rev 2.0
E6300
1Gb G.Skill RAM
2x 120MM fan mounted on case, 2x fans in PSU to draw out heat, 1x 120MM fan on the Ninja

I knew I had to file away some mm of metal on the LG775 bracket in order to make it fit. Fine, no problem, went out and bought a filing tool.

Installed it, had no problem. Used Artic Silver 5.

However, the performance is poor. I am idling at 37c @ stock speed and 60c when running orthos!! Overclocking to 2.8Ghz raises the temp to 65C when running orthos. Now, I have read about the performance of the Ninja and I know this is bad. I've seen people do 3.6Ghz and still have much better temperatures than me.

I checked my airflow and the air around the Ninja feels very good and cold. The Ninja blades feel cold/ambient/very very light warm when running orthos. The heatpipes feel just warm.

So I read about the problems Rev B has on this board and I was wondering what I should do? I read that a copper spacer might work but where can I get one for Core 2 Duo? I really don't want to spend $15 on a new retention kit. I could have gotten the Infinity!!

Post by **NeilBlanchard** » Tue Feb 06, 2007 8:27 pm

Welcome to SPCR!

Apparently, the issue may be less than optimal pressure against the CPU; or so I have read. I'm not sure if there is a solution, or not -- maybe use the "universal" mounting bracket?

finaldeathh · Post by **finaldeathh** » Tue Feb 06, 2007 8:30 pm

Yea, sucks that I might have to spend $15 more on this HSF and it still might not even work (read somewhere).

Should I add more AS5?

Post by **NeilBlanchard** » Tue Feb 06, 2007 8:35 pm

Hello,

No, more goop might only make things worse -- do you have just a very thin coat on there? Thin and as evenly as possible, with no foreign objects, either.

finaldeathh · Post by **finaldeathh** » Tue Feb 06, 2007 8:53 pm

I think I put quite a bit. Would that actually cause the temperature to go up? If the Ninja or the C2D is not perfectly fat, wouldn't more be able to connect them better? Others have reported the same horrible temps by the Ninja Rev B.

PopCorn · Post by **PopCorn** » Tue Feb 06, 2007 9:09 pm

go out to a hardware store or metal shop and ask for a small piece of copper sheet shine it up with [steel] wool and then put as5 on both sides... inexpensive and easy to see if it works

finaldeathh · Post by **finaldeathh** » Tue Feb 06, 2007 9:27 pm

There isn't a metal shop here and I have a feeling that they won't give me one for free even if I find one.

Any place to buy one?

ultrachrome · Post by **ultrachrome** » Tue Feb 06, 2007 9:51 pm

A hobby, crafts, or hardware store would be likely candidates.

finaldeathh · Post by **finaldeathh** » Tue Feb 06, 2007 10:07 pm

Hm...

Has anyone ever fixed the Rev b problem?

Chocolinx · Post by **Chocolinx** » Tue Feb 06, 2007 10:21 pm

If you go read the other threads I've posted on there are some suggestions. But to make your search shorter here is what I've posted in short:

1. I've tried several different methods of trying to lower my temps with the Ninja.

2. Purchased Unversal mounting Bracket. In short, WASTE MY MONEY! Nothing but the 478 bracket will fit, and it the 478 bracket that comes with the the Ninja doesn't fit on a P5B, touch a lot of resistors.

3. The Unversial Mount cam with a metal piece. Read somewhere that 75lb of pressure is needed to add optimal heat transfer. So I added the metal piece in between the heatsink and the CPU (paste on each side). Result: Lower temps by 5C and 10C on load. Personally though I don't suggest it, could potential warp Motherboard, but it seem to be running fine when I did it.

4. The last method I tried was against all of my heatsink laws. Instead of adding a goop in the middle and squeeze. I applied a thin layer across the cpu, AS5. Doing this for some reason gave me near the same results as the Metal Piece. Somewhere around 4C to 8C drops.

Conclusion: C2D's are most definitely concave to some degree. Adding pressure is what they need but the IHS was only designed to really take pressure from their Stock coolings! If you all look at a stock cooler you'll notice it's surface is a lot smaller than the ISH of the CPU.

1. Find a way to add more pressure safely, best retention are the ones that use screws and have a backplate.

2. Lap your CPU if you're willing to. But if you're not, best thing you can do is to add a slightly thicker layer of Paste to make sure there is contact. If the layer is too thin on a C2D it won't make contact with the Heatsink and you won't get any heat transfer at all making you lose out on some degrees of temperature.

finaldeathh · Post by **finaldeathh** » Tue Feb 06, 2007 10:35 pm

How thick did you make it? Do you have any pictures or any estimates?

Bluefront · Post by **Bluefront** » Wed Feb 07, 2007 3:24 am

IMHO.....forget thick layers of TIM. That is not the answer. TIM closes up air gaps, but cannot compensate for uneven pressure, warping, or insufficient pressure.

A hobby shop that sells model airplane stuff, usually sells small sheets of metal. A copper sheet 4"x10" costs <$5. An aluminum sheet even less.....Cut out a small square with tin snips. Flatten it between two perfectly flat objects and smooth it out with fine sand-paper...shims bend easily. Paste both sides....try it.

Mescalero · Post by **Mescalero** » Wed Feb 07, 2007 7:57 am

Chocolinx wrote: Conclusion: C2D's are most definitely concave to some degree.

Are really the C2Ds concave or are the Ninjas uneven? In the first case, this problem should also be present with other HS, I can't believe they all add a lot more pressure to the chip than the Ninja.

I was planning on buying a Ninja, but I dont want to have to aplly some DIY method to a newly bought piece of hardware...

BTW: Where do you get lapping-equipment from?

Post by **CA_Steve** » Wed Feb 07, 2007 9:42 am

Mescalero wrote:
BTW: Where do you get lapping-equipment from?

Put a sheet of very fine sandpaper on a glass tabletop - do a search for details.

Note that lapping your CPU will void the warranty.

finaldeathh · Post by **finaldeathh** » Wed Feb 07, 2007 10:25 am

Anyone here have an old Ninja LG775 mounting bracket or a copper spacer? I am willing to pay a little bit of money for it. I really want to try it but I don't want to pay $16 for the retention kit.

vincentfox · Post by **vincentfox** » Wed Feb 07, 2007 11:12 am

Hmmm, I am running my E4300 at 3.1 GHz currently. I started up Orthos and checked the temps after a while and Intel Thermal Analysis Tool says 78C which does seem kind high. Touching the heatsink it doesn't seem that hot so you must be right about the poor heat-transfer due to seating. Although, maybe it really isn't that hot. Most programs don't seem to read the E4300 temps correctly, and the one that I thought did, Intel TAT, now reading claims it doesn't give correct numbers either.

What is the danger range on a Core2? System is running just fine. I am using only a chassis fan to cool things by the way there is no fan on the Ninja.

The push-pin seating method did seem kinda weak but I didn't look at the problem hard enough I suppose. Ugh, have to remove the motherboard *again* to do something about this.

Chocolinx · Post by **Chocolinx** » Wed Feb 07, 2007 11:27 am

finaldeathh wrote:Anyone here have an old Ninja LG775 mounting bracket or a copper spacer? I am willing to pay a little bit of money for it. I really want to try it but I don't want to pay $16 for the retention kit.

YGPM!

[quote="vincentfox]Hmmm, I am running my E4300 at 3.1 GHz currently. I started up Orthos and checked the temps after a while and Intel Thermal Analysis Tool says 78C which does seem kind high. Touching the heatsink it doesn't seem that hot so you must be right about the poor heat-transfer due to seating. Although, maybe it really isn't that hot. Most programs don't seem to read the E4300 temps correctly, and the one that I thought did, Intel TAT, now reading claims it doesn't give correct numbers either.

What is the danger range on a Core2? System is running just fine. I am using only a chassis fan to cool things by the way there is no fan on the Ninja.

The push-pin seating method did seem kinda weak but I didn't look at the problem hard enough I suppose. Ugh, have to remove the motherboard *again* to do something about this.[/quote]

I'm guessing you're using a Ninja since you're posting in this thread. Tips on Core 2 Duo Ninja:

1. Make sure the bracket isn't touching any of those resistors thingies around the CPU Socket. If it is, saw or sand it down.

2. If pressure is an issue, try adding a really thin sheet of copper metal (you can purchase this at a hardware store like, Rona or Home Depot). Should be no thicker than 1mm, any larger could harm the motherboard.

3. Since the Core 2 Duo is concave, try and figure out how concave it is. Because if you're not willing to do any of the DIY tips then your best bet is to spread a thin layer across the IHS. But since it is CONCAVE, you only spread to the concave area. DO NOT sprea too far out otherwise the paste might bleed and fall on your motherboard. And that's a definite NO NO.

4. Last tip, Lap your C2D. Go to any hardware store, and you can buy different grades of sand paper from 200 grit to 2000 grit. Usually 2000 is the highest you'll need. Any higher and it's a waste. You can easily find some guides google, so use that.

mbetea · Post by **mbetea** » Wed Feb 07, 2007 1:47 pm

So the problem would seem that the push pins are too long? At least on the Ninja. Would there be any way to remove the pins from the stock hsf and try them on the Ninja, or at least compare them? I've seen some pics of Intel's mounting scheme, it looks like a more elaborate style of push pins that you see on northbridge sinks and some graphics cards. And I know I've had problems in the past using aftermarket coolers on a northbridge until I used the pins from the stock sink.

I don't know, just throwing out suggestions.

vincentfox · Post by **vincentfox** » Wed Feb 07, 2007 3:38 pm

Scythe uses their goofy push-pin system that release when you rotate. I don't see a way to do a simple substitution or fix for this. I emailed [email protected] to see what they have to say about it.

finaldeathh · Post by **finaldeathh** » Wed Feb 07, 2007 8:32 pm

I emailed Scythe about it too.

nick705 · Post by **nick705** » Thu Feb 08, 2007 3:10 am

Chocolinx wrote:Read somewhere that 75lb of pressure is needed to add optimal heat transfer.

I found the thermal design guide I think you're referring to - it deals with socket 478 P4s and mentions a 75lbf "nominal clip load," but I still can't find any socket 775-specific info. It seems an excessive amount especially given the ccruddy design of the stock Intel fitting - I can quite believe each of the pushpins takes at least 19 lb pressure to push through the mobo and latch into place, but that necessarily doesn't mean it's applying that much pressure to the CPU heatspreader once the pins have engaged.

/edit: found it - apparently socket 775 has a *maximum* recommended static compressive loading of 70lbf, and a minimum of 18lbf.

/start rant: Scythe seem to be "improving" all their LGA775 heatsinks by switching to the standard pushpin-style mounting, even for monsters like the Infinity and the Ninja, apparently for the supposed benefit of making them easy to mount with the mobo in place. Well, it might be easy if you have hands the size of a newborn baby combined with the finger strength of a blacksmith (and of course double-jointed throughout), and you also have an eye on an extendable stalk so you can zoom into the case and see what you're doing. Otherwise, with a huge heatsink you'll probably end up having to remove the mobo anyway to be sure of getting a secure mounting (at least, as secure as this type of system can ever be).

I can't believe most of the enthusiasts at whom these aftermarket heatsinks are aimed would prefer the allegedly "easy" mounting of the crappy pushpins over a solid spring-loaded mounting which screws into a rigid backplate, especially when the heatsink in question has the mass of a small asteroid. It's not as if heatsinks are swapped in and out on a daily basis.

Please, Scythe, have a rethink...

burebista · Post by **burebista** » Thu Feb 08, 2007 3:27 am

What is good for Infinity should be good for Ninja rev.B.

Mescalero · Post by **Mescalero** » Thu Feb 08, 2007 4:36 am

CA_Steve wrote: Put a sheet of very fine sandpaper on a glass tabletop

I understand, though it's not lapping in the strictest sense, lapping uses a paste in which the particles are suspended. Sorry for the nitpickeing, but I'm an engineering student and so I wondered whre to get lapping paste and if it's really a good idea to apply it to a CPU. I understand now, though.

This thread really worries me, might have to purchase another HS after all...

Post by **CA_Steve** » Thu Feb 08, 2007 5:27 am

Mescalero wrote:
CA_Steve wrote: Put a sheet of very fine sandpaper on a glass tabletop
I understand, though it's not lapping in the strictest sense, lapping uses a paste in which the particles are suspended. Sorry for the nitpickeing, but I'm an engineering student and so I wondered whre to get lapping paste and if it's really a good idea to apply it to a CPU. I understand now, though.

This thread really worries me, might have to purchase another HS after all...

As mentioned by others, do a google search for lapping cpu's. The above was the short version. None use paste, some use machine oil.

Prowler · Post by **Prowler** » Thu Feb 08, 2007 9:42 am

Rather than start yet another Ninja thread, I felt I might as well add on to this one. I have to say I'm not entirely impressed with the Ninja (subject to change if someone can convince me that my temps are normal).

System:
Asus P5B Deluxe
E6600
Ninja Rev B with stock fan
Antec Solo case with two Nexus 92mm front intake fans and a Noctua 120mm 800rpm exaust fan.

Temperatures:
ambient: ~22C
idle TAT: 34C-37C (one core always seems a bit higher when idle, and strangley enough the warmer core will fluctuate between 37C and 39C and never read 38C, makes me wonder about the accuracy)
100% load both on both cores with TAT after a few minutes I see 63-64C.

I do not have the problem with the mounting bracket resting on the capacitors, and I checked all four pins to make sure the Ninja is securely mounted. I use what I thought was a small amount of AS5 on the center of the CPU speader, and spread it out in a bit of a line (as per the instructions).

I guess these temperatures aren't exactly horrible, but I honestly expected to be nicely under 60C full load with the Ninja, so I am a bit disappointed. It seems like a pretty large delta between idle and full load for a HSF that is apparently very efficient, no?

I'm tempted to try a copper shim, or perhaps the stock HSF just to give me something to compare to, surely the Ninja must be better than the stock HSF.

Post by **NeilBlanchard** » Thu Feb 08, 2007 10:13 am

Hello,

burebista wrote:What is good for Infinity should be good for Ninja rev.B.

This seems to be the best solution, and maybe some of you missed it?

continuum · Post by **continuum** » Thu Feb 08, 2007 12:13 pm

Hmm. Maybe I'll sand my E6400...

I have the original Ninja, though. I just installed a rev B on a customer prototype and it had the push pins-- installing them was no problem at all, although I have questiosn about the security of the stock Intel retail mechanism with such a heavy heatsink being shipped.

With my C2D E6400 @ 3.2ghz I get 69-71C in Intel TAT with Orthos @ 22C ambient and a Nexus 120mm, which seems in line with xbitlabs' results.

Temps on the Rev B and a E6600 @ 3.3ghz were similar, 69-71C @ 22C with a Scythe SFF12 something (800rpm) fan in a P180.

http://www.xbitlabs.com/articles/cooler ... he_11.html

JaRoD · Post by **JaRoD** » Thu Feb 08, 2007 12:43 pm

Don't have a ninja but have heatproblems with my big typhoon or maybe it's a problem with the sensor... At 2.97Ghz on my e4300 my temps with orthos are around 60c. I've actually remounted the cooler 2 times and I have it screwed down so hard the motherboard bends a bit and as5 spread on the entire cpu. Any ideas? Or is it just supposed to run that hot? Seems like it could be bad mounting since temp difference between 5 and 10v is only 3-4c

But the before remounting my temps were atleast 5c higher.

nick705 · Post by **nick705** » Thu Feb 08, 2007 3:40 pm

NeilBlanchard wrote: This seems to be the best solution, and maybe some of you missed it?

Definitely much better, although it's a lot of extra fannying around sourcing the right knobs, screws, springs, washers etc from a hardware store.

You'll also still need to file down the retention bracket on the Ninja rev.B if you're one of the unlucky ones.

I know none of these problems are exactly insurmountable, it's just the thoughtless bad design that irks me, and having to go to extra trouble and possibly extra expense modifying a product to make it do what it's supposed to out of the box.

vincentfox · Post by **vincentfox** » Thu Feb 08, 2007 4:45 pm

I'll repeat what I said in several other threads:

I ran 3 programs and got 3 different results on my now-idle E4300 system:

Intel TAT v2.05.2006.0427: 44C
RightMark CPU Clock Utility v2.2: 18C
CoreTemp Beta 0.94: 27C

Crazy!

Rebooting and checking the BIOS PC Health monitor as fast as I could, the "System temp" was around 31C. So it looks like CoreTemp is closest to agreeing with BIOS. Intel TAT seems way too high. So we may be worrying over nothing. It would be nice to see an update to Intel TAT that would correct this, so we could all reference the same manufacturer-approved tool AND consider it trustworthy.