Instead of invading on Burnout's thread, I'll post my lapping results here! Both my CPU's IHS and my Noctua were concave... The IHS much more so than the Noctua, however the Noctua had machining marks on it which worsened its performance. Pre-lapping temps were 39*C idle, and 67*C full load @ 4.005GHz. Post-lapping temps are currently 39*C idle, and 66/67*C full load @ 4.005GHz. Ambient temps rose between the first test and the second test by what I can only guess to be between 2 and 3*C. I'm using AS5 thermal paste, which has up to a 200 hour break-in period... After tempering the thermal paste, temps can drop anywhere from 2*C to 5*C. Sandpaper grits used: 320, 800, 1500, 2000 No water or oil used, and no polishing after the 2000 grit paper. The pic's immediately before the mirror images are the final product, and are the same finish as the mirror images have. Now, on to the pics! The CPU's IHS: The Noctua:
Wow, I wonder why you only got a 1*C decrease? Did you use the ruler trick to see if your lapping was successful?
Well, the 1*C decrease was pretty much 20 minutes after applying the AS5... After two heat cycles, my current temps are 63*C, which is a 4*C drop overall. I expect the temps to drop at least a couple more degrees before everything's said and done. Guess I shouldn't be so quick to post temps, lol. As for your second question, no... I don't have a good straight edge (razor or otherwise) to test it that way. I did add a small dab of AS5 to the CPU, tighten down the Noctua, and remove again to test the spread pattern though. Spread was perfect, so the lapping was what I consider 'good enough'. I followed AS5's instructions on applying the paste to the CPU, however I've heard the 'gain of rice' method works better thermally. If the CPU and HS are both flat though, it should be able to squeeze out what it doesn't need pretty well.
Nice results already! the general guie lines for appling thermal paste is for stock components where there might be ai pockets need filling. Now that you have smooth flat surfaces, less is more! lol! By the way, i have no issues to anyone hijacking my thread on lapping, by all means post away with your results!
Thanks Burnout! I MAY take the heatsink off one more time today and reapply the thermal paste based on what the CPU looks like coverage wise. I'm always a bit paranoid of using the proper amount, so I'll probably end up doing it, lol.
all i did was prime the surfaces of the CPU and HS by rubbing a little into the surface so it misted up, then dabbed a little extra on in the middle which was less than a grain of rice. Also may i note that i am using cheap thermal compound, the white stuff like tooth paste found in most crap computer shops. At the time when i bought it i was in a rush to get some and had no choice, and ended up with a 100g tube of the stuff for £2.50. Its nearly all gone now after 4 years! yay! you name it and i've used it on it, have even used it as a polish at one point!
Well, I ended up taking it back off... Was a tad more than needed, but less than I was using before lapping. Just used a grain of rice this time around... We'll see if there's any difference. Off the thermal paste topic... When I took the heat sink off, I noticed a few spots on the CPU and the heat sink both that seemed scratched... Now this puzzles me, because both of the surfaces are perfectly smooth, and there's nothing in the area that would cause any scratches... They're out of the way of the cores, but it still puzzles me as to how they got there. If I don't get any better performance (temps raised back to what they were before), I may re-lap both the IHS and the Noctua... Time will tell. Edit: OOOKay... Temps are hitting 70 right off the bat, soooo, I dunno... Off with her head?
Arctic Silver 5... I've always used it. I just re-lapped both the IHS and the Noctua heat sink, filmed both of them with thermal paste, put a small dab in the middle of the processor, and put them back together. Initial temps dropped 2*C so far. Brand new thermal compound, so as before, I expect that to get better. Also, ambient temps today are another few degrees warmer than they were yesterday. I would like to see closer to 5*C drop at least, but if I only get 2*C out of it, I suppose it's still worth it for the work I put in. I really need to get some better airflow into my case. I may have to order another Noctua fan for the HS, along with a couple Scythe, YL, or Noctua fans for the rest of the case. Edit: Back to no difference 67 before, 67 now... WTF? As concave as BOTH my heatsink, and my IHS were, I would think I'd get AT LEAST 3~4*C. Is AS5 really THAT good at heat transfer, or do I just need to let it heat cycle itself for a while and check temps again? Edit 2: Temp on both cores hit 68... Hotter than it was before I lapped it. I'm at a loss as to what's going on...
I'm pretty sure that it's the room temp causing this. I too am using AS5 and it so far the best thermal compound I've use (and I've used over 10 different compounds).
Room temp shouldn't cause that much fluctuation in temps... I'm wondering how accurate the temps actually are... After a bit of searching, it seems a lot of people are having issues with finding the 'actual' temp of the E8400. Most of the other temps I've seen on the 8400 are about 20*C lower than the temps I'm getting, and with faster speeds... I don't know what to believe. The Noctua is every bit as good as the TRUE, and a lot of temps I'm seeing are with TRUEs.
TJMax is set at 100*C. I'm actually not even going to go by the temps anymore... Instead I'm going to go by the DTS temp, which is the CPU throttling point. The only problem, is that I don't have a screenshot of my original DTS before I lapped. Here's info posted by unclewebb @ xtremesystems... "What Intel says is that as long as the value of the DTS is a positive number then you are operating it safely and within their intended design specifications. When overclocking, for maximum stability, we all need to keep as far away from DTS=0 as possible. Everyone running around comparing absolute temperature values that are not even documented by Intel is pointless, especially when Intel fully guarantees their processors at TjMax. Functionality, electrical specifications and reliability commitments are guaranteed at maximum Tj as measured by the DTS." Edit (seem to have a lot of these...): I just pulled up a screenie of my system in my memory thread running @ 4.00GHz. The DTS temps were about 5*C LOWER at idle speed / temp than they are now (meaning the total distance til zero was lower, or the chip was running higher). Based on that, I'm going to say that I gained a total of 5*C so far in my DTS headroom (which, according to everything I've looked up on the 45nm chips, is the important thing, NOT the 'actual' core temp). Edit 2: I still don't know how accurate that ACTUALLY is... Idle doesn't tell me much, but I don't have a loaded screenie to compare it to.
one thing i have noticed is that alot of temp apps report different valves, so i tend to use realt temp, speed fan and smart G to draw a rough average temp. High temps after applying thermal paste, not surprised at all, suggest you let it burn in for a few days at a lower clock speed just to keep temps down for a little.
I have Everest Ultimate, as well as Real Temp, both reporting the exact same values. I'm still just going to stick with the DTS numbers.
Nice work JaredC01. One thing i've noticed, is that when a thicker layer of compound is needed to fill all the gaps, as5 works better than most. But with lapped smooth surfaces, a much thinner layer is neccessary to fill the gaps, so a relatively much thinner compound is needed. AS5 can't go as thin as many other thermal compounds, such as ceramique and many others. So allthough as5 is great, it's hard for it to compete when it's thicker. Using diamond powder based compound, when mixed well with the correct type of grease, is probably the best, even for lapped surfaces iirc, provided the powder is at least a certain micron wide... i forget, google knows
I've got some thermal paste that was included with the Noctua, but I think I'm just going to see how well the AS5 works out. VERY thin layer on it right now, so we'll see what she ends up doing. The old AS5 that was on it before I lapped the CPU was on there for a while as well. Either way, I'm happier with two smooth, flat surfaces, than two concave surfaces... Even if the temps aren't much different... Time will tell.
