|
|
Post by ShrimpBrime on Mar 2, 2019 23:09:52 GMT -5
Well, it was a pretty smooth de-lid. The glue is cheaper and easier to cut. I noticed the adhesive on the cpu covering the PCB is much cheaper also. You can pick and peel it off with your finger nails. Other than that, First attempt successful the processor came right apart. The only issue is I only have cheapy white thermal compound and first post gave displeasing temperatures. I'll run to the store tomorrow and pick up a tube as I miss placed the one I had.... here somewhere.... in one of the boxes. So lets just say at 4Ghz with cheap paste cpu was at 71.5c by the end of Wprime 1024M and I'd hate to test with anything heavier. Then twisted my evening pretty good by slapping a plate with a TEC just to see temps. Well, This cpu at stock easily soaks the TEC even at 3.7ghz no boost at 1.199v 16T. Eeek. Spikes hard. Long story short, Down to 4 cores no SMT 3.7Ghz and 1.199v and stays frozen Idle @ -20c. Couple of photos here...    |
|
|
|
Post by ShrimpBrime on Mar 2, 2019 23:18:54 GMT -5
This would be idle temp screen shot at -19c lid-less PGA.  |
|
|
|
Post by ShrimpBrime on Mar 2, 2019 23:35:30 GMT -5
Back the clocks a bit down further and drop the voltage, can stay in the negatives with an X4. Not bad I suppose, I wasn't sure what to expect from the way it acts on air cooling haha.  |
|
|
|
Post by ShrimpBrime on Mar 2, 2019 23:38:29 GMT -5
And gotta have the frozen pic for proofs I suppose.  |
|
|
|
Post by ShrimpBrime on Mar 3, 2019 2:18:20 GMT -5
OK got 3.2Ghz 8/16T under a volt. It's about the best this TEC can handle. Did some adjusting on the cooling seemed to help too.  |
|
|
|
Post by alpi on Mar 3, 2019 17:48:09 GMT -5
So, You have a Tec between core and the block. (like a band, core & the block  ) I'm curious how will it perform under lünger loads. How big the peltiers output ? |
|
|
|
Post by ShrimpBrime on Mar 3, 2019 19:01:08 GMT -5
TEC # 12715 Qmax of 136w - Temp difference (from cold side to hot) = Tmax 70c - 15v Max current. Max operating temp of 90c. www.thermonamic.com/TEC1-12715-English.PDFIt's not big enough to sustain a load for an extended period of time. The plate I'm using is slightly smaller in mass than the IHS plate, so it doesn't hold a temperature for very long. I didn't think a TEC would freeze 16 threads at all, So this is a surprise to me. |
|
|
|
Post by ShrimpBrime on Mar 3, 2019 19:46:22 GMT -5
OK got the TG-7 installed, removed the TEC and now on the Geothermal Loop. I have no screen shots at this time. Looking for beyond 4.5Ghz clock frequency @ 4 cores only. Working my way up, right now at 4450Mhz x4. Max temp in AquaMark was only 42c. reached a max of 46c with cinebench R15 x4.  |
|
|
|
Post by ShrimpBrime on Mar 5, 2019 22:30:10 GMT -5
Now installed a larger Cold plate. About double the mass and size of the IHS plate. Can retain a heat much longer. Results seem to vary depending on idle time to bring cold plate to temp or actively using. Almost always stays frozen unless loaded for long periods of time. 100 seconds of WPrime 1024M brought temps to a max or 0.9c at 3350Mhz .981v Lowest witnessed idle temp was -28c with this set up so far.  |
|
|
|
Post by ShrimpBrime on Mar 5, 2019 22:40:53 GMT -5
Here's what the plate looks like. you can see the size relationship to the 40mm TEC which is the same size as the HIS plate.  |
|
|
|
Post by ShrimpBrime on Mar 7, 2019 18:05:57 GMT -5
So with the above picture, started overclocking. with a slight under-volt I can get benching around 1.1990v which is actually a slight under-volt, and run the Cpu @ 4Ghz. WPrime - Stays frozen on the 32m run, Idle @ -25.5c. 1024M it hit just over 20c and was slowly climbing. I am not interested in heat soaking the TEC and cause issues between condensation and overheating simultaneously. Been there tried that and results are always horrible. I feel comfy with a TEC that can maintain under 30c and not a stitch more. That is the reason I always try to keep frozen. But since we are here, I'll go into ambient temps with the TEC for viewing purposes. Will try to get up a Cinebench score at this setting also.  |
|
|
|
Post by ShrimpBrime on Mar 7, 2019 18:18:46 GMT -5
OK Cinebench was also a successful run. Here is temps just before the run completed and the score just after I saved the screen shot. Basically in under 60 seconds, this TEC on the Geothermal water block can get temps well into freezing. The second picture is what ever time it took to save the first screen shot and save printscrn the score. However as fast as this TEC can freeze the chip the processor can heat up the TEC. I am being probably overly cautious, but for good reason. I didn't fry the 2700x during the de-lid, darn sure I'm not going to fry it with a Peltier.   |
|
)
