@@pete0274 Hey, thanks for your comment! What you can see on the pcb is not silicone grease but silicone oil. Also silicone grease is widely used as corrosion inhibitor because it is chemically inert and therefore does not react with anything. However it is also hygroscopic which means over time it attracts moisture. This moisture in combination with oxygen then could cause corrosion. But that is very very unlikely as oxygen does not simply pass through silicone grease as the moisture does. Now if that moisture were to be acidic we could observe the acid reaction to the underlying metal. But that also is very unlikely.

@@PhysicistAnna Thanks for the info. Yes, sorry, I meant oil not grease. One question though, why silicone oil is hygroscopic?

@@pete0274 I'm happy I could help. About why it is hygroscopic, that is a good question that unfortunately I can't answer as this is beyond my knowledge. This would be something for a chemist to answer. When you commented I knew silicone oil or grease is chemically inert but I double checked my knowledge either way. I also knew of oil dryers/ dehumidifyer specifically used for silicone oil, I just didn't knew that silicone oil or grease was hygroscopic. That I found out thanks to your question. I always thought it was contamination of silicone oil and not its chemical property to attract moisture.

​@@PhysicistAnna I think is safe to say any dust(with or without silicon grease) will attract moisture depending on the humidity of the room where you running the PC. Silicon oil/grease will keep the dust in one place and prolly will cumulate there bringing more moisture. BUT the card temps will dry out the moisture and than we are left with the fact that silicon grease/oil will mix with other compounds present on PCB, composition is changed in God knows what and that can favor the corrosion. I'm thinking that way. IMO your aim Anna, mine to, is to cool down properly the VRAM, GPU itself can throttle down, VRAM can fry, will show artifacts(but not necessarily!! don't want you to think you'll have that warning, in many cases chip died out without artifacting first) and one of mem chip can die bricking your card. Manufacturers does usually the minimum, is our job to cool our cards properly even if we don't like it, will take time and tests and will take some money to. For example your 380 and mine to, cause I have the same, will clearly need more heat transfer from PCB to back plate. You have but a single patch of thermal pad on power delivery, nothing on VRAM. In this case backplate will not be saturated with heat and again is acting more as heat trap. I will use thermal putty as I said first on my main setup 1080 Ti EVGA and than on my Nitro 380 secondary setup. Maybe you know but, I will mention it anyway, when GPU is let's say 65 C you VRAM will be at least 15 C higher(can be also 20 C higher depending on some factors) so VRAM will be 80 C or 85 C. I consider critical 100 C on VRAM I don't care what manufacturers say. Rear fan of your case is very important in cooling you GPU if that GPU got enough thermal pads in between PCB and backplate. Look for a tear up video of 1080 Ti EVGA ICX you can see there TP are there in some quantity for a reason. Backplate is to take a way the heat and disperse it. Only than case fans can take away the heat away from your GPU. On the other hand lack of textolite in the layers of a PCB will not stop the heat traversing PCB from one way to to the back.(research this material is very interesting and worth your time>>> Textolite) Textolite should be in any GPU PCB exceeding 150 Watt TDP IMO. On my 1080 ti with 290 W max TDP(380 Nitro is only on 190 Watt so less heat) my VRAM atm sits equal or under GPU temps which was my aim. Initially my 1080 Ti sitting in a XL case was reaching 75 C on GPU meaning that my VRAM was 90 C or over while playing RDR2 at 2k, high settings. So I proceed with my mods : 1. Deshrouding the card changing 3x 88 mm/ 11 mm thickness original fans with 3 X 92mm/25mm thickness(more airflow and less rpm /less noise) 2. Cutting the grill for my rear fan of my new expensive case cause was hindering the efficiency of that rear fan with with 15-20 %. Research yourself the tests made by Puget Systems in 2011 of various fan grills. 3.Adding of another 92 mm fan in PULL not PUSH on the back plate of my GPU in the exact spot you see in the video kzbin.info/www/bejne/g5fEh4l_o7-na6M at 2:44 you see test results (this mod will work poorly on the 380 by Nitro because lack of enough TP in between PCB and backplate, also the backplate is plasticated on the 380 and is not on 1080 TI. 1080 Ti has no plastic layer sticked on backplate. Note that the test in my video are 2x instances of Valley benchmark in same time and last only 10 minutes but in AAA games(1 hour) is showing even greater results. ATM the back plate fan was increased to from 1000 to 1200 RPM still inaudible - is A Noctua Redux 92 mm. At 55 C GPU in 2 K with VRAM temps ranging from 50- 57 C for example in Assasin Mirage. Also take in consideration I changed only thermal paste on the GPU, the pads are leaking and wasn't changed from factory defaults. My 1080 Ti has various thickness pads and many so is costly. I'm gonna use putty as is superior and cheaper. This year I will use the putty , test it and make a video. Why thermal putty? First, is cheaper secondly, we have and a very good tester, this guy . He is hashing his cards and OC is VRAM on cards consuming over 400 Watt. That's a lot of heat right there. kzbin.info/www/bejne/fYvKhHaNqZuosKc 42 :48 a comprehensive chart of different thermal putties. I will stay on UTP 8 and prolly CX 1300. Hope it helps

Welcome to my channel and thanks for the comment! Indeed at that time I did not realise that the frequency has such an impact on the measurement because I was used to circuits that have a capacitor in parallel to the load. In such a case you can accurately measure dc to dc converters that operate outside of your multimeters limitation. Another reason I think was that I knew my multimeter can read frequencies of up to 10 MHz and I didnt realise that on the ac measurement it is limited to only 400Hz for accurate measurements. Really good catch!

Thanks for watching! I'm glad the video helped you fix your GPU. In your case as another mosfet died I would say it is definately worth replacing all of them. It seems like they got hit hard when the first one went out.

Use Thermal Putty instead > UTP 8 by Upsiren way much better than any thermal pads, not to say is cheaper. Dries way slower and don't leaks silicon grease. Add more strips of putty including over VRAM, Sapphire did a lousy job here, backplate for 380 is more of a heat trap compared with their 390 Nitro. VRAM is the most heat sensitive and has to be cooled.

@@pete0274 Thank you Anna, I will try to use Thermal Putty the next time. Question: Do you recommend Thermal Putty on PS4/PS4 Pro RAMS instead thermal pads?

Don't, especially on plated probes (like the gold plated ones she is using here). You will sand off the plating and they will never be as good after that. You will have to keep sanding more often just to keep them clean and sensitive.

@@cowsongs I hear what you're saying and perhaps I should have been more specific: if you have Probe masters (pr something like that), sanding is a very bad idea, but .... if the leads are cheaply made and aren't sensitive/responsive, then I still think that sanding the metal parts 'lightly' with very fine sandpaper is the way to go.

If you want to listen to a technician you should go watch Northridge Fix. She said she wasn't good.... And it shows actually. Using a heat gun around capacitors (those cylindrical components) is asking for a DISASTER. Those components blow up when there's too much heat!!! They should be protected with some sort of shield so there isn't heat transfer... And has she said... She had to apply a lot of heat to the mosfet. Which is VERY close to the capacitors!!! It's LUCKY she didn't got injured! SAFETY people!!! And bro... She's limited on equipment... No voltage injection and heat camera?!? Yeah... She ain't going far with this