An HP runnng hot. Do I, Don't I?

[Loose rivets]

delve in and re compound the chips?


Tonight, I pulled out the Rivetess' HP, the one I found for her when her American one needed re-balling. The bits of that are still in Texas.

Anyway, this one is like a Rolls Royce. Special Edition, Altec Lancing sound. Great paint job, but just like the American one, it runs hot. Speccy shows 60+ on both the CPU and GPU.

Looking at tutorials of how to get to the heatsink brought back some fearful memories. I did in fact temporarily cure the balling problem, but it failed again during the detailed rebuild.

This one I got given when I rescued the pictures from it, and it runs okay for a modest chip. I managed to pull the Key with Belarc from a Vista machine and download Office 2007. Again, the key was required before download was allowed. A @$#%@#$5 Capatcha needed 4 goes.

It seems a shame not to used such a nice bit of kit, but the 26 screws just on the back cover is a bit daunting. Sooooo many connections before ripping the top trim and keyboard off that it becomes far from a fun project. But the local guru assures me that the heat compound will make a tremendous difference - just de-fluffing is not going to hack it.

What think you?

[G0ULI]

60-70 Degrees Centigrade is okay but that partly depends on the current air temperature. So the answer depends on how hot you expect it to get where the computer is being used. The temperature is never going to be a fixed value, it will be a differential between the processor heatsink temperature and the temperature of the incoming cooling air.

So long as the vents in the case are clear and the airflow paths as dust free as possible, you shouldn't have any problems. HP will have tested the machines in fairly extreme enviornments to make sure they keep working.

Lower temperatures tend to increase the reliability of components such as electrolytic capacitors which contain a liquid that can dry out if they are subjected to too much heat.

Unless you are experiencing problems that are clearly due to overheating such as the computer unexpectedly shutting down, it is better in most cases to leave well alone. If it ain't broke, don't fix it!

[Loose rivets]

I'd certainly rather not do this one as it is a long procedure. There were some symptoms before the Rivetess gave it back during her Christmas trip to the US. She said she'd downloaded fresh Nvidia drivers for some reason or another. It wasn't as bad as the American one which had an intermittent blank screen.

On two Youtube things I've watched, neither of the guys used paste on the video chip! If they can strip this one, they can do anything, but do they know, or is there something I don't know? Very strange as historically, the video one was the one the caused the end of a lot of high end HPs.

This is the Texas one, and they used a fat pad of grey stuff, which is counter to the tutorial for the gold goo. (They had it going on very thinly.) Maybe that's why I didn't see them put stuff on, because the used the old pad.

http://img.photobucket.com/albums/v703/walnaze/PpruNe/DSCF1503.jpg (http://smg.photobucket.com/user/walnaze/media/PpruNe/DSCF1503.jpg.html)

[yellowtriumph]

I'm not sure I'm actually following what you're trying to do?

Are you are asking 'I wonder if it would be a good idea to re-apply the heat compound to some chips in a computer that I am re-furbishing?'

if that's the question you are asking I would advise that in my 30+years of electronic maintenance on some very hi-end electronics we would routinely replace the heat compound on chips and transistors as and when we came to re-furbish electronic circuit boards. We regarded this as good practice.

You can either clean off the existing heat sink compound with a cotton bud and some alcohol etc and re-apply some new compound, or, latterly heat sink compound per se has gone bit out of fashion and it was quite usual to see newer equipment fitted with heat sink pads. These pads were usually cut from a grey blank sheet about 6 inches square so you could cut it to any size or shape. It's actually a very effective replacement, but, you should consider the thickness of the material to make sure it would not cause any mechanical problems in close fit mechanical finishes.

Available from RS components etc in the UK, not sure where you are from your signature. I hope this may help you in trying to achieve whatever it is you're trying to achieve. Whilst there is great deal of common sense in the if it ain't broke don't try to fix it scenario, I would, but then again I am very confident working with electronic components.

Edit. And I hope you're taking the usual anti-static precautions when dealing with circuit boards and components!

[G0ULI]

New Nvidea drivers caused problems with my HP laptop. Downloaded the recommended drivers from the Nvidea site as prompted by Windows 10 and the system became very unstable with blank screen and the whole system locked up.

Reverted to basic on board graphics and removed the Nvidea drivers and went to HP's website where I downloaded their set of driver files. Reinstalled the HP version and everything was fine, except for the message from Nvidea that popped up suggesting that a newer, better set were available from their site.

DON'T DO IT!

The drivers supplied from Nvidea and with Windows are generic versions that kind of work with many machines, but the computer manufacturers are responsible for fine tuning and making sure the graphics drivers actually fully work with their products. Stick with the HP drivers and things work. Upgrade(?) to the latest and greatest versions on the Nvidea site and problems arise.

[Loose rivets]

Gouli - That's interesting and could explain a lot. One will look for some drivers. It may all become academic since it's still 'updating'.


yellowtriumph That's also interesting. I'd wondered about that pad's efficiency. As mentioned, a forum member, talking about cleaning and reapplying the gold compound, said if the metal was perfectly smooth there wouldn't be any need to apply it, so all we're trying to do is compensate for the unevenness of the surfaces. He put a 3mm blob in the middle of a round heat-fan and said it would spread out. (Intel i5) I used the spatula spreading method, but did it thinner than usual.

Back to the grey stuff. Back then, 3 years or so, folk were putting copper coins on the video chip. Real old copper ones, I hope. :8

The chip as you may know is the one at 45 degrees and the daughter board is stuck with a red goo as a fix for their balling failures. That stuff was so hard I couldn't cut it, or melt it with an iron. But I think they used a bit of imagination if they thought that would compensate for balls that hadn't flowed. They would be - and were on most failures - just like any other bad joint after a year or so.

When I read the BIOS going by I assumed I'd had some success and took great care with the proper rebuild. It was then that I found I was out of luck.

A local chap here - I'm in Essex - claims some success with Coke tin shields and a heat gun. A plasterer by day, he seems to have a good reputation.

I'm totally at a loss as to how the re-balling works. They seemed to just clean up and sprinkle .6mm balls into the space and then shove them about with a brush. I can not imagine any other way than to stick them with flux and register every ball with NOT ONE TOO MANY! The heat gun was obviously very expensive kit but the timing was just waiting for the chip to 'drop'. Very odd way to save money in manufacture.

[yellowtriumph]

I see. The re-balling is quite straight forward but it does require some confidence and experience. Heat guns are not that expensive in the overall scheme of things but probably not something you buy for a single use only.

Solders a remarkable product, hard one minute - soft the next. It's getting a uniform heat across the chip thats the tricky bit that requires experience. Solder will attract solder and that's what they are relying on for re-balling. If the 'ball' is touching the original solder pad beneath it it will flow onto the pad when it melts - self centring when it does so. Don't go there if you are in any doubt - you will more than likely destroy the board.

I see the 45degree chip you are referring to and I sort off assume you are thinking of attaching a heat sink to the chip - fair enough. You don't have to glue on copper coins or coke tins as you can buy heat sinks from RS etc or Maplin etc. Maybe something like this?

ICK BGA 14x14x10 | Heatsink, BGA, 27.4K/W, 14 x 14 x 10mm, Adhesive Foil, Conductive Foil Mount | Fischer Elektronik (http://uk.rs-online.com/web/p/heatsinks/6744756/)

Am I barking up the right tree?

[Loose rivets]

Yes, but not about the duality of the heatsink. The graphics sink is part of the conductive bar and in the photo has the grey pad left stuck to it. I imagine the folk on that forum were talking about jamming the coin either in, or in stead of the grey pad.

Thanks for the briefing on the re-balling. I'd love to see it done in real life. Not that one can see the actual flowing.

Really, I think this machine will have to live or die. I don't think I'll launch into that complexity now I don't have a proper hobbyshop.

The update - if that's what they could call it - has finished. I'll pop over to that thread with some 'interesting' comments.

[yellowtriumph]

Ah I see now, I was thinking the parts to the Rhs 'flipped' vertically not horizontally. Ok two things. Do not mix heat sink compound and flexible heat sink interfaces (the grey pad). Use one or the other not both.

As you know you need an absolute flat face to face 'surface to heatsink' interface. I can see the idea of putting a coin between the top of the chip and the lower surface of the grey pad, but, from what you have said I think a piece of flat metal would be better so the suggestion of a flat piece of metal foil would seem a good one. Good luck.

Ps. I can't see what the graphics chip is but it looks as though it has soldered edge contacts around all four sides. If you can get a powerful magnifying glass carefully look at the soldered edge connections. In the past I have found chips that had been unsoldered or 'dry jointed' connections and if you see one, and have a suitable soldering iron, you can usually touch them up even though they are very small.

[Loose rivets]

In this case one had to remove the daughter board and chip as one item. The problem there is there were numerous tiny R's and C's on that board and to us olduns, heating stuff like that is beyond wicked. ;-)

Given I once held one of the very first Darlington Pair transistors which had to be soldered soooo carefully, I of course never believed there would be devices measured in billions that could be raised to those temperatures. In fact, I'm not sure I believe it now.

[Booglebox]

We had a bunch of these HPs at an old job and they ran pretty hot. Things can be improved by taking it apart and removing all dust.
Replacing thermal pads with thermal paste is a tricky proposition though. You really should use the same HP thermal pad as a replacement.

[obgraham]

Your skill levels are far beyond mine at this sort of thing, Rivets, but all I'll say is every HP lappy we've had in the past few years runs so hot I need something on my lap to avoid 2nd degree burns.

[Loose rivets]

That's alarming. Just imagine what the battery has to pump out when nowhere to plug into mains.


ob, skill levels? Nothing has been as challenging as the Seiko 7A series. I had one since 85 and set about refurbishment. Now I've got 20. But, some mystery neuropathy is becoming tedious and some days drop things. Best I only play with bricklaying on a bad day.

Utterly absorbing. I get lost in a different world while footling with these.

http://www.seiko7a38.com/apps/forums/topics/show/7517616-inside-the-7a38

[boguing]

I mentioned my HP experience in a previous thread of yours concerning bios upgrades. It was running so hot that it would shut down (ungraciously) doing relatively menial tasks. I'd cleaned the air path and even cut the plastic grilles out to improve things. Bios upgrade cured it completely.

I don't think that there's much point in re-doing the paste. For one, I think that the claims to huge differences is akin to snake oil. Why would a big name brand skimp on fractions of a cent on something important? Secondly - where would the existing paste have gone to? It's still in there. I did a test last summer on an old laptop that I didn't mind frying. Put it together with no paste and compared the indicated temperature. The difference was about 5 degrees and it didn't fry!

One thing that you could do if the bios thing doesn't work is to put an ssd in it - less heat being generated.

[yellowtriumph]

I mentioned my HP experience in a previous thread of yours concerning bios upgrades. It was running so hot that it would shut down (ungraciously) doing relatively menial tasks. I'd cleaned the air path and even cut the plastic grilles out to improve things. Bios upgrade cured it completely.

I don't think that there's much point in re-doing the paste. For one, I think that the claims to huge differences is akin to snake oil. Why would a big name brand skimp on fractions of a cent on something important? Secondly - where would the existing paste have gone to? It's still in there. I did a test last summer on an old laptop that I didn't mind frying. Put it together with no paste and compared the indicated temperature. The difference was about 5 degrees and it didn't fry!

One thing that you could do if the bios thing doesn't work is to put an ssd in it - less heat being generated.
The heat sink compound doesn't disappear, after prolonged periods at high temperature it will bake/harden and could therefore degrade it's heat transference abilities. It's there for a good reason, omitting it is asking for trouble in the long term, and as you say the cost is trivial really.

[Loose rivets]

Yes, SSD fitted.:ok:

I agree about the compound drying. It often looks in a poor state after parting the surfaces, though at this time I'm not sure about the temperature differences after re-application in a laptop since I've never done a really critical trial. As mentioned, the American that sold me the swish i5 Asus said he'd put a very expensive product on. In that slim shard of executive expensiveness, that was quite a job since even the memory and battery are soldered in.

I am however concerned about how much to put on having seen a site that fairly shouts 'don't put more than a droplet in the middle of a round (Intel type with fan) heat-sink'.

My surfaces were about 25mm diameter and it was suggested even a small pea sized blob would just flow out over the veins. However, this was for my i5 PC and not that critical.

[boguing]

I don't think that it does dry out. New cpus come with a 'sticker' of conductive goop - a very thin film. When we take the cpu off the heat sink we muck it up and it does look very dry, but if you push it around with a warm finger it becomes much softer, and I think that's how it works - heat softens it.

Anyway, if you've wiped it off and have to put new on, the pea is the size of a petits pois, not a garden one.

[G0ULI]

The principle is quite simple. Air is an insulator and pretty good at resisting heat transfer. The surface of the chip and the heatsink are not perfectly flat and may be slightly misaligned leading to small pockets of air between the chip and the heatsink. By appying a very small amount of heatsink compound the air gaps are eliminated and the efficiency of heat transfer is improved. Too litle compound and the air gaps remain unfilled. Too much compound and an additional layer is formed between the chip and the heatsink. Better heat conductivity than air, but not as efficient as it could be. The right amount is probably half of whatever you regard as the minimum sized blob. Something the size of a 16 point O will be more than enough for most purposes.

[Loose rivets]

The bloke on that 'tutorial' even said if the X method is used, (put an X of the stuff on and then spread it with a spatula) when the surfaces are put together, start at one side and roll it on so as to push out the air.

Well good luck with that one. With all the veins and gubbins, it's nearly impossible to have one side lower than t'other for a smooth rolling action.

much as we were saying:


The thermal paste itself is not as thermally conductive as metal, however, so it is important that not too much is used. If that happens, the thermal paste will behave more like an insulator than a means of improving thermal performance.

Gone to the trouble of making a Plexiglass surface to look through.


https://www.pugetsystems.com/labs/articles/Thermal-Paste-Application-Techniques-170/

[G0ULI]

Complete and utter overkill in terms of researching the problem. The rice size dot is clearly good enough for anything but the most extreme build.

The reason why a blob in the centre of the chip works best is because that is where the majority of the heat is being generated and hardest to dissipate. Around the edges of the chip the heat is all coming one way, out from the centre. The edge is next to cooling air in any case.

The single central blob method minimises the amount of product that needs to be applied, saving costs.

The single blob can be applied quickly and provided the heatsink is applied carefully and with even pressure across the chip, the heatsink compound will spread evenly in a circle, covering the areas where most heat will be generated.

Looking at the tables, there was only a couple of degrees difference between all the methods. Not worthwhile worrying about unless you need the computer to run at midday in the middle of Death Valley.

[yellowtriumph]

Complete and utter overkill in terms of researching the problem. The rice size dot is clearly good enough for anything but the most extreme build.

The reason why a blob in the centre of the chip works best is because that is where the majority of the heat is being generated and hardest to dissipate. Around the edges of the chip the heat is all coming one way, out from the centre. The edge is next to cooling air in any case.

The single central blob method minimises the amount of product that needs to be applied, saving costs.

The single blob can be applied quickly and provided the heatsink is applied carefully and with even pressure across the chip, the heatsink compound will spread evenly in a circle, covering the areas where most heat will be generated.

Looking at the tables, there was only a couple of degrees difference between all the methods. Not worthwhile worrying about unless you need the computer to run at midday in the middle of Death Valley.
I agree, don't overthink it.