Contribute
Register

NUC in a Mac Mini case

Status
Not open for further replies.
...

BTW, cable shortening of the Bitfenix unit was aborted yesterday when I got too hasty and ended up messing up one of the USB 3 connections…..I'll come back to that later!

And I was hoping that you would share a workable process for shortening the USB3 cables. I looked at different cable routing to avoid shortening and it is not good.

And again your work on the mini is very interesting and I enjoy the photos.

Good modding,
neil
 
Not that you would use it in this build, but have you considered using custom bent heat pipes in any of your builds? In a case like that, it would be tempting to try them.
 
Not that you would use it in this build, but have you considered using custom bent heat pipes in any of your builds? In a case like that, it would be tempting to try them.

I did think of that for another build, but in the end I decided not to - going in that case for a direct contact approach. Space was so tight for where I was thinking of using them that it was difficult to work out the clamping devices at each end and a good route for the pipes without compromising device positioning. I still have in my garage a 1 metre length of 10mm x 60mm copper bar that I was looking at getting machined down to make some blocks and clamps, also have a 50cm length of 1 inch by 1 inch copper bar for making 1inch cubes of material for making CPU contact blocks. Time and too high labour costs for getting things machined though put that into the status of "something to get around to when time permits".
 
And I was hoping that you would share a workable process for shortening the USB3 cables. I looked at different cable routing to avoid shortening and it is not good.

And again your work on the mini is very interesting and I enjoy the photos.

Good modding,
neil

Thanks Neil.

I will share method when I have a good one…

I should get time tonight or tomorrow for "take 2" on it. My mistake yesterday was not masking off contacts adjacent to where I was soldering, and not taking the time to clean the solder tip after each connection. As usual a case of less haste, more speed.
 
Wow, it was definitely a good move to change the cooling method for the mSATA card.

I assembled the now cured parts closed up the case and fired up the NUC. It took a long time to boot initially, so I shut down and checked the bios. For some reason the disk boot order for my Fusion drive had changed in bios to be using my slow laptop drive (mechanical laptop drive of 1TB). So I changed it back to the mSATA and was greeted with my usual fast boot. [As a side note interesting that when you have a Fusion drive you still need to physically set the boot to be from the SSD part to get max performance].

Anyway, once up and running I set to seeing how far I could elevate the mSATA temp by using geek bench and Black Magic Speed Test at the same time. After a while the mSATA temp rose to 38C and then stayed there!

I stopped the stress testing and let temp.s settle and the mSATA has since that point remained at 38C and I have not yet seen it climb beyond that - so if that situation remains then it appears I have shaved 17C from the operating temperatures with my new heat clamp.

Now, I'm leaving it running while I go off to have a beer and a pizza. :)
 
I stopped the stress testing and let temp.s settle and the mSATA has since that point remained at 38C and I have not yet seen it climb beyond that - so if that situation remains then it appears I have shaved 17C from the operating temperatures with my new heat clamp.

Wow.Glad to read that! A couple of questions:

I'm trying to read my Crucial's mSATA temps but HWmonitor doesn't report anything. Any thoughts ?
( NUC runs ML OSX)

You mentioned sleep issues due to the fusion drive. Did you solve them?

Cheers
F
 
Wow.Glad to read that! A couple of questions:

I'm trying to read my Crucial's mSATA temps but HWmonitor doesn't report anything. Any thoughts ?
( NUC runs ML OSX)

You mentioned sleep issues due to the fusion drive. Did you solve them?

Cheers
F

Hi,

I am on Mavericks (in fact reg'd as a developer for pre-release) and as up to date as possible, so I don't know about using ML.

I realise there may appear to be issues involved in doing the upgrade to Mavericks, but once you are there I think that you will probably have a better overall experience, and then if you upgrade to latest FakeSMC and HWMonitor you certainly should be able to see the mSATA temps. I'd advise doing a clone of your current working system to USB, make that bootable and then do your upgrade - that way you can always go back to your working system.

As for sleep with the fusion drive, I still need to do some troubleshooting as it will always wake from power button, but seems to have issue at the moment waking from USB. I have put that issue to one side though, as it is tolerable and I will come back to it once I have finished my hardware mods.

But yes, I am REALLY pleased with the temp.s now running purely passive. There are things you need to be very careful with in construction as taking some time over the thermal interfaces really pays dividends. I'll take pic.s of the final mSATA cooling arrangements a little later, to illustrate more clearly what I have done, but I suspect that if I had taken the time to really polish the CPU contact plates and prepare the junctions to a higher standard I would be running even cooler across all components.

Another thing I have noticed is that the unit as a whole tends to act as a very slow reacting heat store. So once you do a lot of hard stress testing the case will be at around 40C itself and then this means that it is slow to drop the temp.s of all internal components. In contrast if you only ever do tasks that are low stress (with some occasional peaks of stress) then the entire system stays very cool for extremely long periods. As an illustration, the ambient sensor now is saying 31C, the CPU is saying 31C and the mSATA is saying 31C - stress it and you will find all temp.s rise to around 40C. So it seems to me the Mac Mini case is, as a lump of aluminium, just about the right size to be able to cope passively at a good temp with these components.
 
Detail of the mSATA contact plate. The larger pad is from my 3mm copper stock, cut down to contact the two NAND Flash chips, the smaller pad is 2.6mm copper (I think) to contact with the Marvell controller chip which sits slightly higher than the flash chips.

IMG_1238.jpg

This is the final state of the heatsink. The thing on top is a piece of compressible packaging material that I have attached with double sided tape and this is thick enough so that when the plastic base is on it exerts a moderate force onto the copper plate to ensure a good contact is maintained between my plate and the mSATA card. Without it, I think that in time the plate might otherwise sag and efficiency could be lost.

IMG_1239.jpg
 
You mentioned the card has flash memory chips on both sides. Is there enough space between the card and the mobo to slide in another L bracket, attached to the same set of screws to "pinch" the card between the two?

Or, even more fancy, if it reached around from the center side of the mobo and hooked over the top plate as well to apply even pressure (with shims if needed). Would be a pain to get the combined assembly in though.
 
You mentioned the card has flash memory chips on both sides. Is there enough space between the card and the mobo to slide in another L bracket, attached to the same set of screws to "pinch" the card between the two?

Or, even more fancy, if it reached around from the center side of the mobo and hooked over the top plate as well to apply even pressure (with shims if needed). Would be a pain to get the combined assembly in though.

Maybe I was wrong about there being flash chips on both sides as the card is only 120GB perhaps the flash chips are all on the top surface. Certainly for the larger capacities I believe that both sides are used for flash.

In any case, I think I am going to leave it as it is for the moment! While it is not hard to remove the clamps, it does mean renewing the TIM paste each time, so I'll take a look when I have another good reason to take the card out. My feeling is though that if there are flash chips on the underside then it is unlikely there will be enough room for an extra contact plate to reach them as the mSATA card sits on top of the wifi/BT card so getting another piece of copper into the gap might cause me other problems!!

As it is, I am happy enough with the currently reported temperatures as well as the combined performance of my home brewed 1.12TB fusion drive - so I'll leave any further tinkering with the cooling for another time.

EDIT: by the way, I think if there was room for a second plate then the clamping arrangements you suggest would work really well and should not be hard to implement as the upper L bracket would allow adjustment of clamping force on the card prior to joining it to the lower L - and as the card is inserted into the Mini PCI-e socket it should easily pivot around until the holes of the upper L come into alignment with the slots of the lower L. So that would be a very elegant solution…..
 
Status
Not open for further replies.
Back
Top