Salta al contenuto principale

A1418 / EMC 3069 / 2017 / 3.0 GHz quad-core i5, 3.4 GHz quad-core i5 or 3.6 GHz quad-core i7 Processor. Released June 8, 2017.

138 Domande Visualizza tutte

Why does my iMac only have one diagnostic LED after upgrades?

I did a few upgrades to a 2017 21.5” Retina iMac, and unfortunately, upon reassembly, it isn't powering on. Here's a photo of it (https://imgur.com/a/BOzDy4Q), display disconnected. In the second photo (https://imgur.com/1EtPsCV). I've labelled things for reference.

Block Image

Status:

  • In the photo, AC is plugged in
  • Of the 5 diagnostic lights, the first one (bottom), is lit up. This indicates that the machine is receiving power
  • The others don't LED's up at all (light 2 would indicate the machine is powering on)
  • When I push the power button, nothing happens at all. The fan doesn't move, no sign of activity, like a brick.

What I repaired:

  1. Switched out the disk with a new SSD (currently blank, was going to install MacOS from USB)
  2. Switched out stock 8GB RAM with 32GB from OWC
  3. Applied a [graphite cooling pad](https://www.enostech.com/innovation-cool...) I had instead of the stock thermal paste on the CPU (I regret doing this because it was really difficult to align it on the non-level iMac surface)

My experience:

  • I'm not necessarily an advanced hardware person, but my background is in computer science
  • I'm a software developer
  • I've worked with Apple products for years
  • I've repaired a about 30 different Apple products (various iPhones, iPads, Macbooks, iMacs and iPods, although not too many in recent years)
  • It's been a few years, but I don't think I've had a single problem in an Apple repair or upgrade before. I'm careful and I'm familiar with best practices.

Things that could have gone wrong:

  1. I'm really nervous about the thermal paste replacement, mostly because I've never used one of those graphite pads before. They seem to work fine for everyone though, so I tried it. I cleaned the copper heat sink and top of the CPU chip of all paste with 95% isopropanol. It took me _many times_ to align the stupid graphite pad well, and I can't say for sure that it stayed perfectly aligned when I attached the heat sink, because you're supposed to attach it, then flip it over to screw it tight (I did follow the cross pattern for gradually tightening the corners of heat sink pressure)
  2. Although I looked at the iFixit tutorial beforehand, one of the things I didn't realize was that removing the heat sink from the CPU would also remove it from the GPU and its 4 VRAM chips. These 5 chips had a thick thermal paste (like K5-PRO) on them. I didn't reapply any additional paste when I closed them back up. I think this is normally okay, but now I'm not so confident.
  3. For a brief moment the CPU fell (was stuck to the heat sink), and landed on the aluminum chassis of the iMac from a height of maybe 15cm (6"). I cleaned both sides with alcohol, inspected it. Seemed fine. Aligned it and put it in the CPU socket and continued.
  4. I was really careful with the power supply and made sure that all connectors were fully in place. Actually I was very careful with any junctions or connectors, not touching exposed parts with my hands, making sure they were fully inset to click, etc.. But since the device isn't powering on, I'm a little nervous about the power supply.
  5. There was one weird moment with the PSU. I cleaned everything with compressed air and, for a second, I thought I heard the PSU let out a shock sound when the air "froze" the surface of the PSU board. I didn't think much of it, but now I'm wondering if that is the kind of thing that can bridge a connection on a board and short something? Note that the machine hadn't been connected to power for several hours at that point, and I did try to hold the power button at the start of the repair to fully discharge capacitors.
  6. The RAM was the right type, pins aligned and all. But the second chip didn't quite fit in snuggly. I really felt like I had to jam it, and still, the bevel in the corner of the chip was not perfectly flush with the groove on the outline of the RAM socket. It was really close though.

***

Normally these sorts of small potential mishaps don't matter to much in my experience. I accidentally touch an exposed pin? No problem, clean it with isopropyl and carry on. But here something clearly went wrong (a first for me), so I'm really second guessing everything.

I don't have any thermal paste, but I ordered some quality stuff (two different kinds, a thick one and a CPU paste) just in case I now need it.

These new machines aren't easy to work with, everything has to be taken out just to get to the RAM and CPU, so I'm trying to diagnose as much as I can before going back in.

Based on these 6 potential issues, and the diagnostic lights, which seem the most likely? What should I prioritize exploring in the interest of time?

Block Image

Risposto! Visualizza la risposta Anch'io ho questo problema

Questa è una buona domanda?

Punteggio 1
Aggiungi un commento

2 Risposte

Soluzione Prescelta

OK so lets move on…

Your basic problem here is the chip carrier as you likely didn’t screw in the plate evenly or the chip slipped its position due to the graphite pad.

Go back in properly clean the surfaces and do a very detailed inspection of the pins and the chip for damage. Apply just enough thermal paste (good quality) to cover the chip die cover don’t over do it and don’t create air gaps!

Questa risposta è stata utile?

Punteggio 3

Commenti:

So the (i5) chip in question is the Kaby Lake generation of Intels. They don't use traditional pins and instead have a grid of contact areas on the underside of the processor (land grid array). So the socket itself doesn't have a bunch of holes like the old surface mounts do.

Actually as soon as the chip fell I rushed to look for bent pins and was happy to see that there were no pins to be bent.

I'm worried about alignment too. I'll get the paste in later today (assuming no Amazon COVID delays lol) and then try it out.

I have 2 follow up questions:

1. Would 1 LED light be showing up if the processor was not mounted correctly?

2. In this iMac, the GPU, surrounding 4 VRAM chips and CPU are all held together with the same heat sink assembly. I'll of course use new paste on the CPU and apply it carefully in accordance with the manufacturer recommendation (either a point or a vertical line usually). But what about the VRAMs and GPU? Should I clean those and use a thick paste like K5-PRO on all of them?

re: GPU and VRAMs, there's some dried thermal paste on there that's just super thick. Except now it has some cracks from removing the heat sink)

da

You are not looking at the socket! Thats where the pins are not on the Intel chips. Inspect the socket for bent pins.

As far as your # points:

1 - The LED #1 is just telling us its detecting AC power its not telling us if the power supply is good or anything on the logic board is working correctly. LED #2 as well as the other LED's

2 - The VRAM chips use a special gapping paste. Frankly, I prefer thermal pads over paste Thermal Pad Vs K5 PRO Viscous Thermal Paste

Here's another case thats similar to yours Dual drive setup- PCIe SSD with SATA 3 SSD. He too had issues with the CPU not sitting properly.

da

@danj Amazing. Thanks so much, this is really helpful.

da

@orun bhuiyan, had this problem been solved? Did one of the pins shift?

da

Saw you posted on stack…ever resolve this? Thanks!

da

Mostra 1 altro commento

Aggiungi un commento

Lets first talk about thermal conductance:

Imagine having two sheets of steel and you place a long piece of chain between them in a serpentine pattern covering the surface of the lower plate and then placed the top plate on top. for the sake of argument the plate and the chain are all the same material. So will the chain conduct the heat from the lower plate effectively to the upper plate? Sadly no! The chain links have air gaps and the amount of surface touch on both the top and bottom plate is very limited.

Clearly this is not effective!

Now let’s look at using ground-up metal (metal dust) the fineness of the dust and the thickness of the dust coating then defines how effective the thermal conductance of the dust (a thin coating is all we need). Unlike the chain links the dust has more effective surface contact so it offers better thermal transmission!

So dust is better! But wait!

Clearly, any thermally conductive dust would be very messy! If it was also electrically conductive too it would make a mess in computer! So we can’t use a dust straight up but using a binding agent like oil we can make a paste!

Now let’s look at what a sheet of graphite is using, a polymer instead of an oil. Does that work? It’s better than the chain but not as good as the dust held in a paste!

This is where things get into the mating surfaces smoothness. If I have a mirror finish on both then the space gap is the square of the area of the chips surface. Now lets say the surfaces are ground to 45 angles the surface area would be more than the flat, almost twice the surface area! So now we have the two extremes.

The surfaces are not a mirror finish and ideally a matte finish on both so we need something that fills the gaps of the two matte surfaces. The graphite sheet is not able to fill the micro imperfections of the surfaces so its not as good as the paste which can. Then the thickness of the paste defines the amount of dust to carrier ratio the thicker the better if its an oil and the lighter the oil viscosity the better. And lastly the better the thermal conductance the oil has the better! And again the thinness of the fill between the two surfaces is also important!

So is the graphite pad better than a good paste? No!

Questa risposta è stata utile?

Punteggio 1

Commenti:

First of all that was a great explanation and I'm really appreciative that you took the time to write it.

Yeah, honestly, the graphite paper goes against my better judgement, because it just feels strange to "dry mount" a heat sink to a CPU no matter the case. I ordered some high quality paste that overclockers use (and operates really well at normal temp).

Yeah, nothing will get in there like a viscous liquid of course, especially when we're imagining grooves at a near-molecular/nanometer level. I just figured that the graphite must have heat dissipation that's so effective that it dissipates heat as well despite the lack of perfect surface contact.

I wish I never went down this rabbit hole now of course, lol.

(I'll reply to your other answer in a separate comment).

da

Aggiungi un commento

Aggiungi la tua risposta

orun bhuiyan sarà eternamente grato.
Visualizza Statistiche:

Ultime 24 Ore: 0

Ultimi 7 Giorni: 0

Ultimi 30 Giorni: 3

Tutti i Tempi: 169