Today, I almost...or, pay attention to your lapping process

[Railgun]

This seems like the best place to pop this.

My sig has my forthcoming rig that I'm working on. For no particular reason, I decided to lap this thing. The CPU, given what it is, took a little forethought and I thought I'd had it figured out.

I'd used the socket blank from another board to attach to the back side of the CPU, both for protection and a little extra grabbing surface. A little electrical tape to seal them together (which worked quite well) and away we went.

After a few hours on that as well as the heat sink, I decided that this is a bit of a PITA and found that there's a local lapping supplier and service company practically just up the road. For a relatively low price, I decided to see whether they would do it. Fortunately they will and I'm dropping it off tomorrow.

However, in cleaning up the CPU from the initial work I did, I quickly discovered the error of my ways.



But wait...



Ehhh...



Yeah. That had me britting shicks. I learned that, while the CPU blank will cover things nicely, it also partly sits directly on the pads. D'OH!

The motion and the slurry I'd created with some of the finer grit paper got where I'd not intended it to go and...there ya go.

Fortunately, it seems it's superficial as there doesn't seem to be an issue with the CPU based on some basic initial tests. This could have been a stupidly expensive lesson learned. Fortunately, it has become an inexpensive, but very valuable lesson learned.

And for the record, I'd seen info in the past that some previous gen Threadripper IHSs were not copper. This most definitely is.

 

[Flamethrower1972]

This seems like the best place to pop this.

My sig has my forthcoming rig that I'm working on. For no particular reason, I decided to lap this thing. The CPU, given what it is, took a little forethought and I thought I'd had it figured out.

I'd used the socket blank from another board to attach to the back side of the CPU, both for protection and a little extra grabbing surface. A little electrical tape to seal them together (which worked quite well) and away we went.

After a few hours on that as well as the heat sink, I decided that this is a bit of a PITA and found that there's a local lapping supplier and service company practically just up the road. For a relatively low price, I decided to see whether they would do it. Fortunately they will and I'm dropping it off tomorrow.

However, in cleaning up the CPU from the initial work I did, I quickly discovered the error of my ways.

View attachment 366688

But wait...

View attachment 366687

Ehhh...

View attachment 366686

Yeah. That had me britting shicks. I learned that, while the CPU blank will cover things nicely, it also partly sits directly on the pads. D'OH!

The motion and the slurry I'd created with some of the finer grit paper got where I'd not intended it to go and...there ya go.

Fortunately, it seems it's superficial as there doesn't seem to be an issue with the CPU based on some basic initial tests. This could have been a stupidly expensive lesson learned. Fortunately, it has become an inexpensive, but very valuable lesson learned.

Thats not good. Can you fix it?

 

[EarthDog]

Holy....you dodged a bullet there so far... wow!

 

[Railgun]

Thats not good. Can you fix it?

So far, nothing needed to fix it.

Holy....you dodged a bullet there so far... wow!

No doubt. I need to find a pinout for the chip to see exactly what these are. I think I'd found one for the Milan EPYC once, but nothing so far on these. My Google-fu is lacking lately.

 

[Railgun]

So, luckily for me, I've found a local company that specializes in lapping and will actually do CPUs and blocks to spec. I've only done a pretty basic lap (relatively speaking) to a matte finish, but it's considerably more flat than it was before. I'll post images shortly.

There's some light wringing between the parts, to which I will call good enough.

Now on to other stupidity.