News Noctua Paste Guard for Ryzen 7000 Prevents Thermal Paste Spillovers

Giroro

Splendid
So in order to keep a hot surface clean, this Igor guy used vinyl electrical tape? Has he like, never worked with electronics before, or is this just his first time using that tape?
Thermal paste is going to be way easier to clean off than the rubbery gummy mess that the tape will turn into when it rots, which will take about 5 minutes at CPU temperatures.
Not that I think it's even a real problem if a non-conductive thermal paste overflows a little bit.
 

Kamen Rider Blade

Distinguished
Dec 2, 2013
1,452
996
20,060
The AM5 IHS is such a stupid stupid design.

Prove me wrong.
I concur, I've hated the design since I first saw it, especially the 8-legged spider/octopus design with gaps on the side to the internal SMD capacitors/resistors.

But I understand that they had to do it for "Manufacturibility" reasons.

That being said, I hope Zen 5 has a new IHS design that isn't nearly as ugly and problematic.

I hope they fill out those cut-outs and offer a more standard design that won't let ThermalPaste spillover and potentially touch any of the SMD capacitors/resistors or contact pads.
 
Last edited:

abufrejoval

Reputable
Jun 19, 2020
601
435
5,260
So in order to keep a hot surface clean, this Igor guy used vinyl electrical tape? Has he like, never worked with electronics before, or is this just his first time using that tape?
Thermal paste is going to be way easier to clean off than the rubbery gummy mess that the tape will turn into when it rots, which will take about 5 minutes at CPU temperatures.
Not that I think it's even a real problem if a non-conductive thermal paste overflows a little bit.

Igor's been around the block for a couple of decades. And while he may not be downright lovable, he's ackknowledged for being generally extra thorough.

I don't know how Austrian Noctua and German Igor got put in the same boat, but when it comes to quality, it's more than a little likely they share a mindset, that might be explained by geographical proximity.

As far as the spider design is concerned, it most likely is anything but 'stupid'. Again, it's appearance may not be everyone's favorite, but 'de gustibus non est disputandum' was a bonmot already among the Romans for a reason.

It's most certainly the maximum number of contacts per substrate area, implying the maximum of I/O and power to fit into a surface constraint. Sacrificing esthetics for that in a context where even in these days of RGB exhibitionism nobody will ever really look after assembly, seems very smart and reasonable to me, especially if there are no real downsides with non-conductive pastes.

Perhaps someone will come up with ways to fill the space between the spider legs with something colorful or even blinking and all of a sudden this 'brainy' design will become iconic, but in the meantime I'll simply run with the basic assumption that AMD's leading engineers and CEO are not stupid.

Like most everyone else I don't think of myself as above average stupid, but I'll conceed that Lisa & friends may have a few more insights than I.
 

DonQuixoteIII

Commendable
Aug 16, 2021
33
18
1,535
Igor is my favourite successor to Dr. Pabst.... He is also even more OCD than I am. However, other than being somewhat unsightly, an excess of Noctua paste is not in any way a dangerous to your components as it is electrically neutral.

(For those that don't know, Dr. Thomas Pabst founded Tom's Hardware)
 
  • Like
Reactions: abufrejoval

InvalidError

Titan
Moderator
I concur, I've hated the design since I first saw it, especially the 8-legged spider/octopus design with gaps on the side to the internal SMD capacitors/resistors.

But I understand that they had to do it for "Manufacturibility" reasons.
There is no manufacturability reason for this: AMD could have gone with a 2mm wider+longer substrate and socket designs to fit everything under the IHS without cutouts, spared itself the extra IHS machining steps and still maintained backwards HSF compatibility with HSFs that use the stock AM4 backplate.
 
  • Like
Reactions: KyaraM

Kamen Rider Blade

Distinguished
Dec 2, 2013
1,452
996
20,060
There is no manufacturability reason for this: AMD could have gone with a 2mm wider+longer substrate and socket designs to fit everything under the IHS without cutouts, spared itself the extra IHS machining steps and still maintained backwards HSF compatibility with HSFs that use the stock AM4 backplate.
Then why didn't they go with that extra 2mm wider + longer substrate?

Would it cost them a new mold, is their extra costs to a 2mm wider + longer PCB substrate?

Is there still enough room within the Socket for that extra 2mm?
 

InvalidError

Titan
Moderator
Then why didn't they go with that extra 2mm wider + longer substrate?

Would it cost them a new mold, is their extra costs to a 2mm wider + longer PCB substrate?
Going from PGA to LGA was going to cost them a new socket design with all new molds and tooling no matter what the socket size was going to be, so the only real cost of making the socket ever so slightly larger just to accommodate an extra 1mm of mechanical space all around for the IHS to rest on with all SMDs moved underneath are the extra gram of plastic, slightly larger substrate area and a few extra grams of IHS copper. Machining slots in the IHS likely costs much more than the extra material would have.

AMD simply chose to prioritize maintaining the same substrate size for some arbitrary reason and had to use the "spider IHS" design to achieve that.
 

Kamen Rider Blade

Distinguished
Dec 2, 2013
1,452
996
20,060
Going from PGA to LGA was going to cost them a new socket design with all new molds and tooling no matter what the socket size was going to be, so the only real cost of making the socket ever so slightly larger just to accommodate an extra 1mm of mechanical space all around for the IHS to rest on with all SMDs moved underneath are the extra gram of plastic, slightly larger substrate area and a few extra grams of IHS copper. Machining slots in the IHS likely costs much more than the extra material would have.

AMD simply chose to prioritize maintaining the same substrate size for some arbitrary reason and had to use the "spider IHS" design to achieve that.
I wonder if AMD was able to reuse the original 40mm x 40 mm base PCB mold and didn't want to pay for a completely new PCB substrate size and just had to redo the traces and contacts only. That would explain why there is no dimensional difference between AM4 & AM5 on the PCB size.

Maybe there was some other automated machining/assembly benefits that we don't know about that was saved by not having to go to a larger PCB substrate size.
 

InvalidError

Titan
Moderator
I wonder if AMD was able to reuse the original 40mm x 40 mm base PCB mold and didn't want to pay for a completely new PCB substrate size and just had to redo the traces and contacts only.
PCBs aren't molded, they are cut out of large copper-clad dielectric sheets glued together two layers at a time with an additional dielectric sheet in-between, then cut out from that panel and the machines doing this can cut any arbitrary shape out of each panel. 40x40mm doesn't align particularly well with common PCB panel sizes, so AMD should have wiggle room in at least one dimension there without sacrificing a row's worth of yield per panel... unless AMD is using a custom panel size that doesn't have millimeters to spare.
 

Kamen Rider Blade

Distinguished
Dec 2, 2013
1,452
996
20,060
PCBs aren't molded, they are cut out of large copper-clad dielectric sheets glued together two layers at a time with an additional dielectric sheet in-between, then cut out from that panel and the machines doing this can cut any arbitrary shape out of each panel. 40x40mm doesn't align particularly well with common PCB panel sizes, so AMD should have wiggle room in at least one dimension there without sacrificing a row's worth of yield per panel... unless AMD is using a custom panel size that doesn't have millimeters to spare.
Then I have no clue. AMD chose 40mm x 40mm for some reason, and with Socket AM5 set in stone for it's generation, it has to deal with the consequences of choosing such a weird IHS because they choose to keep the PCB the same.

I wonder how many dependencies there are on the 40mm x 40mm PCB size for their automated manufacturing?
 

InvalidError

Titan
Moderator
I wonder how many dependencies there are on the 40mm x 40mm PCB size for their automated manufacturing?
PCBs get populated before the panel gets split in individual pieces so apart from panel size constraints, I don't think there is anything to it manufacturing-wise. Test jigs would need to be completely changed over between PGA and LGA, so nothing reusable there.
 

Kamen Rider Blade

Distinguished
Dec 2, 2013
1,452
996
20,060
PCBs get populated before the panel gets split in individual pieces so apart from panel size constraints, I don't think there is anything to it manufacturing-wise. Test jigs would need to be completely changed over between PGA and LGA, so nothing reusable there.
Then I don't understand why they didn't increase their PCB by 2mm on both LxW and just have a normal IHS with little wing tabs for the IHS Retention Frame to apply pressure to.