Question Why don't some CPUs just drop flush into the socket?

[ptTimeBldr]

So, why don't some CPUs just drop flush into the socket? I discovered recently that I had to push down on the corner where the yellow triangle is to get the Ryzen 5600G to click or slot into place, so that no pins were showing and that it was completely flush with the socket. Just allowing it to drop in doesn't do this. One end of the CPU is slightly raised, so you have to push it down using medium force where the yellow triangle is.

 

[Ralston18]

If the pin/socket connections where loose enough to do that then the connectivity would not be tight enough to maintain steady/stable contact.

At the frequencies involved solid contact is needed. Any interruption on any active pin likely to crash everything.

Just my thoughts on the matter.

 

[drea.drechsler]

It's probably something peculiar to your specific CPU and socket combination. Something like a pin that's just slightly bent making a tight fit on a contact that's just slightly skewed.

It's a bit scary when they don't just drop into place, I know I'd be leery about pushing down a corner. I'd probably pick it back up first and then run a credit card between each row of contacts, in both directions, to make sure they're all perfectly aligned upright. It really should drop in place with no more than some slight jiggling side to side to help.

 

[ptTimeBldr]

If the pin/socket connections where loose enough to do that then the connectivity would not be tight enough to maintain steady/stable contact.

At the frequencies involved solid contact is needed. Any interruption on any active pin likely to crash everything.

Just my thoughts on the matter.

Socket connections for AMD processors are an array of slots. How can they be loose? I also checked the Ryzen afterwards and all rows of pins are completely straight and not loose. However, when I went to replace the chip, it did this time just drop in completely flush; I didn't have to push down on the corner where the yellow triangle is. I then tested it again, yet same result without issue. The connection is completely tight with no movement at all.

I then watched a YT video (don't have the link...) and the presenter said that Ryzen CPUs that have the yellow triangle on the underside usually just fall in flush without needing any external force, but that those with the triangle on the top don't always fall straight in - you sometimes have to push down where the triangle is.

 

[ptTimeBldr]

It's probably something peculiar to your specific CPU and socket combination. Something like a pin that's just slightly bent making a tight fit on a contact that's just slightly skewed.

It's a bit scary when they don't just drop into place, I know I'd be leery about pushing down a corner. I'd probably pick it back up first and then run a credit card between each row of contacts, in both directions, to make sure they're all perfectly aligned upright. It really should drop in place with no more than some slight jiggling side to side to help.

I certainly think there's something peculiar about the Gigabyte motherboard. Not sure about the chip. I did do a pin test, but with a utility knife blade not a credit card, and there were no snags in either direction.

 

[drea.drechsler]

Socket connections for AMD processors are an array of slots. How can they be loose? ....

The way AMD's PGA socket works....the pins go through holes the plastic top, then into a V shaped socket contact like a hotdog in a bun. When you close the lever it slides the plastic top sideways, moving the entire CPU and pushing the pin into the V shaped contact to make a tight electrical connection.

The V shaped contact could be just slightly skewed to one side so that the pin rubs against it. And if a pin is just slightly bent, not even enough to see, it might not perfectly align with it's hole in the top. There's over a thousand pins, the tolerances are tight and some may be equally slightly bent in the opposite direction. It doesn't take much for it to interfere on insertion and need a little assist but that's also the way a pin can get bent and a reason why I like to make sure pins are in alignment before pushing on it.

A credit card seems to be just the right thickness to set spacing between the pins. That's why it makes a good tool for the purpose of aligning pins in all directions.

And BTW: always make certain the lever is fully raised so the pins have maximum clearance of the contacts when you place it on the socket.

 

[Ralston18]

It does not matter which component has the pins and which has the slots.

The idea is that they fit together snugly and my thought is that "snug" does not include simply dropping into place.

However, AMD does seem to have "drop in".

https://www.amd.com/en/support/kb/faq/cpu-100

Keyword being "should".

And, if there is a pin out of alignment or slightly bent that would indeed require some additional force to get it all seated together. And if seriously bent or out of alignment then the install would fail with likely damage.

Manually straightening could salvage things but one result would be that a bit of a helpful nudge would be needed to seat it all.

Many electronic devices are assembled by robots that can precisely line up components and use just the right amount of force to seat things into place.

Much harder for us people types to do or do consistently perhaps.

Zero Insertion Force

https://en.wikipedia.org/wiki/Zero_insertion_force

All in all most likely a function of the physical measurements of pins and slots. Especially tolerances with respect to width, length, diameter, etc.

A pin pushing the max diameter would likely have some "drop in" problem with a slot pushing the minimum diameter.

With the numbers of pins growing it may all become even more problematic.....

Now yielding to the engineers.

 

[Karadjgne]

Think about the size of the pins, the size of the holes and your eyesight accuracy. Those connections need a nano-micrometer to guage the accuracy of the pin angle and socket V angle differences. If 1 pin is off even 1/2 degree, it can make a physical difference, but your eyesight is not capable of judging that discrepancy.

So you drop in the cpu, need to tap the corner, which re-aligns the pins to the socket. On second and consecutive drops, that 1 pin is now perfectly aligned, so no more tap required, or it didn't budge and another tap is required.

This was always a possibility with the PGA sockets. With the LGA sockets, if a single or more springs is slightly higher or lower than the rest, it doesn't matter, the cpu and hold down will squash them all to the same contact height anyway.

 

[ptTimeBldr]

Think about the size of the pins, the size of the holes and your eyesight accuracy. Those connections need a nano-micrometer to guage the accuracy of the pin angle and socket V angle differences. If 1 pin is off even 1/2 degree, it can make a physical difference, but your eyesight is not capable of judging that discrepancy.

So you drop in the cpu, need to tap the corner, which re-aligns the pins to the socket. On second and consecutive drops, that 1 pin is now perfectly aligned, so no more tap required, or it didn't budge and another tap is required.

This was always a possibility with the PGA sockets. With the LGA sockets, if a single or more springs is slightly higher or lower than the rest, it doesn't matter, the cpu and hold down will squash them all to the same contact height anyway.

I have a magnifying lens if truth be known, and I could not see any misaligned pins. They were all perfectly straight and perpendicular for every row.
All I know is, that the first time I had to push very slightly (not with great force) on the corner where the triangle is, and then it became completely flush all round with the socket!
I then repeated the test twice, again checking to see that all pins were straight - and they were, and on these occasions they dropped-in flush without any assist.
This tells me that there's nothing wrong with the chip, but it did get me to some extent the fact that I had to push it down on the first attempt. I've never experienced this before with AMD chips.

 

[ptTimeBldr]

It does not matter which component has the pins and which has the slots.

The idea is that they fit together snugly and my thought is that "snug" does not include simply dropping into place.

However, AMD does seem to have "drop in".

https://www.amd.com/en/support/kb/faq/cpu-100

Keyword being "should".

And, if there is a pin out of alignment or slightly bent that would indeed require some additional force to get it all seated together. And if seriously bent or out of alignment then the install would fail with likely damage.

Manually straightening could salvage things but one result would be that a bit of a helpful nudge would be needed to seat it all.

Many electronic devices are assembled by robots that can precisely line up components and use just the right amount of force to seat things into place.

Much harder for us people types to do or do consistently perhaps.

Zero Insertion Force

https://en.wikipedia.org/wiki/Zero_insertion_force

All in all most likely a function of the physical measurements of pins and slots. Especially tolerances with respect to width, length, diameter, etc.

A pin pushing the max diameter would likely have some "drop in" problem with a slot pushing the minimum diameter.

With the numbers of pins growing it may all become even more problematic.....

Now yielding to the engineers.

Thanks. I've read that page a few times already and I know it says that you shouldn't use any force for inserting the chip, but I had no choice and I only used slight pressure on the very edge of that corner with the triangle. That is where it was slightly raised, where the pins were just showing.

 

[Karadjgne]

I have a magnifying lens if truth be known, and I could not see any misaligned pins. They were all perfectly straight and perpendicular for every row.
All I know is, that the first time I had to push very slightly (not with great force) on the corner where the triangle is, and then it became completely flush all round with the socket!
I then repeated the test twice, again checking to see that all pins were straight - and they were, and on these occasions they dropped-in flush without any assist.
This tells me that there's nothing wrong with the chip, but it did get me to some extent the fact that I had to push it down on the first attempt. I've never experienced this before with AMD chips.

And what I'm saying is the tolerances are so tight between cpu and socket that 1 or several pins could be off angle as little as the thickness of a human hair and even with a magnifying glass you'd not be able to tell because it's too small to tell, but physically those pins would contact the \ / higher up the slope of one side / and not hit the space between them dead center. As soon as you gave it slight pressure, either the \ / spread or the pin aligned and requires no more pressure.

Either way, not a big deal.

Der8aur has a video on that subject, somewhere. Has to do with how the pins are soldered to the pcb and often are not exactly 100% 90° exact perpendicular to the chip, but can be a hair off angle.

 

[ptTimeBldr]

And what I'm saying is the tolerances are so tight between cpu and socket that 1 or several pins could be off angle as little as the thickness of a human hair and even with a magnifying glass you'd not be able to tell because it's too small to tell, but physically those pins would contact the \ / higher up the slope of one side / and not hit the space between them dead center. As soon as you gave it slight pressure, either the \ / spread or the pin aligned and requires no more pressure.

Either way, not a big deal.

Der8aur has a video on that subject, somewhere. Has to do with how the pins are soldered to the pcb and often are not exactly 100% 90° exact perpendicular to the chip, but can be a hair off angle.

So you think the chip should be ok? I hope so as well obviously, and I hope it's not a big deal in the end, because the new cooler has arrived and it fits the backplate perfectly, so I'm now very close to completing this build. If I get 'nada' or similar on the first power-up, I won't be happy.

 

[Karadjgne]

If you look at a Ryzen G, it's a sizable chip, and thats it. Anything remotely close to that corner is just pins, not cpu. I don't see there being an issue.

 

[ptTimeBldr]

If you look at a Ryzen G, it's a sizable chip, and thats it. Anything remotely close to that corner is just pins, not cpu. I don't see there being an issue.

Yes, the chip itself is well clear of the corners so agree there shouldn't be any problems, but the fact that it didn't go in flush the first time without assist, means I can't be 100% certain.