Nicked an SMD or 2 off a RAM stick

[integral]

So long story short, I wanted to take a look at the RAM chips under the heatsink, but ended up slipping and knocking off 2 SMDs (I think they're capacitors?).
I have a sacrificial RAM stick which I can use to replace the lost SMDs, but I'm unsure if it's possible.
The solder pads are still intact, so I think it should be doable.
I have no experience with soldering nor do I have any solder equipment.

Images (sorry for potato quality)

Thoughts?

[pix3l]

Is it possible? Definitely. For a total beginner? Probably not. Depending on what the stick is worth (to you) you might be able to find someone close to you with the tools and experience to do it..

[integral]

Thanks for a reasonable reply. I asked on a tech forum and they told me the stick is as good as dead, even though I stated that it still works.

[sleemanj]

If it's just a couple of caps and it's still functional, I wouldn't worry about it, run through a ram test.

[helius]

One of the functions of decoupling caps is to block switching noise from propagating to other components in the computer. Certainly it may still work without them, but with a lower level of assurance against data corruption, so act accordingly.

[Mr. Scram]

If it's just a couple of caps and it's still functional, I wouldn't worry about it, run through a ram test.

There's a reasonable chance things won't be stable under specific conditions, though. Bad memory can ruin a lot including your data, so only use the stick with data you can lose without issue.

[integral]

If it's just a couple of caps and it's still functional, I wouldn't worry about it, run through a ram test.

It passed 1 hour of Karhu RAM Test, so I think it's still good.

One of the functions of decoupling caps is to block switching noise from propagating to other components in the computer. Certainly it may still work without them, but with a lower level of assurance against data corruption, so act accordingly.

If it's just a couple of caps and it's still functional, I wouldn't worry about it, run through a ram test.

There's a reasonable chance things won't be stable under specific conditions, though. Bad memory can ruin a lot including your data, so only use the stick with data you can lose without issue.

I'm an avid overclocker, so I tested each stick on its own. The stick with the missing SMDs couldn't clock nearly as high as the good stick (DDR-3600 vs DDR-4000+). Yes, I know I won't see the difference in real world scenarios, but as I said, overclocking is my hobby. I specifically bought these sticks in hopes that they use Samsung B-die ICs (they do) so I could have a bit of fun with them.

[james_s]

If you're worried about it, install a few 100nf chip capacitors in the appropriate physical size in their place. I'd bet that's what was originally there. The value of decoupling capacitors is usually not terribly critical and in most cases stuff will work acceptably with some of them missing.

When I'm designing things I tend to sprinkle in decoupling pretty liberally since they're cheap insurance against glitches, I don't bother trying to calculate the bare minimum I can get away with and I think a lot of gear is designed similarly.

[Mr. Scram]

If you're worried about it, install a few 100nf chip capacitors in the appropriate physical size in their place. I'd bet that's what was originally there. The value of decoupling capacitors is usually not terribly critical and in most cases stuff will work acceptably with some of them missing.

When I'm designing things I tend to sprinkle in decoupling pretty liberally since they're cheap insurance against glitches, I don't bother trying to calculate the bare minimum I can get away with and I think a lot of gear is designed similarly.

DDR4 isn't your typical kind of electronics, though. It's rather picky and systems involved need to be designed with much more care than usual.

[Mr. Scram]

It passed 1 hour of Karhu RAM Test, so I think it's still good.

I'm an avid overclocker, so I tested each stick on its own. The stick with the missing SMDs couldn't clock nearly as high as the good stick (DDR-3600 vs DDR-4000+). Yes, I know I won't see the difference in real world scenarios, but as I said, overclocking is my hobby. I specifically bought these sticks in hopes that they use Samsung B-die ICs (they do) so I could have a bit of fun with them.

The problem is that you don't know whether or not you used up your margin of error. Things like differences in temperature, wear or other things may result in errors you'd normally not encounter. Repair isn't easy though. If you're going to attempt that, I'd definitely recommend finding someone with equipment and some experience. The right person should be able to do it with some confidence.

[OwO]

I'm an avid overclocker, so I tested each stick on its own. The stick with the missing SMDs couldn't clock nearly as high as the good stick (DDR-3600 vs DDR-4000+). Yes, I know I won't see the difference in real world scenarios, but as I said, overclocking is my hobby. I specifically bought these sticks in hopes that they use Samsung B-die ICs (they do) so I could have a bit of fun with them.

That indicates it has already affected signal integrity and eroded margins. Since the RAM still works without these components I would say it's fairly certain they are bypass caps. I would personally just solder some 100nF caps there and run the tests again.

Definitely don't put small value C0G or any high-Q caps there, it can create all sorts of resonances on the power distribution network and worsen signal integrity.

[OwO]

DDR4 isn't your typical kind of electronics, though. It's rather picky and systems involved need to be designed with much more care than usual.

There is nothing special about DDR4, if you can design RF circuits up to a few GHz then DDR3/4 is no different. The only real hard part is layout design, since you have over 100 transmission lines to route.

[integral]

It passed 1 hour of Karhu RAM Test, so I think it's still good.

I'm an avid overclocker, so I tested each stick on its own. The stick with the missing SMDs couldn't clock nearly as high as the good stick (DDR-3600 vs DDR-4000+). Yes, I know I won't see the difference in real world scenarios, but as I said, overclocking is my hobby. I specifically bought these sticks in hopes that they use Samsung B-die ICs (they do) so I could have a bit of fun with them.

The problem is that you don't know whether or not you used up your margin of error. Things like differences in temperature, wear or other things may result in errors you'd normally not encounter. Repair isn't easy though. If you're going to attempt that, I'd definitely recommend finding someone with equipment and some experience. The right person should be able to do it with some confidence.

I'm an avid overclocker, so I tested each stick on its own. The stick with the missing SMDs couldn't clock nearly as high as the good stick (DDR-3600 vs DDR-4000+). Yes, I know I won't see the difference in real world scenarios, but as I said, overclocking is my hobby. I specifically bought these sticks in hopes that they use Samsung B-die ICs (they do) so I could have a bit of fun with them.

That indicates it has already affected signal integrity and eroded margins. Since the RAM still works without these components I would say it's fairly certain they are bypass caps. I would personally just solder some 100nF caps there and run the tests again.

Definitely don't put small value C0G or any high-Q caps there, it can create all sorts of resonances on the power distribution network and worsen signal integrity.

Thanks for the advice. I'll see if I can find someone who can resolder the missing SMDs.

[Mr. Scram]

There is nothing special about DDR4, if you can design RF circuits up to a few GHz then DDR3/4 is no different. The only real hard part is layout design, since you have over 100 transmission lines to route.

Whether it's special or not depends on your definitions, I'm just explaining why it's not just your run of the mill electronics and doesn't have as much margin for error as you may be used to with lower frequency stuff.

[integral]

I managed to take off the heatsink on both my sticks.

https://1drv.ms/f/s!AkfaD7fIG1SUhxz5yI3R6-pMfM5I

The missing SMDs seem to be pretty common place. I spotted similar looking ones on my 2GB DDR3 SODIMM sticks, which were left over after I upgraded my MBP 2012 to 2x4GB.

I emailed a soldering service and they said it looks fixable. Just waiting on their response after I emailed them those photos.

[Mr. Scram]

I managed to take off the heatsink on both my sticks.

https://1drv.ms/f/s!AkfaD7fIG1SUhxz5yI3R6-pMfM5I

The missing SMDs seem to be pretty common place. I spotted similar looking ones on my 2GB DDR3 SODIMM sticks, which were left over after I upgraded my MBP 2012 to 2x4GB.

I emailed a soldering service and they said it looks fixable. Just waiting on their response after I emailed them those photos.

I'd definitely like to see some controlled before and after tests to see the performance difference and how it stacks up to the undamaged unit.

[integral]

Soldering service successfully resoldered the missing SMDs.

Before

After

Can't wait to test the fixed stick.