How to identify RAM chips on PCB?**How do you line these things up????**

[rudeguy]

I'm bored at work and want to see if I can upgrade the RAM on a PCB. Problem is I can't figure out what chip is the RAM and there is no schematic for the board in working on.

Lil help?


-------------------

UPDATE:
Now that I have the RAM chips ID'd and new ones in my hand, I'm having trouble getting them lined up. They aren't like normal SMD's where the connections are outside of the chip, these are under it. So you put the chip down on the board and you can't see any of the connections.

Is there some type of guide that I am missing?

Of course, pics:

chip:

board:

 

[William Gaatjes]

What kind of ram do you wish to identify ?
Static ram or dynamic ram ?
Ball grid chips or chips with pins on the side ?

Can you post a picture of the pcb ?

You can check for cypress, oki or samsung as manufacturer, just to name a few brands.

 

[esun]

Probably your best bet will be to Google whatever markings you can see on the chip. If you can't see any markings or Google turns up nothing, you're probably not going to get far.

 

[PottedMeat]

Probably your best bet will be to Google whatever markings you can see on the chip. If you can't see any markings or Google turns up nothing, you're probably not going to get far.

this is about all you can do. also if you're dealing with standard DIMMs (DDR/2/3) using CPU-z and reading the SPDs may help

 

[rudeguy]

What kind of ram do you wish to identify ?
Static ram or dynamic ram ?
Ball grid chips or chips with pins on the side ?

Can you post a picture of the pcb ?

You can check for cypress, oki or samsung as manufacturer, just to name a few brands.

Pretty sure I figured it out. It looks like a Cypress chip and the darn thing is even still in production. I'm going to order some today to see if its possible to do.

Here is a pic of what I'm looking at with the chip I assume to be the RAM highlighted:

 

[Ketchup]

Is that even RAM? The Samsung next to it looks more like a RAM chip than the blank one does.
You've got my curiosity going though. What type of card is that?

Also, I haven't been in one of your threads since you left the cable company (good for you). Where are you working now?

 

[rudeguy]

Is that even RAM? The Samsung next to it looks more like a RAM chip than the blank one does.
You've got my curiosity going though. What type of card is that?

Also, I haven't been in one of your threads since you left the cable company (good for you). Where are you working now?

Its actually not a card, its a board out of a credit card terminal. The Samsung chip is an ARM chip, so I assume that isn't RAM.

Here is a link to the chip that I think is the RAM:
http://www.cypress.com/?mpn=CY62157EV30LL-45BVXI


Now I'm at a small company. We distribute prepaid cell phones and airtime to stores. The stores use the credit card machines to put minutes on customer's phones. I repair and program the terminals. Its fun because I basically get to tinker with stuff and solve random problems. For instance, the machine I want to upgrade the RAM on is old. I actually recommended we end of life it because it doesn't have enough memory to support newer applications. But then I got bored and started trying to see if it was even possible to upgrade those machines. Now that it looks like I can upgrade them, so I may have just saved the company a butt ton of money. Previously I found a way to do a repair the company was paying $60 to have done for $2. Considering the number of machines we have (its in the thousands), I saved the company a ton of money there as well.

Its a lot better than the cable company.

 

[Ketchup]

Cool, thanks for the info in the chip. Glad you enjoy the new job. I have yet to take one of those apart. I have two VeriFones in my "lab." One is quite old (I remember my first job at a grocery store had them) but it still works perfectly.

 

[imagoon]

Good luck on the replacement but know at times there are other chips that might need to be replaced or flashed. A version of an "SPD" chip that describes the RAM may exist. The processor also may be hardcoded to only use xMB because the vendor wanted to use easier to get larger chips then limited it for compatibility / price point. May just work also.

 

[rudeguy]

So how do you align connections you can't see?

 

[SecurityTheatre]

It's a ball grid array (BGA), and is generally not considered home-serviceable.

Here's a guy trying to use a pancake griddle to do it...
http://www.youtube.com/watch?v=OkjOwuSEzKU

I believe they either use machinery, or hot air... and verifying the work requires an x-ray.

Here is another video...
http://www.youtube.com/watch?v=KjKEmKUatJ4

 

[rudeguy]

It's a ball grid array (BGA), and is generally not considered home-serviceable.

Here's a guy trying to use a pancake griddle to do it...
http://www.youtube.com/watch?v=OkjOwuSEzKU

I believe they either use machinery, or hot air... and verifying the work requires an x-ray.

Here is another video...
http://www.youtube.com/watch?v=KjKEmKUatJ4

I have hot air, infra red....

My work setup:

no xray though

Thanks for giving me the proper term for it.

 

[imagoon]

When balled properly the chip on top of the BGA should "float" to the proper position. As mentioned above though checking requires an xray.

 

[rudeguy]

When balled properly the chip on top of the BGA should "float" to the proper position. As mentioned above though checking requires an xray.

OK thanks. I saw a couple of how-to's on proper "balling". This little side project is turning into a nightmare. But I'm in too deep to quit now!!!

 

[MrDudeMan]

I routinely solder 0.3mm pitch QFP and QFN parts all the time and those are a cake-walk compared to the alignment of many BGAs. I've successfully soldered BGAs using a skillet, but it's much easier in a reflow oven. I have a 2 stage reflow oven in my garage for such matters, but alignment is the hard part as the soldering can be done with hot air (usually). Hot air isn't good for high density BGAs because some of the balls will fail to flow correctly, which leads to fragile joints that will split if conditions are right.

Verifying via xray isn't absolutely necessary if the pitch of the BGA is relatively large. 1mm pitch and above should be possible if you are careful not to bump the PCB after you take your best shot at the alignment. Flipping the IC over and drawing a box around it helps to make sure you're in the center of the landing zone.

 

[rudeguy]

I routinely solder 0.3mm pitch QFP and QFN parts all the time and those are a cake-walk compared to the alignment of many BGAs. I've successfully soldered BGAs using a skillet, but it's much easier in a reflow oven. I have a 2 stage reflow oven in my garage for such matters, but alignment is the hard part as the soldering can be done with hot air (usually). Hot air isn't good for high density BGAs because some of the balls will fail to flow correctly, which leads to fragile joints that will split if conditions are right.

Verifying via xray isn't absolutely necessary if the pitch of the BGA is relatively large. 1mm pitch and above should be possible if you are careful not to bump the PCB after you take your best shot at the alignment. Flipping the IC over and drawing a box around it helps to make sure you're in the center of the landing zone.

Thanks.

I've been digging and it looks like an endoscope would be perfect for me. I'm now trying to find one at a reasonable price that I can use with my existing IR/hot air station. Or a combo station that isn't massive.

The machines I work on are worth roughly $100 each and I probably have 50 that I could use this on. Combine that with the fact that I may need to repair machines for other companies that have this problem and I can most likely justify spending $1000 to $1500. But if I can find a $200 scope to prove my concept, I could justify a more expensive purchase to use long term.

This may be a niche that I can really capitalize on. But I still can't prove that replacing these chips fixes the issue. Hence wanting a cheap way to see if it works, then going big afterwards.

 

[rudeguy]

We ended up ordering a BGA-300 from SMT Tools yesterday.

Definitely more expensive than my original estimate but we should be able to get some decent use out of it. They had a demo unit that saved us about $2k over what everyone else is charging for these.

God I hope I don't screw this up...

 

[redyouch]

Will the firmware automatically address the additional memory?

 

[MrDudeMan]

We ended up ordering a BGA-300 from SMT Tools yesterday.

Definitely more expensive than my original estimate but we should be able to get some decent use out of it. They had a demo unit that saved us about $2k over what everyone else is charging for these.

God I hope I don't screw this up...

How much was the BGA-300? Did you get quotes for the 100 or 200? If so, what were they?

Will the firmware automatically address the additional memory?

My guess is no. Usually in an embedded system there's module specific parameters in the firmware. The firmware defines a certain number of address bits, bank bits, data bits, and all of the timing parameters.

 

[rudeguy]

How much was the BGA-300? Did you get quotes for the 100 or 200? If so, what were they?



My guess is no. Usually in an embedded system there's module specific parameters in the firmware. The firmware defines a certain number of address bits, bank bits, data bits, and all of the timing parameters.

We spent a total of $4100 on the 300, upgraded prism, software, fiber optic light and reballing kit.

I did get a quote on the 100. Retail is $3k and they had a refurb for 1300. We probably could have made that work but the boss wanted to get the best. I was ok with that. If you are in the market, I can give you the guys info I worked with. He was awesome and really took care of me.

As far as the firmware, yes it will need to be upgraded. But I'm pretty sure I already have the software for that. But maybe not. I do already have some chips with the correct firmware. So if I can't update the machine from the computer like I normally do, then I'll be looking for a way to transfer chip to chip.

If that's even possible. ...

 

[sm625]

It is designed to be placed by a machine. You can place it yourself, as others have said when it is heated and fluxed it will float onto the lands, provided you get it somewhat close. But you gotta be real sure you have the right temperature, and subject it to a good temperature cycle test after it is soldered, cleaned, and dried. That's the only way to be reasonably sure you got a good solder connection, unless you want to pay to have it Xrayed.

 

[rudeguy]

It is designed to be placed by a machine. You can place it yourself, as others have said when it is heated and fluxed it will float onto the lands, provided you get it somewhat close. But you gotta be real sure you have the right temperature, and subject it to a good temperature cycle test after it is soldered, cleaned, and dried. That's the only way to be reasonably sure you got a good solder connection, unless you want to pay to have it Xrayed.

you just...stopped reading the thread huh?

 

[rudeguy]

well D Day is almost here. I am nearly ready to try reballing and upgrading these chips.

The scope just arrived late last week along with the reballing kit. Now I'm just waiting on the actual solder balls. I will be using my X-Tronic 5040 station for the work and the SMT Tools BGA 300 for inspection.

Here is my current plan of attack:

Tape off sensitive components using heat resistant tape.

Preheat the board using the IR preheat. With the preheat maxed out, I can get the board temp to the 130 C range.

Once the board is heated to temp, use the hot air portion to heat up and remove the old chip.

Clean old solder off the board using standard flux and solder wick.

Use the reballing kit and hot air to attach solder balls to new chip.

Use tacky flux on board and seat the new chip using the scope to ensure alignment.

With the preheat station still on, heat the board to temp.

Use the hot air gun with as low of air setting as possible to avoid moving the chip.

I'm figuring 30 seconds at 180 C for solder balls to melt and set.

Remove board from heat and let it cool.

Inspect again using scope.

I'm really hoping this works. There are just so many variables with the chips being so small and the contacts being extremely small (.3mm). If I can prove this, we will probably upgrade to a real BGA rework station.

 

[Colt45]

With the small amount of balls on that chip, I've never had a problem with just hot air by hand.

I've done it something like this
add solder to board, so the board has a full set of uh.. half 'balls'
add sea of flux
heat
drop on BGA
heat a bit more, prod it a bit
once satisfied it is centred on it's pads, push it down just slightly,
remove heat, presto

Always seemed to work alright. I wouldn't try it with anything with hundreds of balls or so, though, or send it to war.

 

[rudeguy]

With the small amount of balls on that chip, I've never had a problem with just hot air by hand.

I've done it something like this
add solder to board, so the board has a full set of uh.. half 'balls'
add sea of flux
heat
drop on BGA
heat a bit more, prod it a bit
once satisfied it is centred on it's pads, push it down just slightly,
remove heat, presto

Always seemed to work alright. I wouldn't try it with anything with hundreds of balls or so, though, or send it to war.

So you didn't use actual solder balls?

The balls came in yesterday and I gave it a quick run though. Looks like I had the heat up a bit too high and the balls overmelted.