Does Intel ever ship defective product in sealed retail box?

[pm]

John3850 [first]!! How would I KNOW if it's a ground contact?! I agree that there may be more than one pad devoted to the same ground, but even if they labeled those little gold pads, I'd need an electron microscope!!

Did anybody ever document which pads are either unused or grounds?

The pin out spec for the Core i7 2700K is here on page 90:
http://www.intel.com/content/dam/ww...eets/2nd-gen-core-desktop-vol-1-datasheet.pdf


And on the general topic of the thread, I work for Intel as an engineer, and I have spent a lot of time involved in the manufacturing and test aspects of our designs, and I have a lot of hours logged in the high-volume testing machines writing and optimizing test programs. And based on my experience I can say with absolutely certainty that every CPU that comes out of Intel is tested at a minimum of three times with increasingly larger and more complex test programs that are designed to check the correct operation of every transistor (at least that's the goal) and all of the pads on the package. I can easily imagine a manufacturing packaging visual defect (like a discoloration) slipping by - we don't tend to visually inspect each CPU for things other than gross defects - but I have a hard time trying to figure out how a serious electrical issue like a package pad being an open circuit would slip past. I guess it's possible - in quantum theory all things are possible - but it seems statistically improbable given the testing regimen that I know is used. If you do have one of these rare defective CPUs, then like the others, I encourage you to ask for an RMA and if you have issues doing this, then go ahead and email me - my email is in my Anandtech profile - and I'll see if I can poke someone from the inside.

Patrick Mahoney
Server Processor Division
Intel Corporation

* Definitely not a company spokesperson, merely an Intel engineer who posts on Anandtech *

 

[BonzaiDuck]

The pin out spec for the Core i7 2700K is here on page 90:
http://www.intel.com/content/dam/ww...eets/2nd-gen-core-desktop-vol-1-datasheet.pdf


And on the general topic of the thread, I work for Intel as an engineer, and I have spent a lot of time involved in the manufacturing and test aspects of our designs, and I have a lot of hours logged in the high-volume testing machines writing and optimizing test programs. And based on my experience I can say with absolutely certainty that every CPU that comes out of Intel is tested at a minimum of three times with increasingly larger and more complex test programs that are designed to check the correct operation of every transistor (at least that's the goal) and all of the pads on the package. I can easily imagine a manufacturing packaging visual defect (like a discoloration) slipping by - we don't tend to visually inspect each CPU for things other than gross defects - but I have a hard time trying to figure out how a serious electrical issue like a package pad being an open circuit would slip past. I guess it's possible - in quantum theory all things are possible - but it seems statistically improbable given the testing regimen that I know is used. If you do have one of these rare defective CPUs, then like the others, I encourage you to ask for an RMA and if you have issues doing this, then go ahead and email me - my email is in my Anandtech profile - and I'll see if I can poke someone from the inside.

Patrick Mahoney
Server Processor Division
Intel Corporation

* Definitely not a company spokesperson, merely an Intel engineer who posts on Anandtech *

Couple things . . . I will PM you on this.

The darn things seems to work perfectly in the Z77-A mobo where it's currently installed. I ran OCCT:CPU on it for 20 minutes at stock settings with the stock CPU cooler. I think the temperatures climbed into the low 60's C, but what would you expect with that cooler at those settings? The temperatures were remarkably close together on all cores. It just kept ticking along like the pink bunny.

I tried raising the multiplier to 4.2 leaving the voltages on "Auto." Under the same OCCT test, the temperatures climbed to a point where the two highest reached about 87, and OCCT terminated the test. Again, what else would I expect? I think the thermal monitoring -- all the motherboard settings in fact -- was set to default or "auto." And it was indeed the stock cooler. But I haven't turned up any kind of real failure.

As I mentioned, this is a gold contact pad on the row closest to the edge. The gold plate seems missing on the outside half. The remaining gold plate seems to extend up to -- or into -- the tiny dimple in the middle of those pads. This contact is on the side of the processor opposite the hinge of the motherboard latch-plate -- the side where the notch locks into place under a screw. I remember -- because when I inserted it, I tried to seat it as far forward toward the hinge before dropping the latch and securing the lever.

I'd seen that document yesterday, but I couldn't discover where to find if the pad was "unused" or "ground." Couldn't find any pinout diagram. Maybe you can point me to the right chapter and page(s).

More in the PM. Thanks for your post.

Oh -- OK -- I missed it -- you say page 90. Check your PMs.

 

[Burpo]

Wow, there's a lot more pins not used than I thought.. Looks like more than 100..

 

[BonzaiDuck]

Wow, there's a lot more pins not used than I thought.. Looks like more than 100..

I sent PM a PM. Basically, anyone can answer this question who knows more than I do.

There are really a few questions -- which I'll pose in a moment. I'm guessing there's a 60% chance from my lack of knowledge at this point that the flaw of the missing outer half of the gold plate on the contact is essentially a cosmetic defect. I say that first, because PM cites his knowledge that Intel electrically tests these processors three (freakin'!!) times before they go into the blue box. That's what I would expect of Intel. He later goes on to say that someone may give them a "visual" inspection, but the flaw may have been missed. I throw out the 60% chance just off the top of my head, because there are so many "black" pinouts in the page 90 schematic.

So . . .

1) What exactly do the black squares represent? "Dead" or unused pins? Or "ground?"

2) If "ground," do the other black pinouts continue to supply that full function to the processor? Or, is any particular ground contact discrete and separate to some subset group of pinouts?

3) The critical question. Which diagrams and which sides of those diagrams represent the edge of the circuit-board and the outermost pinouts closest and adjacent to the screw where the retaining lid locks down so you can secure the spring lever on the latch mechanism? The defect is among the outermost row of contacts/pinouts on that particular side of the processor.

Once I can match that edge to the diagram, I can determine whether the defective gold pad is among the black boxes in the schematic.

But the simple fact remains. They tested it electrically three . . . freakin' . . . times. And there is a somewhat high probability that the relevant motherboard pin is making contact with the remaining gold, and not just the "gray matter" (solder?) beneath it.

 

[Replay]

Of course, when you tell the truth, and say "three" (Windows 7 family pack), the recorded voice unhelpfully says you have violated the licensing (not true) and must buy another copy of Windows.

Or you tell the truth and say "one", (for this pc that you are trying to activate), and you must discover that the only correct answer is zero (a lie). Yet a slick operator can avoid that nonsense, just like an oem, and leave the retail keys in their packages (don't ask).

To the OP, bought, sometimes fixed, several boards with bad socket pins. There are so many duplicate pins for chip ground and voltage source that if your dodgy chip pad matches up with source or ground I would not worry at all about it operating ok.

 

[BonzaiDuck]

Or you tell the truth and say "one", (for this pc that you are trying to activate), and you must discover that the only correct answer is zero (a lie). Yet a slick operator can avoid that nonsense, just like an oem, and leave the retail keys in their packages (don't ask).

To the OP, bought, sometimes fixed, several boards with bad socket pins. There are so many duplicate pins for chip ground and voltage source that if your dodgy chip pad matches up with source or ground I would not worry at all about it operating ok.

I'm not. Worried anymore, that is. I just want to confirm that the contact pad with half the gold on it is a "black" box on the page-90-etc. pinout diagram.

Who . . . can answer . . . that question? Right now, it's a matter of probability. But even the testing regimen ALMOST makes it a moot point. I'd feel more comfortable with Intel manufacturing my automobile than if it were Toyota or Honda. I KNOW enough about QC statistics. I worked with W.E.D. for two years. I like to tell people the story about the old man going deaf, and how I got "picked up" by his junior colleague.

But . . . it's old war stories. Point is -- I don't know what they do with the OEM "tray" processors -- probably the same thing. But, given PM's explanation after I speculated about "what they do?" They have their freakin' act together!

 

[piasabird]

Maybe the retailer got ahold of a bad lot or someone threw some bad CPU's in some retail packages. If you make a million CPU'S some % of that amount have to be bad, based on statistics.

Hard drives get damaged a lot in the shipping process but CPU's tend to be light and probably just hold up better in shipment. Here is where maybe you need a photo of the CPU and the new retail box and send that into Intel as proof. CPU's are expected by most retailers to be returned to the retailer immediately if damaged when they arrive. Otherwise you have to get the warranty coverage from Intel. Good luck with that.

 

[Diogenes2]

Actually, once the box is opened. photos of something that was not visible from the outside of the sealed box, is not proof of anything..

Just talking about " proof " , not the validity of the situation as described.

Good luck with that.

I don't think Intel has a reputation for not treating warranty claims fairly.

 

[BonzaiDuck]

Maybe the retailer got ahold of a bad lot or someone threw some bad CPU's in some retail packages. If you make a million CPU'S some % of that amount have to be bad, based on statistics.

You overlooked PM's post. All the Intel processors go through three rigorous electrical tests -- brief, but rigorous -- before packaging. If they fail, they don't get to the shipping department. The visual inspection is the point where some minor flaw as the one I described would be caught -- or missed. So the probability of an CPU rendered unusable is almost 0; the probability of a cosmetic flaw is small, but higher.

They define their warranty period to eliminate as much as possible any RMA and replacement costs, based on that initial testing. Thus, the exception of "over-clocking insurance" we'd come to hear about.

QUESTION: Probably can be found in the 100-page Intel specifications, but maybe someone knows or has a fresh memory of it.

There is always a "pin 1" identifier on various parts, motherboard ports etc. In exploring this current issue, I came across a reference to such an identifier on the Intel CPU PCB. I think there's a little gold triangle on one corner, but maybe that isn't it.

Can anyone confirm? Once I can identify pin #1, the diagrams on page 90-etc. can be applied to determine whether the defective contact pad I'd discovered is among the "common ground" pads identified in the color black on the pinout diagrams.

And -- also. I had speculated that the "ground" contacts might be separate -- that a damaged ground would cease to function for the parts of the chip it "served." That can't be true. I can only imagine a "common ground" circuit-trace in the PCB connected to all those ground pins. If one of the ground contacts was "bad" -- the remaining ground contacts would equally serve.

 

[Diogenes2]

There is always a "pin 1" identifier on various parts, motherboard ports etc. In exploring this current issue, I came across a reference to such an identifier on the Intel CPU PCB. I think there's a little gold triangle on one corner, but maybe that isn't it.

Did you say you have the color coded diagram That shows the blacked out pins, with pin A1 in the upper right corner?

The blacked out pins are not pins, they are the blanked out notches in the socket. You should be able to align the notches ( key ) that have a unique pattern, so that pin A1 can be located..

 

[BonzaiDuck]

Did you say you have the color coded diagram That shows the blacked out pins, with pin A1 in the upper right corner?

The blacked out pins are not pins, they are the blanked out notches in the socket. You should be able to align the notches ( key ) that have a unique pattern, so that pin A1 can be located..

I'm confused about what you're saying. Are there not the slightly-ellipsoid gold contact pads associated with those black boxes?

What color are the "ground" and "unused" pins, then?

As I said to PM who is assisting on this, I wasn't going to remove the processor from its current working motherboard location until the next couple days. It could be tonight; it could be Sunday or Monday. I await a package in the mail with a better motherboard.

I think this will be resolved. Only mystery to me, is a reference to "blanked out notches." Or I'm wondering if the black squares in the pinout diagram have gold pads, or no? Because they are certainly included in the diagram and seem indicative and consistent with what I'd seen on the bottom of the PCB. There were colored pads in between the black boxes, and I know how the rows at the edges of the PCB are broken by absence of gold pads.

I won't be able to look at the CPU bottom until sometime in the next day or two. But I said that, so . . .

 

[Diogenes2]

Let me double check what I was looking at..

What I thought I saw, was that the black pads did not have associated pins/contacts in the socket... I'll double check and get back to you..

Here is what I came up with..

You are correct. All of the black slots are not devoid of pads.
It appears the only real keys in the package are the half circle cutouts, and not the rectangular void areas around the edge. That was what confused me earlier.. The location of pin A1 ( no actual pad ) should be the lower right corner when viewing the package from the bottom with the triangle in the lower right corner..

 

[BonzaiDuck]

Let me double check what I was looking at..

What I thought I saw, was that the black pads did not have associated pins/contacts in the socket... I'll double check and get back to you..

Here is what I came up with..

You are correct. All of the black slots are not devoid of pads.
It appears the only real keys in the package are the half circle cutouts, and not the rectangular void areas around the edge. That was what confused me earlier.. The location of pin A1 ( no actual pad ) should be the lower right corner when viewing the package from the bottom with the triangle in the lower right corner..

Well, Diogenes2 !! The odds that I have a common ground or simply unused pad with half the gold missing has gone up!!

I just pulled the 2700K from the socket. Flipped it over with the notches fitting the socket away from me.

There are six black squares in the A1 corner. They apparently have no gold pads. Just as an observation, going up from there on the right side, there are alternating green and light blue (white?) double boxes -- some which have a gap in the gold pads -- lower right quadrant. That's not inconsistent with the contact being made in the vertical column just to left of the right-most column, there just aren't any gold pads counting upward from the highest black box -- upper corner of the black triangle for A1 -- going up about five or so. That's where there are three pads missing, but I suspect intended contact in the column to the left. So if the rightmost diagram in your post has this as the correct "interpretation", counting left from the black box in the lower left corner at A1 shows that my defective pad is the seventh to the left, or counting two from the rightmost red box on the bottom row -- lower right quadrant.

It's black. So it either has no function, or it's common ground. I can look at the original Intel document that PM posted for pages 90 through maybe 95, but it appears that that defective gold pad serves no purpose that isn't fulfilled by another pin if it's just a "dead," unused pin/pad.

Schweet. Nice to know that for sure!! Or, if I go back to the 100-page intel document, I know now which gold pad it is.

Nothing to worry about with this processor!! RMA be damned!! Now I can even go ahead and lap the sucker, because I'd probably never need to use the warranty anyway.

I'm going to double-check this some more, but I can now match the defective gold pad to the pinout diagrams.

Also -- are those faint blue "minus" signs in some of those black boxes? I think so.

 

[BonzaiDuck]

Yeah! Confirmed. First, if dark-BLUE boxes look like they have a minus sign, it is really characters identifying VCCIO.

And lucky me. The pad corresponding to the box/pin-out counting to the right from the right-most red box/pad, bottom row toward A1 -- is just "a black box."