<ddark>
Are there any backup plan for the RAM chip?
<DocScrutinizer05>
you mean for soldering?
<DocScrutinizer05>
there's not really an alternative to PoP, and the unique 1GB RAM chip I sourced is not the cause of the problems Nikolaus mentions
<ddark>
yes, the results for GTA04 seems to be really bad ones
<DocScrutinizer05>
it puzzles me how they build their Pyra, it also has PoP afaik
<ddark>
previous gen of GTA04 is also PoP ?
<DocScrutinizer05>
then OTOH I really dail to figure how the yield is so low, it can't be *that* hard to solder those chips, I mean Nokia and BeagleBoard both built a bazillion devices with OMAP3
<DocScrutinizer05>
yes
<DocScrutinizer05>
s/dail/fail/
<DocScrutinizer05>
not to mention TI themselves and their Zoom
<DocScrutinizer05>
and the millions of Blackberry and Samsung using OMAP4
<ddark>
on paper it should work, even with this combination - Nokia N9
<DocScrutinizer05>
yes, as I said above
<DocScrutinizer05>
and OMAP4 is PoP too afaik, as is OMAP5 probably. Just the chips are even larger than OMAP3, and thus even harder to solder
<ddark>
The problem appears to be the thermal effects during heat up and cool down how the DM3730 warps. And sometimes it warps 5% more and squeezes the BGA balls. Sometimes it warps 5% less and the balls do not give contact to the PCB.
<DocScrutinizer05>
doesn't make sense to me, last RAM chip and SoC used in GTA04 are exactly same as used in N9
<DocScrutinizer05>
I tend to blame their thermal reflow profile
<DocScrutinizer05>
however Nikolaus' rationale about plans to use same factory is sound
<DocScrutinizer05>
*maybe* I (or community) can get into contact with guys who used a succeeding thermal profile
<DocScrutinizer05>
... and procedure
<DocScrutinizer05>
(there might be more to it than just thermal profile. E.g. presoldering the PoP stack off-PCB. Or applying pressure on PoP stack during soldering. Or whatever)
<DocScrutinizer05>
on an unrelated note: first day of this year where I try to crank up my productivity to nominal, after flu and other health issues
* DocScrutinizer05
knocks on wood
<ddark>
So the community quest is it to find the factory that made PCBs with those chips.
<ddark>
btw how different is AM335x? can it be used as replacement for dm3730 for worst case scenario?
<DocScrutinizer05>
AM has different function blocks abd isn't a replacement. It's anyway not an alternative by formfactor
<DocScrutinizer05>
worst case scenario is: we have same issues with soldering _and_ a reduced feature set that's not backward compatible to N900
<DocScrutinizer05>
when using AM335x
<DocScrutinizer05>
((community quest)) exactly
<ddark>
... Salo production facility, its last in Finland, was to close ... Salo facility made Nokia's N8, N9 and E7 phones.
<DocScrutinizer05>
find engineers who worked there ;-)
<DocScrutinizer05>
I *think* (Ex-)Nokia-employees are tightly networked to their peer workers
<DocScrutinizer05>
so it might be possible to make the right contact by only 2 hops, ince you find a production dude
<DocScrutinizer05>
once*
<DocScrutinizer05>
I also have a hard time buying that "warp" story, particularly since the RAM PoP has only a few rows of BGA balls around the perimeter, so any warping should be largely irrelevant (and I doubt different makes of same chip differ so much that yield drops through the floor)
<DocScrutinizer05>
migh be a simple matter of (not) applying a larger quantity or different type of glue between the PoP stack 'sheets'
<Joerg-Neo900>
(only a few rows) actually just two
<Joerg-Neo900>
so unless the chip warps like a screw (which would be *really* nasty and unusual) how would and warping in or out (convex or concave) of PoP RAM make any difference?
<Joerg-Neo900>
s/and/any/
<Joerg-Neo900>
also where from does that "warp theory" origin? Where there any X-ray microscope shots done to support this hypothesis? Or how else did that get determined?
lobito has joined #neo900
lobito has quit [Remote host closed the connection]
e has quit [Quit: edk]
e has joined #neo900
ddark has quit [Quit: Page closed]
Kabouik has joined #neo900
jake42 has joined #neo900
jake42 is now known as jake42_
wpwrak has quit [Ping timeout: 240 seconds]
<jake42_>
the problem appears on the bottom of the DM3730, which has an outer and an inner ring with solder balls