PLCC socket problems
nospam
Has anyone experienced problems with PLCC sockets where the IC contact
seems intermittent and possibly affected by variations in temperature.
We have a problem with a Z80 micro and PLCC socket where the micro
seems to behave erratically and a socket problem is suspected.
Changing socket manufacturers didn't solve the problem.
1 Lucky Texan
I have had to 're-form' both socket contacts (cantilever, and
bifurcated haispring) as well as flip chips over and carefully re-form
their legs with a scribe.
More problems with the chips I think. Particularly if they are handled
roughly/frequently in partial tubes. They band into each other and the
pins suffer. Mostly the corner pins.
Try upgrading to bifurcated contacts and inspect the chips.
You don't think a thermal spec is being violated do you?
good luck
D Yuniskis
> Hi
>
> Has anyone experienced problems with PLCC sockets where the IC contact
> seems intermittent and possibly affected by variations in temperature.
What led you to believe/suspect this? Do mechanical tweaks to
the assembly *while* it is misbehaving cause noticeable changes?
I.e., power down, reseat the device and power back up again, etc.
> We have a problem with a Z80 micro and PLCC socket where the micro
> seems to behave erratically and a socket problem is suspected.
And you're *sure* it's not software, right? :>
Does "erratically" mean "totally crazy" or just "not as we expect
it to behave"? I.e., if you were to write a piece of code that
just ran a simple scope loop, would you expect it to *also*
"behave erratically"? (i.e., this would be something easy to prove!)
> Changing socket manufacturers didn't solve the problem.
Are you sure the contacts on the Z80 aren't oxidized or
otherwise contaminated with something that is preventing
good contact?
With a high quality ohmmeter, have you measured the
resistance between the pin *at* the PCB foil and the
device itself? This requires some careful positioning
of "probes". You can also "gut" a package (one part is a
small price to pay for an answer) and probe from the bond
wires "out".
Have you tried *soldering* the device to the board (make or
buy a little adapter that lets you solder the PLCC to the
adapter and the adapter to the circuit board)? This allows
you to remove the socket from the equation entirely (at least
while you attempt to identify the real problem)
HTH
larwe
> of "probes". You can also "gut" a package (one part is a
> small price to pay for an answer) and probe from the bond
> wires "out".
You make this sound so easy! Decapsulating plastic parts to the
leadframe - even without trying to preserve the silicon - is pretty
darn difficult.
I'd pick an EPROM or ceramic part in a compatible LCC footprint and
crack the lid off - much easier.
David Brown
For sockets that were used a lot, I've sometimes had to use a fine bent
pin to pull out all the connection springs to stretch them a little -
that has often helped.
But most of the problems I've seen with PLCC sockets where when
customers changed the chips themselves. I've seen what happens when
people can't work the chip extractors, and use a large pair of pliers
instead. And it doesn't matter how well you label the board and the
chips, and tell people they only fit one way - with the help of a
hammer, customers can always get the chip to fit the way they want it to.
Rocky
We once found the a similar problem that was caused because the
sockets were gold plated and the ICs were tin/lead. The lead and gold
form a non-conductive amalgam. It turned out that it was possible to
buy tin plated sockets which solved the problem.
D Yuniskis
That doesn't tell you anything about the Z80 that is believed to
be (at least part of) the problem!
larwe
> >> of "probes". You can also "gut" a package (one part is a
> >> small price to pay for an answer) and probe from the bond
> >> wires "out".
>
> > I'd pick an EPROM or ceramic part in a compatible LCC footprint and
> > crack the lid off - much easier.
>
> That doesn't tell you anything about the Z80 that is believed to
> be (at least part of) the problem!
Your posting that I have requoted above appears to be asking the OP to
do a continuity check from the leadframe to the PCB foil, nothing to
do with the die.
D Yuniskis
Correct. His concern is that there *appears*/suggests that there
is a problem in which "the IC contact seems intermittent". This
involves two components -- an IC and a socket. *That* IC and
*that* socket.
Note that I never made any claims about the *die*. Rather, I was
commenting on the PHYSICAL PART that the OP was having problems with.
In particular, how it mates with the other physical component that
the OP was having problems with -- the socket.
Are you advocating he check continuity from ANY similarly sized
PLCC package from ANY (possibly different) manufacturer and, from
that, come to a conclusion about *this* PLCC package from *this*
manufacturer??
Why not test some *other* PLCC part in some other *vendor's*
PLCC socket and come to conclusions that you hope apply to *this*
part in *this* socket??
If you think you can come to conclusions using a *different*
PLCC component than the one in question, then just buy a PLCC
test harness and plug *that* into the socket in question.
Google (images) for "plcc plug". You'll find lots of off the
shelf solutions for this problem (I have a PLCC68 in my cable bin).
Buy one. Probably $100 (less than an hour of your time).
And, from that, you will know whether *that* PLCC harness/plug works
in *that* socket -- and very little more about why the Z80 PLCC
appears *not* to.
D Yuniskis
This is usually a longer term problem caused by the corrosion
("galvanic corrosion") resulting from the dissimilar metals
being in electrical contact with each other.
larwe
>
> Correct. His concern is that there *appears*/suggests that there
> is a problem in which "the IC contact seems intermittent". This
> involves two components -- an IC and a socket. *That* IC and
> *that* socket.
Poor debugging logic. If he tests a shape-compatible LCC part and it
does have a problem, then the case is more or less proven. You are
asking him to jump straight to a process that ideally requires fuming
nitric acid and a fair amount of skill. Note: Not that I necessarily
believe the problem has been correctly diagnosed. A quick squirt of
freeze spray and/or a puff from a heat gun would probably help him
learn a lot. At the speed of a Z80, I wouldn't expect serious issues
with an in-spec socket and in-spec chip body.
Michael Karas
@w36g2000vbi.googlegroups.com>, zwsd...@gmail.com says...
As has been stated before in this thread...your idea to change the part
in evaluation is nonsense. It really does need to be his part in his
socket.
To bare the lead frame of a PLCC chip does _not_ require working
with hazardous chemicals. A piece of fine grit carborundum abrasive
cloth or paper laid on a nice flat surface can be used to sand off the
top of the PLCC chip in short order.
--
Michael Karas
Carousel Design Solutions
http://www.carousel-design.com
malcolm
We may have seen that - google fretting corrosion. IIRC.
There are some vendors of pre lubricated PLCC sockets, but not easy
to source.
eg
http://media.digikey.com/pdf/Data%20Sheets/Mill%20Max%20PDFs/Standard%20PLCC%20Sockets%20pg.93,%20Series%20940.pdf
We changed process to include a contact lubricant on the PLCC
devices, and have since had no reports.
D Yuniskis
> On Apr 16, 8:10 pm, D Yuniskis<not.going.to...@seen.com> wrote:
>
>>
>> Correct. His concern is that there *appears*/suggests that there
>> is a problem in which "the IC contact seems intermittent". This
>> involves two components -- an IC and a socket. *That* IC and
>> *that* socket.
>
> Poor debugging logic. If he tests a shape-compatible LCC part and it
> does have a problem, then the case is more or less proven. You are
He's already tried a different manufacturer's socket. So, either
two manufacturers have produced defective parts *or* the problem
lies with his Z80's (or, the problem is "otherwise").
> asking him to jump straight to a process that ideally requires fuming
> nitric acid and a fair amount of skill. Note: Not that I necessarily
No, you only need fuming nitric acid if you want to decap the
device with the intent of salvaging the die. I had stated that
the component could be *sacrificed*. His assertion is that the
problem is not something internal to the package -- so, assume the
die, bond wires, etc. are working perfectly... he assumes the
problem is with the connection between "chip" and socket.
He has stated that he has already tried another socket vendor.
And the results are unchanged. So, if the problem *does* lie
with the chip+socket, then it has to be THE *chip* that is tested,
not some other that could be manufactured differently, etc.
(e.g., what if the contacts are oxidized, enameled, etc. Or, the
dimensions of the leads and/or chip body are not "to spec")
Testing some other device's "connectivity" with the socket tells
him nothing about the device that he is actually *using*.
D Yuniskis
Exactly. An end mill -- or a surface grinder if you are *really*
skillful -- will make quick work of shaving the top off the
device. The trick will be ensuring that you don't significantly
impact the position and retention of the lead frame within the
*remaining* bits of epoxy. As such, you might have to mill off
different parts of different "samples" so that what remains is
still "testable" (*not* the electronics but just the lead frame).
Recall the Z80 is a rather small die -- compared to the size of
the PLCC. So, you have a fair bit of "mass" of lead frame that
can remain embedded in non-abraded epoxy while still exposing
"enough" for a test probe.
My money is still on something *other* than a bad *mechanical*
fit -- which is what the "connection" hypothesis seems predicated
upon (unless the sockets are deforming in the board preheat
cycle -- is this SMT or thru-hole assembly?)
Glenn
Do not use sockets. Socket usually sucks over about 10MHz ;-)
Instead use in-circuit upgrade?
PS: Could it be a faulty PCB/via hole?
D Yuniskis
> Do not use sockets. Socket usually sucks over about 10MHz;-)
>
> Instead use in-circuit upgrade?
Z80 has nothing internally to upgrade. I'd guess the socket
is there to:
- allow SMT component (the Z80) to be used on a thru-hole PCB
- allow ICE connection for development
- allow easy access to all the I/O's for cheap ICT, etc.
- allow optional daughterboard connection "on the cheap"
etc.
> PS: Could it be a faulty PCB/via hole?
OP hasn't given enough information about his implementation,
the actual symptoms observed, or the actual conditions under
which it works/fails. The focus on "bad socket" (despite the
fact that two different vendors' parts exhibit the same symptom?)
seems either premature or there is a lot we haven't been told...
(e.g., did the design *ever* work? Is the oscillator running
properly? etc.)
Dennis
"nospam" <nos...@nospam.nospam> wrote in message
news:jg6jq61ji1kj7qu2i...@4ax.com...
We had this problems with SMD sockets, different socket manufacturers didn't
fix the problem. Changed to a through hole socket fixed the problem. an
extreme fix but it seemed to work for us.
D Yuniskis
On 4/17/2011 1:06 AM, Dennis wrote:
> "nospam"<nos...@nospam.nospam> wrote in message
>> Has anyone experienced problems with PLCC sockets where the IC contact
>> seems intermittent and possibly affected by variations in temperature.
>> We have a problem with a Z80 micro and PLCC socket where the micro
>> seems to behave erratically and a socket problem is suspected.
>> Changing socket manufacturers didn't solve the problem.
>
> We had this problems with SMD sockets, different socket manufacturers didn't
> fix the problem. Changed to a through hole socket fixed the problem. an
> extreme fix but it seemed to work for us.
How did you determine that the sockets were at fault?
As you said, a bit "extreme" to alter the layout (so,
presumably, you had convinced yourselves that this
*was* the problem before undertaking such a big step).
Were you ever able to determine what *aspect* of the
sockets was the problem? E.g., deformation during
assembly, bad wash, "made out of spec", too flimsy,
contaminants, etc.? Or, perhaps some subtlety of
the *layout* that changed when you moved to the
thru-hole layout? Did you ever get any feedback from
socket vendor(s) regarding the problem?
As with the OP, did the fault appear to be temperature
correlated?
I.e., do you now have a "policy" of *never* using SMD
sockets in designs? Or, have you been able to qualify
*when* they can be used and when they shouldn't?
(device size, types of signals involved, etc.)
Out of curiosity, what size socket and what sort of
device (technology) was *in* the socket?
Dennis
"D Yuniskis" <not.goi...@seen.com> wrote in message
news:iofb91$g1i$1...@speranza.aioe.org...
> Hi Dennis,
>
> On 4/17/2011 1:06 AM, Dennis wrote:
>> "nospam"<nos...@nospam.nospam> wrote in message
>
>>> Has anyone experienced problems with PLCC sockets where the IC contact
>>> seems intermittent and possibly affected by variations in temperature.
>>> We have a problem with a Z80 micro and PLCC socket where the micro
>>> seems to behave erratically and a socket problem is suspected.
>>> Changing socket manufacturers didn't solve the problem.
>>
>> We had this problems with SMD sockets, different socket manufacturers
>> didn't
>> fix the problem. Changed to a through hole socket fixed the problem. an
>> extreme fix but it seemed to work for us.
>
> How did you determine that the sockets were at fault?
> As you said, a bit "extreme" to alter the layout (so,
> presumably, you had convinced yourselves that this
> *was* the problem before undertaking such a big step).
>
The problem could be "fixed" by tweaking the contacts or squeezing the sides
of the socket in by hand. ie Squeeze it works, let go it fails.......
> Were you ever able to determine what *aspect* of the
> sockets was the problem? E.g., deformation during
> assembly, bad wash, "made out of spec", too flimsy,
> contaminants, etc.? Or, perhaps some subtlety of
Poor cleaning was something we suspected but ruled out. We also looked at
oxidation of the contacts on the package. The socket frames deformed
slightly when the ic was inserted.
> the *layout* that changed when you moved to the
> thru-hole layout? Did you ever get any feedback from
> socket vendor(s) regarding the problem?
>
Our volume was so small we didn't worry about conversing with the socket
manuf.
> As with the OP, did the fault appear to be temperature
> correlated?
>
Probably not.
> I.e., do you now have a "policy" of *never* using SMD
> sockets in designs? Or, have you been able to qualify
> *when* they can be used and when they shouldn't?
> (device size, types of signals involved, etc.)
>
I prefer to not use sockets anywhere if possible.
> Out of curiosity, what size socket and what sort of
> device (technology) was *in* the socket?
IIRC 128 way Lattice pld or similar, it was about 10 years ago.
nospam
We suspect fretting corrosion. Tapping the board or the micro can
make it reset. Removing the micro and cleaning the pins helps for a
while but only for a few weeks/ months.
What lubricant do you use and how do you apply it? Do you use it in
the field?
D Yuniskis
On 4/17/2011 6:22 PM, Dennis wrote:
> "D Yuniskis"<not.goi...@seen.com> wrote in message
> news:iofb91$g1i$1...@speranza.aioe.org...
>>> We had this problems with SMD sockets, different socket manufacturers
>>> didn't
>>> fix the problem. Changed to a through hole socket fixed the problem. an
>>> extreme fix but it seemed to work for us.
>>
>> How did you determine that the sockets were at fault?
>> As you said, a bit "extreme" to alter the layout (so,
>> presumably, you had convinced yourselves that this
>> *was* the problem before undertaking such a big step).
>
> The problem could be "fixed" by tweaking the contacts or squeezing the sides
> of the socket in by hand. ie Squeeze it works, let go it fails.......
So, the socket wasn't mechanically stiff enough (?).
Interesting. Most of the PLCC sockets I've used were
large and bulky. "Flimsy" wouldn't be a term I would
have applied to them.
>> Were you ever able to determine what *aspect* of the
>> sockets was the problem? E.g., deformation during
>> assembly, bad wash, "made out of spec", too flimsy,
>> contaminants, etc.? Or, perhaps some subtlety of
>
> Poor cleaning was something we suspected but ruled out. We also looked at
> oxidation of the contacts on the package. The socket frames deformed
> slightly when the ic was inserted.
So, the pins nearest the corners probably exhibited the least
number of problems while those midway along each side would
be most vulnerable (?)
>> I.e., do you now have a "policy" of *never* using SMD
>> sockets in designs? Or, have you been able to qualify
>> *when* they can be used and when they shouldn't?
>> (device size, types of signals involved, etc.)
>
> I prefer to not use sockets anywhere if possible.
Agreed. Sockets/Connectors and Caps are the root of all evil.
But, in the event you *need* a socket in the future, you'll
just look for something "beefier"?
Dennis
Yep, mid way along the sides showed the biggest problems. Could have been
"fretting corrosion" which someone else mentioned. Re-seating the ic a
number of times would fix the problem temporarily.
>>> I.e., do you now have a "policy" of *never* using SMD
>>> sockets in designs? Or, have you been able to qualify
>>> *when* they can be used and when they shouldn't?
>>> (device size, types of signals involved, etc.)
>>
>> I prefer to not use sockets anywhere if possible.
>
> Agreed. Sockets/Connectors and Caps are the root of all evil.
> But, in the event you *need* a socket in the future, you'll
> just look for something "beefier"?
Preferably dont use sockets! Same for any interconnects, get rid of as many
as possible.
1 Lucky Texan
> Hi Dennis,
>
> On 4/17/2011 6:22 PM, Dennis wrote:
>
I have seen PLCC sockets that flex due to hairline cracks in the
corners. Likely caused by customers monkeying around or inserting
backwards as mentioned in other posts.
Dennis
"1 Lucky Texan" <alck...@swbell.net> wrote in message
news:cfe64582-2187-47d3...@t16g2000vbi.googlegroups.com...
=============================================
Also happens when the "base plate" is picked out to make acess for hand
soldering easier on prototypes.
D Yuniskis
On 4/18/2011 5:51 PM, Dennis wrote:
[attributions elided]
>>>> I.e., do you now have a "policy" of *never* using SMD
>>>> sockets in designs? Or, have you been able to qualify
>>>> *when* they can be used and when they shouldn't?
>>>> (device size, types of signals involved, etc.)
>>>
>>> I prefer to not use sockets anywhere if possible.
>>
>> Agreed. Sockets/Connectors and Caps are the root of all evil.
>> But, in the event you *need* a socket in the future, you'll
>> just look for something "beefier"?
>
> Preferably dont use sockets! Same for any interconnects, get rid of as many
> as possible.
Of course, that's not always possible (e.g., connecting to
the Field, etc.).
I find my biggest use of sockets is when using SMT devices
on boards that must be thru-hole (e.g., many devices intended
for 2nd and 3rd world countries where "service facilities"
are often scant and "behind the times").
The biggest problem I see with socketed components is that
it encourages (hmmm... perhaps too strong a word? maybe
"tempts"?) folks to remove the socketed device -- if only
to immediately reseat it. As if they are thinking, "Hmmm,
maybe this component is 'broken'?" even though they don't
really *do* anything with it once removed.
Often, they don't use the right sort of extractor to remove
the device -- it seems a tiny "standard-tip" screwdriver is the
extraction tool of choice -- and the socket gets beat up in
the process (even gouging contacts on the device!)
I had tried gluing devices into their sockets (this can be
a real crapshoot if you're sloppy with the adhesive). But,
this just aggravates the above behavior -- they sit there
*digging* at the component, wondering why it isn't coming out.
<frown>
Most recently, I tried "encapsulating" the device so that
it is obvious that the only way to remove it is to unsolder
the socket. We'll see how that works... :>
Memory modules are a real PITA. They (SIMM, DIMM, etc.) are
an economical and versatile way of handling memory configuration
options. But, where a socketed "IC" might "tempt" someone to
remove it, socketed memory devices seem to SCREAM: "Remove Me!!!"
And, they tend to be easier to mis-install thereafter (put
in the module slightly crooked so it *looks* like it is
in place but really isn't). Making matters worse, many
of these sockets have really low insertion cycle ratings
which leads to flakey long-term behavior.
I haven't, yet, found a way around this that leverages the
economy and ubiquity of the modules... :< Maybe pot the whole
damn device!?!
Hans-Bernhard Bröker
> I find my biggest use of sockets is when using SMT devices
> on boards that must be thru-hole (e.g., many devices intended
> for 2nd and 3rd world countries where "service facilities"
> are often scant and "behind the times").
But for exactly those same reasons it's at least equally bad an idea to
use a socket. You've just laid out the explanation yourself.
Those people who can't properly repair SMT boards for lack of tools or
skills will, because of that same lack, damage your sockets trying to do
the one thing (they think) they can: take out the socketed device, and
put it back in.
So, no, sockets are not the solution to that problem. Daughter boards,
OTOH, might be.
> to immediately reseat it. As if they are thinking, "Hmmm, maybe this
> component is 'broken'?" even though they don't really *do* anything
> with it once removed.
But still, it will appear to them to be the easiest thing they can do to
the whole device. By putting it into a socket, you've just turned that
CPU into the lowest-hanging fruit around, so don't blame people for
picking that first.
Say the only symptoms they've been able to find up to that point were
"power consumption higher than specified" and "CPU rather hot to the
touch". So unplug the CPU and see if that brings power consumption down
to what they usually get for a CPU-deprived board. If so, exchange CPU.
If not, make measurements right there at the empty socket pins until a
difference shows up.
> I haven't, yet, found a way around this that leverages the
> economy and ubiquity of the modules... :< Maybe pot the whole
> damn device!?!
If you're seriously worried about badly unqualified people tampering
with it: absolutely.
Well, either that or put a dye pack into the enclosure that will explode
into the perpetrator's face, plus in the complementary color to the dye,
a _huge_ sticker stating an acceptable equivalent of "Hey moron! We
wrote 'No user-serviceable parts inside! Seriously!' on the cover you
just removed, and we bloody well meant it!"
D Yuniskis
On 4/19/2011 1:08 PM, Hans-Bernhard Bröker wrote:
> On 19.04.2011 20:15, D Yuniskis wrote:
>
>> I find my biggest use of sockets is when using SMT devices
>> on boards that must be thru-hole (e.g., many devices intended
>> for 2nd and 3rd world countries where "service facilities"
>> are often scant and "behind the times").
>
> But for exactly those same reasons it's at least equally bad an idea to
> use a socket. You've just laid out the explanation yourself.
Some parts just aren't *available* in a thru-hole configuration.
Looking for some *other* component often leaves you with a
tougher implementation (if you can even meet the goals at all!)
> Those people who can't properly repair SMT boards for lack of tools or
> skills will, because of that same lack, damage your sockets trying to do
> the one thing (they think) they can: take out the socketed device, and
> put it back in.
Exactly! Just like folks pulling "tubes" out of TV sets (ages ago)
in the hope that this would magically "fix" a broken set...
> So, no, sockets are not the solution to that problem. Daughter boards,
> OTOH, might be.
Any sort of connector invites trouble. People mate things
"backwards" (regardless of the presence of a key), bend
connector pins, etc.
>> to immediately reseat it. As if they are thinking, "Hmmm, maybe this
>> component is 'broken'?" even though they don't really *do* anything
>> with it once removed.
>
> But still, it will appear to them to be the easiest thing they can do to
> the whole device. By putting it into a socket, you've just turned that
> CPU into the lowest-hanging fruit around, so don't blame people for
> picking that first.
That;s why I've resorted to making the socket+chip a single assembly.
In a sense, converting a SMT device into a thru-hole device. This
makes the particular device one of the *least* desirable things to
tamper with -- because the consequences of "breaking it" are very
obvious ("How will we ever repair *that*??")
> Say the only symptoms they've been able to find up to that point were
> "power consumption higher than specified" and "CPU rather hot to the
> touch". So unplug the CPU and see if that brings power consumption down
> to what they usually get for a CPU-deprived board. If so, exchange CPU.
> If not, make measurements right there at the empty socket pins until a
> difference shows up.
>
>> I haven't, yet, found a way around this that leverages the
>> economy and ubiquity of the modules... :< Maybe pot the whole
>> damn device!?!
>
> If you're seriously worried about badly unqualified people tampering
> with it: absolutely.
>
> Well, either that or put a dye pack into the enclosure that will explode
> into the perpetrator's face, plus in the complementary color to the dye,
> a _huge_ sticker stating an acceptable equivalent of "Hey moron! We
> wrote 'No user-serviceable parts inside! Seriously!' on the cover you
> just removed, and we bloody well meant it!"
The problem is making devices so that they *can* be serviced by
"local (um) 'professionals'" with the sorts of tools and skills
that they have available. Shipping something back to the US
for service gets expensive, fast!
Often, a little knowledge is dangerous. I.e., if they are skilled
enough to use a soldering iron, they *can* remove a thru-hole
device and replace it. But, since this represents "work"/effort,
they will look for easier things to try -- without necessarily
being aware of the consequences of those other actions.
I've looked at *soldering* SIMMs to PCB's as a way of "discouraging"
their casual removal/replacement, etc.
I've seen people plug power connectors into ATA drives UPSIDE DOWN!
When I exclaimed, "Didn't it *feel* wrong to be doing this when
you did it? I means, *really* FEEL WRONG???", they just shrugged.
(try it, some time. I can't see how you could *easily* do this
without hands the size of ham hocks *or* a pair of pliers!)
malcolm
>
> We suspect fretting corrosion. Tapping the board or the micro can
> make it reset. Removing the micro and cleaning the pins helps for a
> while but only for a few weeks/ months.
>
> What lubricant do you use and how do you apply it? Do you use it in
> the field?
We use spray electrical contact cleaner, applied to a bush, then to
the PLCC devices edges.
There was a different design PLCC socket from Tyco ( now TE
CONNECTIVITY ) , I think the 'current' number is 1571541-x - not
cheap, but worth a try if you have the slightly larger holes it needs.