need QFP leads bent the other way

J

Jon Elson

Guest
Hello,

We had an ASIC made through MOSIS, and they were supposed to make it die
down, but they have made it die UP by mistake. The only real problem is
the gull-wing bending of the leads goes the wrong way. Does anyone know
if there are tools to bend the leads after they have been sheared from
the leadframe? I assume the normal way is they break the corners of the
leadframe, but keep the edge strips on when forming the gull-wings. Now
that the edge strips are sheared off, it would be a lot harder to keep
the leads straight.

This chip is a 128-pin LQFP with 0.4mm lead pitch, if that helps any.

Thanks much in advance for any ideas.

(Yes, we are going to have MOSIS re-make the chip, but that is going to
take months, for sure.)

Jon
 
In article <bYWdnaoHGKoCIVTQnZ2dnUVZ_hSdnZ2d@giganews.com>,
Jon Elson <jmelson@wustl.edu> wrote:

We had an ASIC made through MOSIS, and they were supposed to make it die
down, but they have made it die UP by mistake. The only real problem is
the gull-wing bending of the leads goes the wrong way. Does anyone know
if there are tools to bend the leads after they have been sheared from
the leadframe? I assume the normal way is they break the corners of the
leadframe, but keep the edge strips on when forming the gull-wings. Now
that the edge strips are sheared off, it would be a lot harder to keep
the leads straight.
Click to expand...
Skilled hand labor with attention to detail will get it done on a small
scale. If the material is not so brittle it won't take backbending, I'm
pretty sure I could pull this off with a pair of pliers, a strong light,
and my fussy-work glasses. Lord knows I've done it with DIPs often
enough to salvage boards other people built. I've raised pins (and
soldered jumper wires on) on SSOPs for rework, but not QFPs that I
recall, but assuming similar material/process they are probably
backbendable.

If there were moderate volume, I'd probably go find a set of wide flat
pliers to make it more efficient. If you need a large volume done
cheaply, I could see a die set to backflip the leads en masse, but there
has to be enough volume to justify the tooling cost and fuss. If you get
fussy enough machining, you can even put guide grooves in those, but if
you handle leads nicely (and uniformly) they don't generally cross.

--
Cats, coffee, chocolate...vices to live by
 
Am 11.05.2011 00:34, schrieb Jon Elson:

We had an ASIC made through MOSIS, and they were supposed to make it die
down, but they have made it die UP by mistake. The only real problem is
Click to expand...
Why not just change the PCB?

Jan
 
On 05/10/2011 09:21 PM, Ecnerwal wrote:

If there were moderate volume, I'd probably go find a set of wide flat
pliers to make it more efficient. If you need a large volume done
cheaply, I could see a die set to backflip the leads en masse, but there
has to be enough volume to justify the tooling cost and fuss. If you get
fussy enough machining, you can even put guide grooves in those, but if
you handle leads nicely (and uniformly) they don't generally cross.
Click to expand...
Well, we have 85 chips or so in this condition. I think you
misunderstand the difficulty of working with leads of 0.4mm pitch, they
are REALLY tiny. It is very easy to pull the leads off the body, and
just looking at them funny will kink the leads sideways.

Jon
 
Jon Elson wrote:
On 05/11/2011 03:47 AM, Jan Lucas wrote:
Am 11.05.2011 00:34, schrieb Jon Elson:

We had an ASIC made through MOSIS, and they were supposed to make it die
down, but they have made it die UP by mistake. The only real problem is

Why not just change the PCB?
Well, we just made a test fixture with a $700 socket to test these
chips. The "foot" that mounts to the PCB is not likely to be re-usable.
MOSIS will have to re-do the chips for us, as they goofed.
So, it won't cost us anything but delay.

If I were to redo the board, it would take some time. The board is
pretty tight spacing, and 6 layers, and they cost us about $1000 to make
a small quantity. Then, I'd have to make a new test fixture to test the
chips before putting them on the boards. So, it would be a BIG hassle
and some expense to fix somebody ELSE's mistake.

Jon
Click to expand...
If the board is populated only one side, just move the pattern to the
other side of the board without changing anything, and mount the chip on
the bottom.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal
ElectroOptical Innovations
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058

email: hobbs (atsign) electrooptical (period) net
http://electrooptical.net
 
On 05/11/2011 03:47 AM, Jan Lucas wrote:
Am 11.05.2011 00:34, schrieb Jon Elson:

We had an ASIC made through MOSIS, and they were supposed to make it die
down, but they have made it die UP by mistake. The only real problem is

Why not just change the PCB?
Well, we just made a test fixture with a $700 socket to test these
Click to expand...
chips. The "foot" that mounts to the PCB is not likely to be re-usable.
MOSIS will have to re-do the chips for us, as they goofed.
So, it won't cost us anything but delay.

If I were to redo the board, it would take some time. The board is
pretty tight spacing, and 6 layers, and they cost us about $1000 to make
a small quantity. Then, I'd have to make a new test fixture to test the
chips before putting them on the boards. So, it would be a BIG hassle
and some expense to fix somebody ELSE's mistake.

Jon
 
In article <y6idnVPMmL_HcVfQnZ2dnUVZ_judnZ2d@giganews.com>,
Jon Elson <jmelson@wustl.edu> wrote:

On 05/10/2011 09:21 PM, Ecnerwal wrote:


If there were moderate volume, I'd probably go find a set of wide flat
pliers to make it more efficient. If you need a large volume done
cheaply, I could see a die set to backflip the leads en masse, but there
has to be enough volume to justify the tooling cost and fuss. If you get
fussy enough machining, you can even put guide grooves in those, but if
you handle leads nicely (and uniformly) they don't generally cross.


Well, we have 85 chips or so in this condition. I think you
misunderstand the difficulty of working with leads of 0.4mm pitch, they
are REALLY tiny. It is very easy to pull the leads off the body, and
just looking at them funny will kink the leads sideways.

Jon
Click to expand...
I'd be happy to give ruining one a try. Under NDA if it makes you happy,
not that I'd blab or snoop without one. Or you could send me a dummy if
you prefer. And if I fail to ruin it I'd be happy to quote you on the
other 84. While I have not tried this exact procedure on those exact
parts, I do understand what pitch you are talking about.

And here's an amusing (though not cheap) tool that might help - if a bit
fancier than I was thinking of going. Sort of one half of the die set I
made reference to.

http://www.topline.tv/QFP_tool.html

--
Cats, coffee, chocolate...vices to live by
 
On 05/11/2011 08:18 PM, Ecnerwal wrote:

I'd be happy to give ruining one a try. Under NDA if it makes you happy,
not that I'd blab or snoop without one. Or you could send me a dummy if
you prefer. And if I fail to ruin it I'd be happy to quote you on the
other 84. While I have not tried this exact procedure on those exact
parts, I do understand what pitch you are talking about.

And here's an amusing (though not cheap) tool that might help - if a bit
fancier than I was thinking of going. Sort of one half of the die set I
made reference to.

http://www.topline.tv/QFP_tool.html
Click to expand...
Holy COW!!! That's phenomenal! Thanks for digging that item up. I'm
not sure we could really do this lead bending on 86 chips, that's eleven
THOUSAND leads to bend manually. Also, we'd need to make up some kind
of frame to go around the chip body to perform the bend with the right
radius, and relieve stress at the package. But, I appreciate
knowing this device exists.

Jon
 
On 05/11/2011 03:13 PM, Phil Hobbs wrote:
Jon Elson wrote:
On 05/11/2011 03:47 AM, Jan Lucas wrote:
Am 11.05.2011 00:34, schrieb Jon Elson:

We had an ASIC made through MOSIS, and they were supposed to make it
die
down, but they have made it die UP by mistake. The only real problem is

Why not just change the PCB?
Well, we just made a test fixture with a $700 socket to test these
chips. The "foot" that mounts to the PCB is not likely to be re-usable.
MOSIS will have to re-do the chips for us, as they goofed.
So, it won't cost us anything but delay.

If I were to redo the board, it would take some time. The board is
pretty tight spacing, and 6 layers, and they cost us about $1000 to make
a small quantity. Then, I'd have to make a new test fixture to test the
chips before putting them on the boards. So, it would be a BIG hassle
and some expense to fix somebody ELSE's mistake.

Jon

If the board is populated only one side, just move the pattern to the
other side of the board without changing anything, and mount the chip on
the bottom.
Yes, that is a definite possibility. There are a bunch of caps on the
Click to expand...
back side, I could flip the caps to the front and put the chips on the
back. Of course, one of the issues is the boards are already assembled
on the back side, but that is just $25 worth of caps and such, not too
big a deal. We'd probably have to make a test fixture for the reversed
chips, too, that's a pain.

Thanks for the suggestion, although I had pretty much already thought of
that.

Jon
 
In article <_6KdnTYRWvnH1FHQnZ2dnUVZ_sednZ2d@giganews.com>,
Jon Elson <jmelson@wustl.edu> wrote:

Holy COW!!! That's phenomenal! Thanks for digging that item up.
Click to expand...
You're welcome. There are a few other types you can find.

I'm
not sure we could really do this lead bending on 86 chips, that's eleven
THOUSAND leads to bend manually. Also, we'd need to make up some kind
of frame to go around the chip body to perform the bend with the right
radius, and relieve stress at the package. But, I appreciate
knowing this device exists.

Jon
Click to expand...
Well, this might illuminate part of your problem approaching the
problem. It's not 11,000 leads to bend. Bending them one at a time is
crazy, and not the good kind of crazy. It's 344 sets of 32 (assuming you
have a square pack - or 172 sets of X and 172 sets of Y where X+Y = 64
for a rectangle) The tool above would be of some use in making sure that
any minor issues with the mass bends got sorted out, but all the way
back to DIPs, the way to bend chip leads is en masse. Grab one whole
side in smooth flat jaws, flattening the lower bend and grabbing at the
point of the upper bend, and back bend the upper - then slide out and do
the lower bend. You can make the radius purely by hand, or you could
dress the jaws to help with it. 8 bends per chip, plus clean-up.

--
Cats, coffee, chocolate...vices to live by
 
On 05/12/2011 04:52 PM, Ecnerwal wrote:

Well, this might illuminate part of your problem approaching the
problem. It's not 11,000 leads to bend.
Click to expand...
Yup, I tried bending them with the handle end of an X-acto blade, and
was able to easily bend 15 leads or so at a time, under the microscope.
It worked way better that I ever would have imagined. This allowed
the chips to be placed in the test socket on our test fixture. The BAD
news: After testing 3 chips, it is looking like the latest minor
revision screwed something up, and one pair of output pins of the chip
are stuck at the rails. These chips might still be usable in some
systems, but probably most users will want that feature that is not working.

Thanks for the ideas!

Jon
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top