in need of plastic IC packager

Forgot to add... E-mail me for a recommendation for an _excellent_ layout guy who knows better than to do that. (He's in Columbus, OH, as if that matters... we often don't see each other for years at a time :-) ...Jim Thompson

One could consider breaking the design to two 4 x 3 chips..

I think what Jim is saying is that it may do that by itself.

Yes. The thermal shock during die bonding may well do the deed... instantly :-( ...Jim Thompson

Yes, a 3D die stacking arrangement. Brilliant!

It might help to understand why they think the wire length is an issue. I don't know the technical name for the problem (not being a package engineer), but the wire gets moved in the flow (wash) of the plastic. The wash can cause wires to short. You might be able to use a thicker wire to get around this problem.

They called this wire sweep, and I assume this is what they meant. My boss is pretty frantic at this point, as if we can't find a simple solution, they will end up wasting perhaps two months of work redesigning the channels of the design to be less elongated. We'd mentioned this a few times to the design team, and they mostly brushed it off. Now, MOSIS may not accept the design. We've skirted the edge of the wire bond design rules before, but don't want to go too far past normal practices.

Jon

Shortening them by a factor of approx. 1/sqrt(2) doesn't sound simple or fun.

eyes will roll.

can the design be tweaked to fit a 7mm x 7mm die, it'll cost more to produce, but if there's a rush.

Well now you've got me curious. Surely this shouldn't exist, then..?

(Everything else on the board is early 2008, so that should be 4th week

Size: 64 pins, looks to be 0.65mm pitch (TSSOP narrow body), unknown if there are exposed pads underneath.

I'm guessing the die is smaller than 8000 x 3000, but if they're using a standard lead frame, I can't imagine the bonding / encapsulation would be all that great?

And how do they fill any of these things in the first place? The resin must be extremely low viscosity, which makes no sense given the amount of solids in it (try burning an IC to ash some time; too hot and it sinters into a ceramic puck, with little shrinkage -- it's a dense composite to begin with). It can't be injected with too much pressure, or else the bond wires would get pushed around. And all the commercially available resins I know of, that have low viscosity, they suck!

Tim

I have a very simple solution. The problem is the die is not square enough. Instead of making one side shorter, make the other side longer. Do you really have to minimize the area of the die? Which is cheaper, a larger die or a complete redesign?

Or just 2 dies side by side? As long as you can reach over the redundant side to get to the chosen bond-pads.

Might help "yield" as well!

Yes! Good idea! Simply push out the pad ring on the narrow side(s). ...Jim Thompson

Ooooops! Maybe. The wire-bond angles may still be an issue. ...Jim Thompson

YUP, that IS the question! Since much of the layout work is being done by grad students, it is not as expensive per hour as a design bureau. Still, it could get expensive if it takes a long time.

MOSIS does charge by the square mm, so making the die square would about double the price, adding at least $20K.

Jon

There's your problem.... grad students >:-}

I'm sure my layout buddy could fix the layout for a whole lot less than $20K ...Jim Thompson

Sounds like the same problem. Just inquire about the fatter wire. It doesn't hurt to ask.

Non-standard bonding wire is one of those things that can get screwed up in high volume production. That is, you specify a fatter wire, and they substitute a standard wire. You might catch the wrong wire in an electrical test. But there is a possibility that the wrong wire size will not be caught at test, but will show up in actual use since heat flow is proportional to the cross section area of the wire.

Of course all sorts of assembly special orders are subject to the same problem. You would be shocked at the amount of product an assembly house can ruin. WTF this has to be done in Asia is a good question.

Fatter wire should have a larger bonding pad, but they may wave that.

How many are you making in total? Is the MOSIS order all that you will be doing ever, or is it to prove the design so that you can then go to production?

If you are only ever going to make a few dozen of these, then perhaps you don't need a standard plastic package. If there is no ceramic package that you could use for testing, perhaps chip-on-board with glob top or something like that.

Even if you want the standard plastic package, if you want to mount a few die with extra-long bondwires you could use MOPs (milled out packages, which are standard plastic packages that were moulded with no chip inside and then a hole was dug down to the leadframe, where this chip should go, using a milling machine). You can put the chip into the MOP, and then cover it with epoxy glob, but since this is not done by injection moulding it may not have the same problems with sweep of the bond wires.

Chris

For less than 20K, you can take the die to a custom lead frame fabricator. But MOSIS might charge you extra for not bonding the die, they would argue with extra precaution in handling the die.