We're working with an EFM32WG940, and noticed an interesting problem with its QFN pads. I also took a look at some EFM32HG310 devices which I have at hand, and they have the same problem. The pads on QFNs normally consist of one metal area which goes from the bottom to the side of the device. However, these EFM32 devices have it wrong: there is some plastic inside the metallic area on the corner.
Here is a picture of the EFM32HG to better describe the problem:
I'm not much of a technologist myself, so we asked a friend who knows her technology well enough, and she said that ― in her opinion ― this package is faulty and when they see something like this, they just send it back.
So my question is, is this a normal occourence with EFM32 devices?
The mechanical specification for the QDN64 does show some plasticization between the pads and the edge of the package:
Looks like this is allowable for this particular package, and I do see this same 64 QFN in use by other vendors.
Sorry, meant to say QFN64 not QDN64.
@JohnB That's interesting. I haven't seen that variant before. So far I've used an EZR32WG and Si446x chips which haven't had this problem.
There do appear to be two different lead frame types for QFN packages.
A certain competitor of ours also uses both and has a document explaining this. As you've noticed, we use both types, as well.
I'm not sure if there's a reason one is used versus the other. It apparently has no impact manufacturability, so maybe it's just a legacy manufacturing decision.
As you've noticed, the EZR32WG uses the QFN64 without the extra package mold compound. I'm guessing this is the preferred package for newer devices, but we couldn't simply switch packages on the EFM32WG without causing headaches for existing customers, so the older lead frame style remains.
Thank you @JohnB for the details. I've got two concerns regarding this:
In my former life at another chip vendor, packages were also chosen based on lowest cost for the pins needed and, if a factor, heat dissipation ability.
Way back when, we used to have two different 144-pin quad flat packs. One was thicker than the other, but both had the same lead pitch (0.5 mm). As soon as we could, we moved everything to the thinner LQFP variant.
Why? The LQFP lead frame was not trimmed whereas the original 144-pin QFP used a molded carrier ring (MCR). The chip was then trimmed out of the MCR. Needless to say, that MCR was the cause of some manufacturing fallout (lead coplanarity rejects), so it had a higher cost even though we'd been in production on that package for some time.
The new 144-pin LQFP required no trimming. It used a pre-cut lead frame, was subject to a much lower rate of lead coplanarity fallout, and thus was a fair bit less expensive (50% lower package cost, IIRC).
To encourage customers to move to this new package, we actually passed along the savings as an enticement. That "carrot" worked very well such that when we finally EOLed the original 144-pin QFP a year or two later, there were very few holdouts.
I suspect the two different 64 QFN variants for EFM32WG vs. EZR32WG have a similar story, but the cost delta is probably not enough to make a wholesale switch to the newer version of the package.
@JohnB Thank you John, that's a nice explanation and a cool story. I now understand.
Ah yes. As a person who's hand-soldered a fair number of QFN packages, it's always a pain when you run across one of these. It's definitely not out of the ordinary, though, but I much prefer the packages with wrap-around metal leads...