Propeller QFN For Sale

JohnC

I've got 130 quantity of the QFN parts for sale. They're still in the tray. I opened the package from DigiKey to see what was inside, but otherwise they're untouched.

DK's best price right now is $7.49 each at that quantity, or $981.19 plus freight for the batch.

Yours for $850. Reply here or via message if interested.

John

JohnC

Any interest at $5.90 / each for the batch?

WBA Consulting

John, that's a very good price and I wish I could take you up on it, but my only released product using the QFN Prop isn't on the top of my to-do list. Too bad my PropBSC didn't get the attention I had desired....

Do you have the capabilities to place the QFNs?

JohnC

That's a nifty little board, Andrew.

We can do tqfp backwards and blindfolded, but our in-house qfn yield is patently dreadful. I bought these last fall as the tqfps went world-wide no stock, just to have for emergencies. Figured it'd be easier to respin a circuit board or two than lose 12+ weeks of sales.

Our inventory of the bigger chips has since been replenished with a big buffer. I wouldn't mind making these work, but currently don't have the clock cycles to fine tune a different process. Sadly, we're outside DK's return window at this point.

Mahonroy

I have been contemplating using the QFN package in some of our products lately (currently using the TQFP chips).

What are some of the problems you are experiencing with getting them to lay down correctly?

WBA Consulting

I would also like to know what challenges you were having so I can offer advice if relevant. I haven't had any issues with my QFN based boards, but I have some tools/tricks not necessarily available to everyone.

Mahonroy

Wanted to check back in on this? Any info?

JohnC

Sorry for being out of touch. The problem I believe has been with our stencil geometry, and later this week I'm going to have a different vendor take a look and make some suggestions.

Currently after reflow, we see high and low spots, which means the chip isn't lying flat on the board, and various connections aren't being made.

In the interim, I'd love to hear any tips and tricks people are willing to share.

John

Tracy Allen

If the stencil does not release the paste it can end up with pads high and dry. Bad release on several pads including the center pad can end up with a chip on a slant. Messy release = short circuits. Andrew gave me some good advice on this and he can speak much more about the specifics. I've gone through this with two different CMs. The CM should have an x-ray to enable looking at the solder wetting under the chip. Another decision is choice of flux, both for wetting and for subsequent cleaning.

There is info in this thread, including some closeup photos I took.
http://forums.parallax.com/discussion/149285/qfn-chip-solder-points-on-both-bottom-and-edges

Oh, I'm using only QFP now. Fewer layout and assembly headaches!

JohnC

What do you recommend for a center contact paste pattern? I've seen arguments for a single square the size of the pad as well as a smaller 'window' pattern of squares.

We've already edited the PCB layout so that the QFN pads extend further out and away from the chip itself.

WBA Consulting

Does your stencil geometry match the geometry in the Propeller datasheet? What type and thickness are you using? Do you have a way to verify paste height after printing? My 4 mil stainless stencil matches the datasheet exactly and works great.

You mention the part not laying flat on the board and that spawns several questions on my end. Mainly, it implies that you may not be getting full reflow on the solderpaste.
How's your reflow profile look?
What solderpaste are you using?
Are the defects in consistent locations on the QFN? If so, are they near copper pours? (thermal mass)

If you have any defective boards and are willing to ship them to California, I would be happy to x-ray them for you and give your more details as to the exact defect scenario.

Tracy Allen

There's plenty of guidance from manufacturers, more than you want to absorb, and much more than provided by Parallax in the data sheet...
-- Texas Instruments: sloa122.pdf QFN Layout Guidelines - Texas Instruments
-- NXP: AN1902.pdf, Assembly Guidelines for QFN (Quad Flat No-lead) and DFN (Dual Flat No-lead) Packages
-- Intersil: tb389.pdf PCB Land Pattern Design and Surface Mount Guidelines for QFN Packages
-- On Semi -- AND8211-D.pdf Board Level Application Notes for DFN and QFN Packages
The CMs I went through this with ended up with good yield after resolving a few initial problems, but I'm sure they were going more by their own experience and industry lore than by perusal of the tech literature.

Another issue with QFNs and DFNs is cleaning. Water soluble solder pastes have relatively high conductivity and it is difficult to clean the excess flux out from the tight QFN or DFN interstices. I became acutely aware of this when testing the conductivity between Prop pins, the same sort of tests I had used with the BASIC Stamp. There was lots of leakage (and by that I mean 10s of megaohms instead of 100s of megaohms). That was solved or improved by a change of flux and by special wetting agents while pressure washing the finished boards, also by decreasing the amount of solder on the center pad. 10s of megaohms don't matter so much in a Propeller circuit, unless maybe it is used for an analog function such as sigma-delta conversion with high impedances. On the other hand, it can be a big issue with analog front end chips, such as a sensitive oscillator or a pH or gas sensor. Those have to be squeaky clean, and it just doesn't make much sense to me to have them in a DFN package. But that is often how they are packaged.

JohnC

Thanks for the links. This'd be so much simpler if those tiny contacts around the outside edge were consistently useful.

WBA Consulting

Yes, there are many good Application Notes regarding QFN/DFN packages by other manufacturers, thanks for pointing some out.

In regards to aqueous flux and QFNs, we only use no-clean processes/solderpastes for BTCs (bottom terminated components) such as BGA, µBGA, QFN, DFN, LGA, etc. As a contract manufacturer, I deal with too many variables between customer designs and component manufacturers to enable a process that guarantees full flux removal under these components, even when we use additives in our wash (such as saponifiers). The cleanest (excuse the pun) way to address the concern is to use no-clean paste. We have also chosen a no-clean paste that can be processed in wash without causing cosmetic issues like hazing, so we able to have assemblies that are processed no-clean in SMT, but TH parts are soldered on selective wave machines with aqueous flux. We use the best tools in each area, so-to-speak.

Tracy Allen

In broad strokes I heard that no-clean pastes are otherwise known as "low residue" and suitable for less critical boards (the majority?), where it is desired to have "no clean" in the manufacturing process. They do leave residues, solids and activators, and those can bake on in a way that makes them harder than other fluxes to wash off if need be. The solder paste under a QFN releases a lot of flux that remains as residue between and behind the pads. I do see that it would be difficult to come up with any paste or process could guarantee squeaky clean. My complaint is more that so many analog chips where leakage and environmental protection are paramount concerns end up packaged as BTCs where cleaning is next to impossible, and there is no recourse to optical inspection.