7
\$\begingroup\$

Quad Flat Package is a rectangular component package with "gull wing" leads on all 4 sides. The smallest common package is QFP-32 (8 leads on each side). The question is specifically about QFP (TQFP/LQFP/PQFP/CQFP), and not QFN.

Were QFP-28, QFP-24 or QFP-20 ever used to package electronics? I have found 3D-models of these packages, but without any reference to actual parts, just the packages. So these models could've just been made by extrapolating the specification of a higher pin-count part.

I believe, I have seen a component packaged in QFP-20. It was an RF filter, and its datasheet had a technical drawing complete with all dimensions. Memory is not the most reliable source, and I couldn't find any information about it online.

\$\endgroup\$

3 Answers 3

9
\$\begingroup\$

Yes. A search quickly turned up the Freescale MC33591FTAR2, in an LQFP24 package.

MC33591FTAR2 in an LQFP24 package.

(source: NXP)

\$\endgroup\$
3
  • 2
    \$\begingroup\$ absolutely love it :) It looks so … biological? Like a squid sandwich? \$\endgroup\$ Commented Dec 12, 2025 at 10:20
  • \$\begingroup\$ Great example! Anything with even fewer pins, maybe? \$\endgroup\$ Commented Dec 12, 2025 at 14:39
  • 1
    \$\begingroup\$ @tpimh I couldn't find anything smaller, but admittedly I only searched for QFP20 and QFP16 on Farnell, Mouser, Digikey, LCSC, and a couple of search engines. The only results were QFN packaged ICs that had been erroneously marked as QFP, either by mistaken data entry or a typo in the datasheet. It is possible that there are other obsolete parts that wouldn't easily show up in those searches, or perhaps even some unusual asymmetric QFP packages that have different numbers of pins per side. I didn't count unpopulated clearance pins on HV parts as being valid answers either. \$\endgroup\$ Commented Dec 12, 2025 at 19:59
7
\$\begingroup\$

May I present to you the lowest possible pin count "QFP" package:

SOT-86

SOT-86 SMD Packages

These are used for a lot of medium power (a watt or so) GHz-class transistors and amplifiers (a lot of cable and satellite TV amps in this package). Maybe not as common as QFP-32, but has pins on all four sides, has four fold rotational symmetry, is SMD, and made in the millions so I think it still qualifies.

\$\endgroup\$
2
  • \$\begingroup\$ Technically, has "gull wing" leads, but not a QFP, body is circular, not rectangular. Nice package, though. \$\endgroup\$ Commented Dec 12, 2025 at 14:37
  • 7
    \$\begingroup\$ @tpimh You may call it a circle, I call it a square with corners heavily rounded off :) \$\endgroup\$ Commented Dec 12, 2025 at 20:59
2
\$\begingroup\$

warning: see end

hm. A "does it exist" question is very likely to be answered with "yes" anyways. You can buy custom packaging services, and I'm sure someone did combine a leadframe with fewer pins with a QFP-32-sized packaging mold. It's hard to stop engineers, especially when they have a customer with unreasonably deep pockets for custom solutions and a high cost for a board redesign (/me waves to the aerospace and especially defense designers), or when their frequency ranges demand very custom packaging anyways (side-eyes the Keysights of this world). That's probably the reason your RF filter had a pretty special package.

Let's rephrase this

Were QFP-28, QFP-24 or QFP-20 ever used to package electronics?

as

Were (or are) QFP-28, QFP-24 or QFP-20 used in mass-market industrially-rated devices whose functionality doesn't demand a non-standard package?

That rules out things like this specimen:

24-CLQFP package

… And then things get really sparse.

For something that's a bus driver, multiple-devices-per-package amplifier IC, or similar "has an input and an output side" thing, for 28 pins (and fewer) and below, the bilinear arrangement of (T)(S)SOP-28 is actually kind of advantageous for layout. (you still find these packages on new audio power amplifier ICs, for example, and on LED matrix drivers, and on parallel data bus drivers, and on polyphase switching power supply ICs, and the like.)

That would leave you with mostly digital logic devices like PLDs, GALs, and microcontrollers. And for these, market forces have mostly divided things into "less than 12 pads" and "more than 28 pads", as far as I can tell – my guess here is that if you're, say, ST, and you make a new microcontroller die, then you don't produce a different die for every different number-of-pins part in a MCU family; you make one die and connect it differently. But then that die needs to support the "largest pincount" in the whole family. Selling a device with fewer IOs than possible at the same production cost has little commercial utility – unless you can reduce device package size for people who care about SWAP. (absolutely hating hand-soldering fine-pitched TSSOP and the like: My guess is furthermore that the yield (or the defect rate) on no-lead packages in a reflow process is significantly better than for QFP of the same pitch.)

Now, these people wouldn't care about QFP at all; they've been in the QFN camp for decades, and have or are moving to WLCSP / BGA parts now, because most low-to-medium-cost board production processes allow them do so, so why waste space on pins that serve no functionality that much finer structures on the board itself could do, for free?

For higher-speed digital devices on multilayer boards, having to bring the contacts to the edge of a package is a complication and potentially a bandwidth limitation, and thus you'd want to have the contacts all across the bottom of the device instead of just the edges, too.

Together with these market and practical consideration, a bit of empirical evidence: When you peruse the filter box on the IC category on a part meta search engine like octopart.com, you'll only find RF-specialized "special" packages that are quadrilaterally flat-pinned and have less than 32 pins.

So, I'll go with a negative:

No, there aren't and weren't any non-specialized-package devices in these packages.

Edit based on counterexample to this answer

Ha! Clearly my answer that it doesn't exist is wrong; thankful for Polynomial to prove me wrong :) . So, not expecting any upvotes on this answer, and edited in warnings to not mislead a casual reader; still think it's worth having been written the way it is, and thus will leave it here.

\$\endgroup\$
6
  • \$\begingroup\$ Good answer but of course someone managed to prove it wrong :) \$\endgroup\$ Commented Dec 12, 2025 at 8:53
  • \$\begingroup\$ @Lundin you couldn't picture me happier about that! I think it's still worth having written what I wrote, so let me just add an addendum \$\endgroup\$ Commented Dec 12, 2025 at 10:21
  • 1
    \$\begingroup\$ "Selling a device with fewer IOs than possible at the same production cost has little commercial utility" That's engineer thinking ;-) The marketing/economics folks - bless them - think differently, and will quite happily use the fewer-IOs device for market segmentation reasons... \$\endgroup\$ Commented Dec 12, 2025 at 11:21
  • \$\begingroup\$ @psmears true! And I'd guess that might be what happened in the counterexample; in my head, however, the market segmentation of the time where QFP-32 became popular for MCUs was that of "low-pin count 8 bitters vs high-pin count luxury MCUs", where there was a conscious choice to kind of atrophy the middle (I can only name 1 (one) 16-bit architecture that's still relevant, for example, and while I fully understand why, with the availability of cheap cortex-M0 cores you'd rather directly jump to 32 from 8 bit as manufacturer, it doesn't quite explain the lack of competition when these were … \$\endgroup\$ Commented Dec 12, 2025 at 12:27
  • 1
    \$\begingroup\$ I don't know that I would call that microwave(?) module a QFP, even if the manufacturer does; it's a solidly RF package design that's well outside any of the common everyday packages. \$\endgroup\$ Commented Dec 13, 2025 at 3:53

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.