We are redesigning a board with a 32 pins chip from a 64 pins. For the old board, we have to use QFN64 due to the size. For the new board, we might be able to use either QFN32 or TQFP32. QFN does not offer much area savings, since we cannot route signals under the ground pad. Can we run a via/trace on the channel between the ground pad and outside pads? Any opinions?
QFN32 vs. TQFP32
Nov 29, 2009
12 Replies
If it passes your design rules, sure.
How many layers will you have?
John
2 layers, 60mmx10mm PCB. it's difficult to route all the signals on the outside. For the QFN64 (5mmx5mm), we currently route some signals on the channel (0.4mm) and the four corners. Vias would be very small and more expensive. There are very little clearances between traces and the ground pad. Soldermask is the only thing keeping them apart.
Do the arithmetic. You'll have a gap between the outer pads and the center pad of 20 to 25 mils. That means a via pad size of 15 to 10 mils if you use 5/5 rules.
If you have the space, easier to use a TQFP. You'll probably find the TQFP saves space since you can pop vias under the part. Plus, much easier to rework TQFPs.
Looking at my designs, looks like I avoid using QFN parts, choosing BGA over QFN.
Mark
I usually have a ground and power plane underneath TQFP packages so I try to avoid putting traces underneath a TQFP package.
Failure does not prove something is impossible, failure simply
indicates you are not using the right tools...
"If it doesn\'t fit, use a bigger hammer!"
--------------------------------------------------------------
Well, 7mm (9mm peak to peak) TQFP does not leave much room on a 10mm width board.
QFN has a big ground pad under the chip. We don''t really need the thermo dissipation, and it limits traces and via under it. We had about 10% burn through (during assembly) with traces under the ground pad.
Yikes, that sort of thing is hard to do on a 2-layer board.
John
? What does that mean ? Did you try to route traces under the PAD, and rely on the solder mask to protect you ?!
If you read the fine print on many QFN packages, they DO require that the inner PAD is soldered, I think there have been issues with the very small outer pads giving way under the cyclic stress
-jg
t
nd
Sorry, it's more like 90mmx15mm. TQFP64 is 14mm QFN64 is 10mm TQFP32 is 7mm QFN32 is 5mm
Actually, it's doable on 2 layers. But without vias on the inside, we are forced to do everything on top. And we want to keep cost low, so
0.3mm traces minimum.
Yes, I did. I confess.
I put in a smaller PCB pad and traces around it. This is just to get by until we can change the design.
Not possible with TQFP64, possible with TQFP32. But for TQFP32, we need another chip to multiplex I/Os.
No BGA available.
Ouch!
If you _really_ have to do this, then you will need to pay more for the solder-resist.
Get the best two-pot resist they have, and also ask about adding glass beads - those can give a proven finite gap. (used for gaping transformers)
- Of course, all of this adds costs....
-jg
nd
kI was thinking about adding epoxy on the edge of the ground pad. But the added thickness makes it more difficult to solder the outer pads.
It's a temporary solution. The boards work if it survives the soldering process.
Join the Discussion
Have something to add? Share your thoughts — no account required.
Didn't find your answer?
Ask the community — no account required