Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

How to mount a QFN chipof 16 pins and the pitch of 0.5mm on a PCB?

Status
Not open for further replies.

mfarhan1

Advanced Member level 4
Joined
Apr 20, 2007
Messages
105
Helped
27
Reputation
54
Reaction score
13
Trophy points
1,298
Location
UK
Activity points
1,999
Dear all
I have to mount a QFN chip of 16pins and the pitch of 0.5mm i can not put it on PCB as i can not put it in place with hand. I dont have pick and place tool. I am thinking of using QFN socket any on with experience of using QFN, freq is 2.5GHz does the socket adds inductance that effect the result?
thanks
 

dendrite growth qfn

Use a tweezer, solder paste, glass magnifier and a heat gun will solve your problem.

Socket will alter your device characteristics and a 50 Ohm RF microstrip contact socket costs around $2000 or even more. If you are not concern with 50 Ohm impedance characteristics then Aries Electronics also makes good HF center probe socket and it costs around $500 ea.
 

qfn solder joint defects

"Use a tweezer, solder paste, glass magnifier and a heat gun will solve your problem. " and a steady hand. Talking from personal experience, its difficult at first (might even damage some chips), but eventually you'll manage so solder these by yourself.
 

qfn bandwidth

We use exensively Johnstech sockets (1mm eval bolt down):

advantages:

-very small
-pcb footprint compatible with QFN; it meand that you can do only one PCB in which you can use the socket or solder the QFN
-very large bandwidth (>20 GHz) and small parasitics

disadvantages:

- only 1000 insertions
-not well suited for temperature tests (they suffer cold tests)

cost is around 2k$

Mazz
 

56-qfn socket solderable

I weep when I see these sort of packages, because it means the demise of being able to breadboard circuits in my basement lab anymore!

In these sort of packages, especially for RF circuits, it is essential that the big central ground pad be attached to the ground on the board. The best way to do this sort of thing is to have a real board made up, and solder paste the various pads, put it on a hot plate, heat it up to soldering temperature, and use tweezers to move it around just a little to make sure it is aligned, and then to remove it from the hot plate and cool it off.

Sometimes you can just solder the center ground pad and use a very tiny soldering iron head to heat up the individual smaller attach pads. Luckily, on RF circuits, a lot of the attach pads are N.C.'w, so you do not have to get every last one to work.

In moments of desperation, I have breadboarded to these chips using purely manual methods. I drill a hole in the board. solder the edge pins to the top side pads. Then turn the board over and solder a small piece of solder wick to the backside of the chip, and tack solder the two ends of the solder wick to the ground plane. It is not pretty, but seems to work up to around 6 GHz.

You will need a microscope on a boom to even attempt any of this.
 

johnsteck socket

Hei!
The simplest way is:
* apply solder on each pad
* heat up with heat gun
* apply small amount of flux
* put IC
QFN have quite good self-aligment properties when heated, so it is not sobig problemm. Just do not affraid and that's all.
Good luck!
 

qfn hard to ground underneath

Some good advice is already given, but here is how we do it on our prototypes. Assumption is that you have double sided PCB with vias on central pad (usually 5, essential for HF applications):
1. apply small amount of solder on central pad vias on bottom side of pcb(with solder iron, just that vias fill with solder)
2. apply good quality flux on small peripheral pads on pcb
3. apply solder paste on central pad (not too much!)
4. mount pcb firmly but so you can reach central pad vias from below with iron
5. position chip on top side
6. heat central pad vias from below with strong iron with appropriate tip (SSC736A on Metcal SP200 in my case)
7. Have prepared tweezers and eventually reposition chip once solder paste melts (time window for positioning once paste has melt is 4..10 seconds or you will damage pcb). Somtimes repositioning will not be necessary (if central pad design is good and not to much paste is applied to it). Not enough solder paste will result in poor central pad connection and very probably with degraded HF performance.
8. remove heating from below until solder is solid (it can take more than few seconds !)
9. Use sharp tip (SSCC745A in my case) and 0.3mm solder wire and "feed" pcb/ic pad gaps (the flux applied in step 2. will help a lot)
10. Check pad connections electrically by understanding internals of IC . Some pad connections can not be checked easyly if there are several pads with same internal connection (Vcc, GND) but they have to be connected properly for correct HF performance. Use good magnifying glass to observer solder joint - after some time you will be able to spot the bad ones.

This works for me your milege may vary.

Regards
 

qfn solder balls

Here are some pics of my QFNs:




Another difficulty with 3x3mm2 QFN-16s is the two die connections on each side of each 4 pads. These die contacts sometimes solder to your outer pads and make a short circuit. Use of magnifying glass, solder wick and good flux paste is recommended.
 

qfn open socket die

You can use solder mask pen, used to repair solder mask defect on PCB, to cover the protected patterns, areas, traces, then apply solder paste to pads and land patterns. This will help tremendously to reduce solder bridges.

Use a tweezer to put the part in place, apply nice pressure with tweezer tip to the top of the package then turn on the heat gun. You can also build a modify nozzle for the heat gun to concentrate the hot air flow to a small selected area. The problem for first time experimenter are mostly at applying too much solder paste, applying too much of solder, and applying too much pressure to hold the QFN package in place and making it slips. Try it, it's not that bad as it sounds.
 

Re: QFN mounting

Biff

biff44 said:
I weep when I see these sort of packages, because it means the demise of being able to breadboard circuits in my basement lab anymore!

why you say this?

I've been using these (or similar) sockets up to 6 GHz and I'm happy with them.
Of course, it depends on what are your needs and your (company) budget.

To validate a new design (this is where I used them), usually you need to run several tests on several samples in your lab. Making 1 board with 1 socket to test 100 parts can be chaper then making 100 boards to test same parts.

With some limitations, of course, but with the advantage to have quick tests and larger accuracy (as the other components are always the same).

But I'm open to change my mind, if you offer valid arguments...

Mazz
 

Re: QFN mounting

Mazz said:
Biff

biff44 said:
I weep when I see these sort of packages, because it means the demise of being able to breadboard circuits in my basement lab anymore!

why you say this?

Mazz

Because they stink!

Connections are 100% under the body of the plastic housing means:
1) you can no longer inspect them to see if the pads are actually soldered (no visible fillet)

2) you can no longer rework them yourself with a soldering iron, you need a hot gas reflow unit heating below the substrate (since heating on top will melt the plastic package). This is especially true of most Microwave packages, that have a huge ground slug underneath them that is hard to solder.

3) Logistics cost money. You have a QFN package on a large board you are testing. It is not working. The first thing you suspect is a bad pad solder joint. You screw around with it as much as you can, and still can not get the chip to work. You drive over to your assy house (assuming it is not in Taiwan), and have them reflow it. A long time later you get it back. Wasted time! If there were at least SOME metal going up the side of the plastic package to allow wetting of solder, you could do a lot of repair work by yourself almost instantly.

4) Reliability. unless it is a big production run, you are never sure of the solder joint reliability, since it is not inspectable. What if you start getting field failures? Dendritic growth because of some trapped solder balls or entraped no-clean flux? Momentary contact due to flexing or thermal cycling of a solder pad joint that worked at initial test, but microcracked because the soldering conditions were not 100% ideal?

I guess if you are making 10,000 cell phones a day (with tightly controlled production), it is not an issue. If you are making 20 complex multichip microwave modules a week, you will get to really hate the QFN "revolution".

Rich
 
Last edited by a moderator:

QFN mounting

Rich

From what you are saying, it seems that the use cases are different. I am talking about protos & lab testing.

My experience in lab is really good, inspection is OK (no solder joint, as they have very small "s" shaped contacts.
Rework is easy, as you can remove the socket in 2 minutes and solder the QFN in the same socket footprint.

Mazz
 

Re: QFN mounting

Mazz, maybe we are talking about two different things. You are using a socket that is soldered onto the board, and then just press the QFN chip into the socket? How does that work, especially for chips like a 1 watt power amplifier that needs a good thermal heat sink to the bottom of the chip center?

Who's sockets are you using?

Rich
 

QFN mounting

Rich
I said it in my first post. Sockets I use are from Johnsteck, they are not soldered, but just mounted with screws. I use it for small signal (RF synth, transceivers). They are largely used in RF ICs mass production tests.
I have done an accurate comparison between socket and soldered measurements (same parts, same boards, only socket/soldered difference;
my results are:
1. Noise figure: 0.5 dB @2.4GHz worst with socket
2. Gain & Linearity: <0.1 dB @ 2.4 GHz difference between socket and soldered solution.

There are different models, if you need a good heat sink, I agree with you.
Some time ago Johnsteck came to us to show some new socket with a metal interposer designed for PAs, but I have no experience with that.

Mazz
Mazz
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top