Printed Circuit Board Assembly & PCB Design SMT Electronics Assembly Manufacturing Forum

Printed Circuit Board Assembly & PCB Design Forum

SMT electronics assembly manufacturing forum.


Excess Flux in BGA rework

monkey

#17683

Excess Flux in uBGA rework | 24 September, 2001

Hi. We are currently using an unconventional method to rework a micro BGA - removing component with hot air, fluxing the area, placing the part with pick and place, then reflowing it in the oven. The reason we are doing it this way and not with a rework station is because we have a large quantity to rework in a short period of time. The problem is that we are having failures and consensus is that it is due to excess flux. I was just wondering what sort of problems excess flux will cause in a solder joint, and how to monitor the amount applied to the pad. Has anyone had any experiences like this? We are using a no-clean flux - specifically Alpha's UP flux.

Thanks! N

reply »

Michael Parker

#17685

Excess Flux in uBGA rework | 24 September, 2001

First question - do you have a rework station? If so, are you experiencing the same type of failures there?

I have successfully used no-clean on uBGA's in the past. 6 mil solder bumps, 10 mil spacing. 88 sites. Are you reusing previously soldered uBGA's? If so, have you guaranteed to your self that the solder balls are sufficient for a second reflow? Can you get a micro screen and apply solder paste to the site instead of just flux? Or preferably, put the paste on the uBGA before placement.

Use your rework station for placement, you can then batch process through the reflow oven.

How are you verifying failures? X-Ray? What are the faults?

Are you very sure that your pick and place are putting the uBGA's on target? A certain amount of self alignment does occur. I could misregister placement by 3 mils in X or Y onto a 6 mil pad and still get good alignment after reflow. If you must use your pick and place, consider the following:

A hint for X,Y programming. If you are using mm, don't, if you have mil. capability. Mils will give you more accuracy. Check it yourself - for example: X= 44.5mm, you want to change to X= 44.6mm, you find out that the change in X is too much, what you really need is X= 44.55mm but the pick and place software won't let you. (not enough significant digits available to the right of the decimal point) Answer- Reset the machine and before you load the specific program, configure the machine for mil. units of measure. When the program is loaded, the operating software should do the conversion from mm to mil. Now you have about 7 more X coordinate choices in mil. than you had in mm and you will be able to specify the X= more accurately. I hope all of this makes sense.

In my past experience, the biggest problems for uBGA was placement accuracy, followed by part coplanarity. Check your paste height and print alignment accuracy to ensure the best chance to solder the parts right.

reply »

#17688

Excess Flux in uBGA rework | 24 September, 2001

Mike's comments are well taken.

Your technique works fine with BGA, but uBGA are less forgiving.

Again, let's take about: * Specifics of the problem. * Amount and method of applying flux. * What's holding the uBGA in-place before the solder ball and pad dress reflow.

reply »

monkey

#17693

Excess Flux in uBGA rework | 25 September, 2001

Hi, thanks for your input.

Yes, we do have a rework station - specifically a SRT Summitt 1000 station, but it causes us many problems and is down more than it's up...I do not think that using it would speed up the process at all. As far as accuracy goes, I believe our pick and place is placing them with a good degree of accuracy.

I think the main problem we have is controlling the amount of flux applied to the pads. We are using new micro BGAs - not the old ones - and applying a thick no-clean flux to hold the part in place. The amount of flux applied depends on the operator applying it, so there is no control here.

We are currently ordering a microstencil in hopes that it will control the amount of flux applied. Any experience with microstencils? We are going to try it with flux and with solder paste and see which gives better results.

We are finding our failures when the boards go through test. The part we are replacing is the power to the pcb, so if it does not power up, then something is not right with the connection. We do x-ray the boards, but our x-ray machine does not have the capability to detect opens.

If we do find misalignment is an issue, that is a good idea to use our rework station to place and then batch reflow. Thanks...

N

reply »

monkey

#17694

Excess Flux in uBGA rework | 25 September, 2001

In regards to your last question, - problem is failures and most think it is due to too much flux - we apply flux with a brush - amount applied depends on the operator applying it - the flux is very thick and will hold the part in place before reflow

N

reply »

Michael Parker

#17696

Excess Flux in uBGA rework | 25 September, 2001

Since you are mostly concerned about the flux and suspect an application issue, you need to do a few things to validate whats going on.

1. Forget about speed of processing, length of time to do the job. Focus on accuracy first, then speed. with that said, get your SRT fixed! Is it out of calibration or do you have inexperienced operators making mistakes? Do you have good profiles or unsure?

2. You are only applying flux? What is soldering the chip to the board? Are you relying on the solder bumps to make the connection? If so, those bumps probably have a higher melt temp than solder paste and what you are getting are cold, insufficient, or open solder joints.

3. Get a micro screen and apply solder paste. Make sure your print alignment is accurate. Use 15X magnification to verify.

4. Take a failed board, remove the uBGA, microscreen solder paste to the chip and put the chip back using the SRT. Reflow. Test. Does it work? If yes, you have just proven that flux alone doesn't work. If it fails, then something other than the flux is the problem.

Good luck

reply »

Hussman

#17697

Excess Flux in uBGA rework | 25 September, 2001

If the flux you are applying by brush is alcohol based, use some alcohol on a few failed PCBs and retest. It souldn't take a lot of flux to reflow the balls to a pad that's already covered with solder and flux from the first BGA - unless you are cleaning them up really good before remounting. I've used what is called "sticky flux" to repair BGAs and all I did was dab the balls into the sticky flux (is dab really a word?). They reflowed fine and looked OK under X-ray. Not the best I've seen, but no voids or failures.

Can you slow your reflow process down to try and burn off as much flux as possible? Just another thought. It might take a couple of passes of the profiler, but could help.

Maybe try a water based flux and wash the boards after reflow? Might be worth a try too.

reply »

#17699

Excess Flux in uBGA rework | 25 September, 2001

Points in response are: * Microstencils work fine with paste, although they are a little fussy to use. * Never have used microstencils for flux. * Like Hussman, we dab the uBGA into a puddle of tacky flux and then place it on the board. * On controlling the amount of flux: - You're correct your paint brush scheme is too flakey - BIG timers have these fancy flux depth controller machines that squirt flux onto a plate. It has a wiper that goes around in circles [driven by a motor under the plate and its shaft poking perpendicularly through the plate ]. But then again, this is a production process thing, not a rework thing. - We use a credit card with a portion of the edge cut away to leave a tooth on each end. We drag it through the paste to level it out, before dipping the component.

The "problem is failures" are you kidding? Is that specific er what??? Failures!!! Git otta taun!!!

reply »

monkey

#17700

Excess Flux in uBGA rework | 25 September, 2001

1. SRT has been recently calibrated - problem is too many hands in the machine and careless mistakes. The profile is good and has been recently checked by a fellow from Genrad.

2. We are relying on the left over solder from the previous joint and the solder bump on the BGA to create a sufficient joint. We have cross sectioned 3-5 of these boards and found fairly good joints - no opens.

3. We are going to try this method. What thickness of screen would you suggest given that there is some solder paste remaining on the pcb? 3 mil? 4 mil?

4. We will try this. We did find that some of our boards passed after going through the SRT - probably because the excess heat allowed for the reamaining flux to burn off...

Thanks! N

reply »

Michael Parker

#17702

Excess Flux in uBGA rework | 25 September, 2001

Scavenge all the old solder off the board and the chip. Micro screen paste to the chip. As far as thickness, check your aspect ratio to determine the spec.

Eliminate just one variable at a time to get to the root cause. Start by not using the tacky flux, and using fresh solder instead. If you are satisfied that all is right but the unit still fails, then the problem is elsewhere.

Good luck

reply »

Casey Scheu

#17730

Excess Flux in uBGA rework | 27 September, 2001

Monkey, I have read all the replies you received. I have extensive experience in reworking these parts. There are several ways to approach your problem. If the chips are bumped you will need to have a stencil to apply the flux. If the chips are un-bumped you will need to stencil solder paste to the chips. The ratio is not one to one, a 1/1 ratio will cause bridging. If you would like to discuss this further please call me at 305-451-4722. I can also send an engineer in to your facility to address your problem. I can show you how to put your chips on the board in 3 minutes apiece. Including prep work. Check out the site. http://www.ape.com

Best of luck, Casey

reply »

Paul Christiansen

#17738

Excess Flux in uBGA rework | 28 September, 2001

I have had success by applying the paste flux with my finger, using a finger cot and spreading it on the PCB. This seems to put it on more consistently than using a brush.

reply »

#17747

Excess Flux in uBGA rework | 1 October, 2001

You got the right feeling for that I guess but what if somebody else is doing your trick, is it still repeatable? Personally I wouldn�t trust that method more than the brushing. Did you try the dipping method ? The thing from Dave with this "card-comb" to get a smooth controlled surface and depth of flux is so impressive, I read it with my mouth open, a superb example for KISS.

Dave you are great !!!

Wolfgang

reply »

PCB Soldering Tools