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

Printed Circuit Board Assembly & PCB Design Forum

SMT electronics assembly manufacturing forum.

SPC for Wave solder (defects on PCB's)

Views: 6439



SPC for Wave solder (defects on PCB's) | 7 February, 2006

Hello everybody,

I have been asked to establish a way to improve our wave solder process. I wonder where I should start. We collect data by inspecting our PCB's visually.

What's the best way to report data? Ocurrence of defects in total among all PCB's? or just defective PCB's? Also, how can you say that the wave process is in/put of control? I understand that a 50ppm defects is a widely accepted process limit control.

I highly appreciate your feedback.

Regards, George

reply »


SPC for Wave solder (defects on PCB's) | 7 February, 2006

hi, the only way you can say that the process is in/out of control is to plot the values (mean) of a metric against the 3 sigma limits for a Xbar chart, or plot a individuals run chart.

you can also do a capability analysis nased on a metric (say, for eg hole fill) to find out how well yr wave matches up to the process requirements


reply »

Cal Kolokoy


SPC for Wave solder (defects on PCB's) | 7 February, 2006

At the wave, you should collect data on the critical process parameters. The cliche being, it's better to control your process via SPC versus performing SPC on defect data. Perform a DoE to determine which variable interactions have the greatest effect on board quality, or if you know the process well enough, you'll be able to identify these yourself. From experience, I can tell you the critical variables include:

* Wave Contact Time * Contact Length (derived from conveyor speed) * Ramp rate and Preheat (temperature before wave contact) * Delta T

Find the appropriate profiling device that will measure the above parameters (there are several on the market), and you as the engineer will have to perform process characterization and then establish your USL / LSL.

reply »



SPC for Wave solder (defects on PCB's) | 7 February, 2006

Cal is right. SPC doesn't work at wave, period. Establishing parameters is very key for each product run. Then perhaps you can start grouping them. A WaveRider works well, but you have to use it everyday to make sure your parameters are correct.

reply »



SPC for Wave solder (defects on PCB's) | 7 February, 2006

I guess I'll start gruping the products and go from there to monitor process not products. Good idea!



reply »


SPC for Wave solder (defects on PCB's) | 8 February, 2006

Just to nitpick a little, collecting defect data and reacting to it really isn't an SPC function. Cal's recommendations (determine critical processes, assess process capability and acceptable limits) are spot on and are pretty much the definition of Statistical Process Control.

Tracking your output quality is helpful but it's not SPC. It's also a good way to come up with numbers for your resume, too.

reply »


SPC for Wave solder (defects on PCB's) | 13 February, 2006

SPC was a fad of the '80's and has been mis-used by the igonorant for decades now! Doing SPC on defects is a case in point.

Hell I implemented wave SPC programs before eating breakfast at 2 previous companies, and they worked great..resulting in was it 50% reduction in defects??!! You're right Steve...wave SPC on defects is good just to get accurate numbers on your most memorable one was on an old Econopak 229 with analog chip and lambda pump controls (10-turn pots) where the wave changed after every PM... we used to get skips on just about every SOT23.. I got it down to ZERO!

Anwywho...You'd think by now, people would've learned how to "engineer" - and how to do the due diligence of process characterization, DOE, and capability analysis, but alas, it goes back to people being more "managers" and not engineers.

reply »


SPC for Wave solder (defects on PCB's) | 15 February, 2006

Yes, doing the DoEs on your wave to achieve optimum settings that deliver the best results on your worst board is correct method. If not clear on the process, there are still wave solder process classes out there you can attend or hire in-house. If you have an Electrovert for example, you can hire one of their applications engineers for a day or two. A wealth of knowledge. The challenge is to keep your wave at those optimum settings. The ECD WaveRIDER instrument was designed just for that function. You run it before each shift. It measures conveyor speed, preheat thermal transfer efficiency, wave temps, parallelism, contact time, bandwidth, and implied immersion depth. This instrument provides a method to monitor the wave machine(s), thereby providing process control. The WaveRIDER software also has the ability to provide SPC charting of the critical parameters. Not to sound like a commercial, but the thing really does work. The tools are out there!

reply »


SPC for Wave solder (defects on PCB's) | 15 February, 2006


In my career, I've used Wave Solder optimizer and the WaveRIDER. Pro's and Con's to both. I did a GR&R on both as well, and WaveRIDER came out on top. The main challenge is to get the operators to use it. I've found that the operators were more comfortable with Optimizer's Direct LCD display and immediate feedback, versus having to deal with a computer and that damn new fangled windows software.

Ideally, a junior/Process/Wave Technician should be the guy who does this every day, but in these lean times, here at my company in Riyadh, I'm the guy doing this.

My latest project is to make "my own WaveRIDER". My SlimKIC 2000 has automatic wave detection and uses TC2 & TC3 channels as its dwell time sensors. Gonna design myself a custom pallet for this - FUN ENGINEERING WORK!

Also, see some past threads here on the Forum. The Malcom Dip Tester got some good reviews on their wave measuring device.

reply »



SPC for Wave solder (defects on PCB's) | 15 February, 2006

Samir, how did you perform the Gage R&R on this equipment? What was your standard?

reply »

Mr. Samir


SPC for Wave solder (defects on PCB's) | 15 February, 2006

Russ, On both instruments, I ran them through the wave 20 times each using identical profiles and wave parameters, and then used statistical formulas to measure the variation and repeatability of each. I had these all in Excel, but that was around 1999.I'll try dig this up somewhere... The parameters which I felt were worth measuring were:

1.) Dwell Time 2.) Contact Length 3.) Parallilism 4.) Delta T (temp delta before the last pre-heat and chip wave) 5.) Conveyor Speed

Note: #'s 2&3 can be derived from 1&5.

Don't use immersion depth. It's a bullshit measurement and neither device had a reliable method to measure theis.

Each gave slightly different readings but both were within 0.5 seconds, and 10 Deg. remember, each instrument's got its own thermal coupon as well so your delta T's will vary a bit.

I would get a statistics book, or google GR&R formulas, or use a statistical software like MiniTAB. If my memory serves, I believe I used traditional statistical formulas like average & standard deviation and the formula for instrument repeatability.

In the end, the WaveRIDER fared a little better, and as others stated, the SPC software was nice. I also liked it better than WaveRIDER because it used the MOLE as its brains, and you could do other stuff like run the Wave Rider whilst profiling a board behind it. At the time, too,I was a MOLE loyalist. The Wave Optimizer, although pretty good when it worked, had some not-so-good features (that ripoff $300 replacement battery pack), and lame software.

The key to success in implementing this is to train the proper personnel to run this, and have a reaction plan in place if one of your measurements falls outside the control limits. In my experience, when I was at a big OEM in Riyadh with 10 lines, the GOOD operators always called me out to the floor during an out-of-control condition. Others didn't bother running it. (we had one on each wave). So, again, half your battle will be getting the proper personnel to run this. At my 1st small company, we used the Optimizer on just one very old 1980's Econopak and it worked great! So, it all depends on how your shop is run.

Okay...enough rambling for one night. Damn, I should move to America and consult for you guys. At least they don't saw your hands off! :)

reply »



SPC for Wave solder (defects on PCB's) | 16 February, 2006

Samir, You are a nut! I wouldn't cut your hands off! I asked because I was wondering about the stability and repeatability of the wave machine itself being used as the standard. I do understand Gage R&R or at least I think I do that was why I asked.

reply »

SMT spare parts - Qinyi Electronics

Electronics Equipment Consignment