Printed Circuit Board Assembly & PCB Design Forum

SMT electronics assembly manufacturing forum.

SMTnet
»
Electronics Forum
»
Yield on 0402 SMT Printed Circuit Board Assemblies

Yield on 0402 SMT Printed Circuit Board Assemblies

Subscribe to Thread

I am looking for data on the yield of printed circuit board ... - Jan 02, 2007 by dyoungquist

Personally I wouldn't care what the industry standard is....... - Jan 02, 2007 by Fastek

way too low of yield. with that technology on the board I wo... - Jan 02, 2007 by RDR

While the 85% yield on a per board basis seems low, we are a... - Jan 04, 2007 by dyoungquist

99% good solder joints = 10000 ppm (defects per million oppo... - Jan 04, 2007 by Steve Thomas

We use DPMO when talking to customer and yield when trying t... - Jan 04, 2007 by RDR

Have you thought about an AOI system? With the system we hav... - Jan 04, 2007 by

Our first pass yield across all our products is above 95% bu... - Jan 05, 2007 by

Craig- Your point assumes most if not all of the problem ... - Jan 05, 2007 by Fastek

as someone had said before, check you stencil and have it re... - Jan 08, 2007 by snaggletooth

I've been assembling thousands of boards for dozens of custo... - Jan 10, 2007 by

dyoungquist

#46437

Yield on 0402 SMT Printed Circuit Board Assemblies | 2 January, 2007

I am looking for data on the yield of printed circuit board assemblies so that I can judge if our SMT line is performing up to typical industry standards. Following I will describe the details of the assembly and what our yield has been.

Each board has 82 resistors and capacitors ranging in size from 0402 to 0805, most being 0402. There are 14 ICs of various package sizes including 8 and 14 pin TSSOP packages. Parts are loaded on both sides. The boards come in a panel that has 63 boards on it. The steps in our process are apply lead free paste to top side of panel with screen printer, load panel with automated placement machine and then run panel through hot air reflow oven. Next repeat the steps for the bottom side of the panel. After the panel is loaded, the individual boards are broke out of the panel and pins are hand soldered to the boards. Then each board is tested for functionality.

Our yield on a per board basis is about 85% meaning that 85% of the boards pass the functionality test and require no rework. My question is this: is 85% a good yield for this type of printed circuit board assembly?

reply »

Fastek

#46438

Yield on 0402 SMT Printed Circuit Board Assemblies | 2 January, 2007

Personally I wouldn't care what the industry standard is....if I was producing something where 15 out of 100 failed, it would not be acceptable.

Determine why this is happening and fix the problem and don't be concerned with what anyone/everyone else might be doing.

I would bet you have 1 or 2 issues that upon resolving will bring your yield up to 98%.

reply »

RDR

#46439

Yield on 0402 SMT Printed Circuit Board Assemblies | 2 January, 2007

way too low of yield. with that technology on the board I would expect about 98% as well.

So what are the defects? and more importantly, is your design, machine(s) and/or process(s) capable of this technology?

Details I am sure are forthcoming

Russ

reply »

dyoungquist

#46496

Yield on 0402 SMT Printed Circuit Board Assemblies | 4 January, 2007

While the 85% yield on a per board basis seems low, we are actually running over 99% yield on a per solder joint basis as there are about 17,700 joints per panel. On average, 9 out of 63 boards in the panel need rework. Rework issues include missing 0402 resistor, lack of solder from poor paste job, solder bridge on IC legs and the occasional shifted part so that one of the legs is not making contact with the pad. Rework time averages 10 min per board.

We are using a Mydata machine for the automated placement that is capable of loading the fine pitch ICs and 0402 parts. However, no machine is perfect and this is the source of the occasional missing or shifted parts. I believe that solder paste application is the biggest source for our low yield and we are in the process of dramatically upgrading our solder paste application equipment.

What this really comes down to is what metric you use to measure yield, a per board basis or a per joint basis. We could have 63 bad joints on the panel which is a 99.6% yield on a per joint basis. However, if it turns out to be 1 joint per board the yield on the panel would be 0% on a per board basis. Any thoughts on which metric we should use to judge the performance of our SMT line?

reply »

Steve Thomas

#46500

Yield on 0402 SMT Printed Circuit Board Assemblies | 4 January, 2007

99% good solder joints = 10000 ppm (defects per million opportunities).

Having only built with 0603s I can't speak for a target but that still seems pretty harsh....with a 75/25 0805/0603 ratio at my last place of employment I think we were running at about 2000-3000 ppm and almost all of our problems were with 0603 caps that didn't work with our minimelf capable nozzles. With good placement you should be able to achieve that as well.

If your problems are mostly print based maybe a stencil design review is in order to go along with your new equipment? I know I had to use a different set of rules for 0603 aps than I did for 0805s.

Edited to add that ppm can be calculated many different ways as far as opportunities go but if you'r looking at connections only it should be pretty cut and dried. You might also find that as your printing gets better you see fewer misalignments because you get better bricks and paste/part contact.

reply »

RDR

#46501

Yield on 0402 SMT Printed Circuit Board Assemblies | 4 January, 2007

We use DPMO when talking to customer and yield when trying to figure out how to make more money in the shop.

Does that make sense?

Russ

reply »

Ben_T

#46503

Yield on 0402 SMT Printed Circuit Board Assemblies | 4 January, 2007

Have you thought about an AOI system? With the system we have where I work, I'm able to run queries that'll tell me DPMO, yield, and pass/fail rates on a board level, part level, and opportunity level. Furthermore, this data can be broken down to see the yields for a particular assembly, or a particular part, or even a package. Using this data, we have been able to see EXACTLY where the problems are coming from. The machine we are using is a YesTech....YTV-2050? I think. Either way, it's got 1 TOP camera, and 4 side views. I'm not gonna sit here and sell it like the salesmen do, but it has definitely improved our yield. I have heard that the Mirtec AOI machines are quite an improvement over the YesTech series, but you might want to do more research.

reply »

Cmiller

#46511

Yield on 0402 SMT Printed Circuit Board Assemblies | 5 January, 2007

Our first pass yield across all our products is above 95% but we include anything that does not meet IPC standards. Probably 75% of our failures are not electrical. AOI is a great tool to get a consistent measure of your process. I dont know how anyone could live without it. We have two Yestech machines, I have heard good things about Mirtec but It would be only a matter of opinion that they are better than Yestech. Looking at first pass yield is statistically inferior to PPM data but it is easier to calculate. The problem is if your boards get twice as complex and your yield goes down 1% you quality improved but your boss will think it got worse. Try to look at your PPM data then take the biggest problems and work on those first. If you cant break down your data into the actual defects then you may want to work on your quality system before you attack your manufacturing procces. Two years ago we had a lot of problems with bridges, they probably accounted for 60% of our defects. We have replaced our screen printers, our placement machines, our stencil frame system, our solder paste and the way we design stencils. We did these one at a time so we could measure the results of each step. Obviously everyone can not take all these measures but The level of improvement was dramatic. You mentioned upgrading your screen printers, once thats done and you can take them out of the equation, take a good look at stencil design and paste. We use QTS stencils and are pretty happy with them, since you only buy the foil, it makes it a lot more cost effective to experiment with stencil designs. I would not reccomend the brands that only stretch the stencil from two sides as we found those systems to be a part of our problem. I would not get too concerned with your numbers compared with other companies, just keep looking at the data and keep working on the biggest issues. Since nobody has your exact setup, there is no fair comparison. If you have marginal placement equipment, missing parts and misplacements would be somewhat common. If you have high end equipment, I would not expect to see them very often at all.

reply »

Fastek

#46519

Yield on 0402 SMT Printed Circuit Board Assemblies | 5 January, 2007

Craig-

Your point assumes most if not all of the problem lies in inferior equipment. I understand your point but I believe that an excellent process can overcome the fact that you don't have brand new and pristine equipment.

As an example you can look at any company that runs three shifts. Productivity will be low on the day shift but the quality will be excellent. Swing shift will have the highest productivity and be able to maintain quality. Grave yard.....no comment.

For me...given capable equipment...it's all about the people you hire and train more than having brand new equipment.

Screen printing is the easiest example of this. As far as these machines have come over the years, they are still operator dependent when it comes to changeovers, set up and consumable replenishment. A bad screen printer operator can really hose a process if he's not on the ball.

Fine tune the process....get the best people you can. If you still have issues, then look at the equipment.

reply »

snaggletooth

#46532

Yield on 0402 SMT Printed Circuit Board Assemblies | 8 January, 2007

as someone had said before, check you stencil and have it redesigned. I design our stencils for my factory and have inproved quality in BGA , QFP and 0603 case sizes. we did have 1 prototype series that used 0403 chips btu we had brand new feeders and all resistors and caps were in paper tape reels, which i believe absorbs so much vibration from the feeder reload that the components keep there correct place much better. so of course you nozzles will be able to locate and pick the component. There are a many possiblilties why you have high failure rate. Much to do with you paste printing and you SMD line. Check you have no 0403 case sized component with plastic tape sometimes caps sometimes have this. Service you feeders use alcohol on all oiled places and use air gun to remove any excess dirt then re-oil them carefully. then check accuracy of nozzle pick place. and also get a new stencil with larger openings and thicker stencil and smaller openings for IC�s. but change it ASAP. AOI will not help or fix problems just spot them for you you cant polish a turd

reply »

SSS

#46586

Yield on 0402 SMT Printed Circuit Board Assemblies | 10 January, 2007

I've been assembling thousands of boards for dozens of customers over the past 5 years with 0402 chip components. 85% can happen very easily even with newer equipment. Defect challanges with 0402 packages is almost always tombstoning. Capacitors about 3x the challange of resistors due to there thicker geometry which aggrevates the tombstoning dynamic.

Here is what you need to know: 1. Proper pad layout is the first step. Center to center needs to be close to nominal component length so that the pads don't compete for the part during reflow. Smaller pads are better.

2. Pad design, aperture disign, paste print accuracy and placement accuracy tolerance stack up needs to allow part to paste interference every time.

3. Less paste, less tombstoning generally speaking.

4. Thermal properties of the two pads, ideally would be similar. If one pad is directly and themally tied to a ground layer it will heat slower, reflow later and not get the part. The hotter pad will melt first and surface tension will lift the part off the colder pad. Direct the colder pad(s) into the inline oven first if possible.

This message was posted via the Electronics Forum @