Defect Rate Measurement Technique

Dear All,

We are a medium sized Electronic manufacturing company manufacturing computer motherboards and Telecommunication products. We have a big debate going on within our company regarding the defect rate measurement technique.

At present we are calculating yield rates at all the stages i.e paste printing, pre-reflow, Post Reflow, Pre-Wave and Post- Wave Inspection satges. We calculate that as follows.

No. of defective boards x 100 ----------------------- No. of boards inspected

Now, using this method, I am listing below two cases which complicate the situation.

e.g. If out of 100 boards 10 boards fail because of one point touch-up the yield rate is 90%, which is considered to be a pretty bad situation. On the other hand if only 3 boards fail and require touch up at more than 100 points the yield rate would be 97%, which reflects a better situation, though in the later case I am doing much more re-work.

Now, the other alternative is to calculate the defect rate in PPM. i.e.

No. of defective joints x 1000000 ------------------------- Total No. of Joints inspected

Consider a case below, that I am doing 1 point touch-up on 100% of the boards. The PPM value would be very low (Happy Situation) since I am touching up only one point but the yield is 0% (Unacceptable).

Can somebdy help me decide which method should be used to take care of both the requirements. Also, please elaborate on the method used at your end and average Yield/PPMs at Post reflow/Post Wave Inspection stages.

Thanks & Regards Vinesh Gandhi

reply »

| Dear All, | | We are a medium sized Electronic manufacturing company manufacturing computer motherboards and Telecommunication products. We have a big debate going on within our company regarding the defect rate measurement technique. | | At present we are calculating yield rates at all the stages i.e paste printing, pre-reflow, Post Reflow, Pre-Wave and Post- Wave Inspection satges. We calculate that as follows. | | No. of defective boards x 100 | ----------------------- | No. of boards inspected | | | Now, using this method, I am listing below two cases which complicate the situation. | | e.g. If out of 100 boards 10 boards fail because of one point touch-up the yield rate is 90%, which is considered to be a pretty bad situation. On the other hand if only 3 boards fail and require touch up at more than 100 points the yield rate would be 97%, which reflects a better situation, though in the later case I am doing much more re-work. | | Now, the other alternative is to calculate the defect rate in PPM. i.e. | | No. of defective joints x 1000000 | ------------------------- | Total No. of Joints inspected | | | Consider a case below, that I am doing 1 point touch-up on 100% of the boards. The PPM value would be very low (Happy Situation) since I am touching up only one point but the yield is 0% (Unacceptable). | | Can somebdy help me decide which method should be used to take care of both the requirements. Also, please elaborate on the method used at your end and average Yield/PPMs at Post reflow/Post Wave Inspection stages. | | Thanks & Regards | Vinesh Gandhi | Vinesh,

From what your describing it sound's like a terminology trap.. What I mean is what is a defect and what is a defective ?? In our plant a defect is a fault found on a board, you can have a yield of 99% ( 1 fail in 100)i.e. 1 defective board but it could have 20 defect's on it. Yield is normally looked at as the number of defective boards vs the good ones e.g. 5 boards have fault's out of 100 = 95% yeild The anaylysis of defectives lend's it's self to the caluculation of PPM which is how many defect's you have per million... but I think the question is per million what ??? boards / joints / components / days'.... I guess for me the purest measure is PPM in term's of number of joint's so say a board has 200 1206's and a few 20 pin SOIC's your total joint's would be 440, so you'd calculate your PPM using that. On thought thou is that maybe the point's your measuring aint the best ones.. I can understand pre reflow..but I would say generally I'd go with post as a gate, sure pre will give you placement fault's but that should be getting fed back real time anyway's, Pre wave and post wave...why inspect pre wave when you've just inspected it from SMT anyway ??? the only thing you should be finding at a pre wave point would be missing component'd which is going to highlight a handling issue of some sort, which is important but I would't think it can justify it's self as a value add operation (and before I get jumped on I know all inspection is non value add..but some can be justified more than others) Post wave yep for sure that has to be a gate..but you can use it for SMT bottomside quality as well! What about ATE..it's the best quality gate you can have if your test coverage is nice and high (I'm assuming you are running ATE)at the end of the day ATE is really a process verification tool rather than testing the boards..that's what functional does.. so the gate's I would choose for my PPM and yeild's would be SMD - Tops Post Wave ATE Final QC

as for what our yeilds are...now that would be telling..but it depend's on the complexity of the product..double sided / singlesided yada yada yada....

JohnW

reply »