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

Printed Circuit Board Assembly & PCB Design Forum

SMT electronics assembly manufacturing forum.


Process control, SMT inspection and rework

Alan W.

#19438

Process control, SMT inspection and rework | 9 April, 2002

I have recently signed onto a company with very little board assembly experience. I am trying to develop a process control and continuous process improvement program for a new surface mount line, but have run into several people who are not really interested in pursuing one or both or have significantly different ideas on them. The main problem is when, how, and who collects defect data. The company is very small and with limited resources (we can�t afford automated inspection equipment). I am advocating starting with a program of Mass Inspection to collect defect data for use in identifying our process problems. We would use this data to chip away at our process issues, then when we reach a defect level we are comfortable with, to implement a sampling method for inspection. My intention was to place this inspection step immediately after each Reflow step. The defects would be identified and documented, then reworked. This inspection would be performed by an experienced SMT inspector, and the rework by an experienced rework technician. My associates are advocating a mass inspection process also, but only after both sides of the board are reflowed, and the inspection, documentation and rework are to be performed by the rework techs. I argue against this because I believe the rework techs are not sufficiently skilled in identifying, categorizing, and documenting defects. You have to have a specific mindset to be a good inspector. Just because you know how to identify a defect does not mean you can stare at boards all day to find them. I know I would go nuts if I was asked to do it. I also think a rework tech would over rework and under report, causing the data to be bogus. Another argument I am getting against my recommendation is that it increases board handling, and the inspection after a first side is useless as it will need to be done again due to introduced defects during the 2nd side process. My experience says if you are creating that many defects on a first side, during a second side process, you have a problem with your second side process. Finally, I say touch-up (I prefer calling it rework) is not inherent in the SMT process and should never be an accepted part of a process, and through proper process development and controls, it can be reduced and even eliminated. The other side says it inherent and acceptable, and should be considered a standard manufacturing step.

Your Thoughts?

This message was posted Add this forum to your site! Click to learn more. the Electronics Forum @

reply »

Alan W.

#19439

Process control, SMT inspection and rework | 9 April, 2002

I have recently signed onto a company with very little board assembly experience. I am trying to develop a process control and continuous process improvement program for a new surface mount line, but have run into several people who are not really interested in pursuing one or both or have significantly different ideas on them. The main problem is when, how, and who collects defect data. The company is very small and with limited resources (we can�t afford automated inspection equipment). I am advocating starting with a program of Mass Inspection to collect defect data for use in identifying our process problems. We would use this data to chip away at our process issues, then when we reach a defect level we are comfortable with, to implement a sampling method for inspection. My intention was to place this inspection step immediately after each Reflow step. The defects would be identified and documented, then reworked. This inspection would be performed by an experienced SMT inspector, and the rework by an experienced rework technician. My associates are advocating a mass inspection process also, but only after both sides of the board are reflowed, and the inspection, documentation and rework are to be performed by the rework techs. I argue against this because I believe the rework techs are not sufficiently skilled in identifying, categorizing, and documenting defects. You have to have a specific mindset to be a good inspector. Just because you know how to identify a defect does not mean you can stare at boards all day to find them. I know I would go nuts if I was asked to do it. I also think a rework tech would over rework and under report, causing the data to be bogus. Another argument I am getting against my recommendation is that it increases board handling, and the inspection after a first side is useless as it will need to be done again due to introduced defects during the 2nd side process. My experience says if you are creating that many defects on a first side, during a second side process, you have a problem with your second side process. Finally, I say touch-up (I prefer calling it rework) is not inherent in the SMT process and should never be an accepted part of a process, and through proper process development and controls, it can be reduced and even eliminated. The other side says it inherent and acceptable, and should be considered a standard manufacturing step.

Your Thoughts?

This message was posted Add this forum to your site! Click to learn more. the Electronics Forum @

reply »

Alan W.

#19440

Process control, SMT inspection and rework | 9 April, 2002

I have recently signed onto a company with very little board assembly experience. I am trying to develop a process control and continuous process improvement program for a new surface mount line, but have run into several people who are not really interested in pursuing one or both or have significantly different ideas on them. The main problem is when, how, and who collects defect data. The company is very small and with limited resources (we can�t afford automated inspection equipment). I am advocating starting with a program of Mass Inspection to collect defect data for use in identifying our process problems. We would use this data to chip away at our process issues, then when we reach a defect level we are comfortable with, to implement a sampling method for inspection. My intention was to place this inspection step immediately after each Reflow step. The defects would be identified and documented, then reworked. This inspection would be performed by an experienced SMT inspector, and the rework by an experienced rework technician. My associates are advocating a mass inspection process also, but only after both sides of the board are reflowed, and the inspection, documentation and rework are to be performed by the rework techs. I argue against this because I believe the rework techs are not sufficiently skilled in identifying, categorizing, and documenting defects. You have to have a specific mindset to be a good inspector. Just because you know how to identify a defect does not mean you can stare at boards all day to find them. I know I would go nuts if I was asked to do it. I also think a rework tech would over rework and under report, causing the data to be bogus. Another argument I am getting against my recommendation is that it increases board handling, and the inspection after a first side is useless as it will need to be done again due to introduced defects during the 2nd side process. My experience says if you are creating that many defects on a first side, during a second side process, you have a problem with your second side process. Finally, I say touch-up (I prefer calling it rework) is not inherent in the SMT process and should never be an accepted part of a process, and through proper process development and controls, it can be reduced and even eliminated. The other side says it inherent and acceptable, and should be considered a standard manufacturing step.

Your Thoughts?

This message was posted Add this forum to your site! Click to learn more. the Electronics Forum @

reply »

#19443

Process control, SMT inspection and rework | 9 April, 2002

Alan, You're right, mass inspection after assembly won't work. The inspectors will miss 25% (or more) of the defects, and the boards will be reworked before SMT operators can properly define defects and their sources. The company I work for tends to view inspction as as a work center function, with each group carrying the responsibility for their production. As a result, SMT has its' own trained inspectors, and so on. In time, we have settled on using AOI for lower-volume assembly and ICT for higher volume. Final visual is only used to inspect assemblies that cannot be cleared by these tests, i.e. brackets, terminal blocks, jumpers, etc. The key to this is providing process instructions for the inpectors to narrowly focus the items to be inspected so as not to re-inspect areas of the assembly that have already been cleared. Just a side note, one of the most expensive downsides of the 'mass inspection' is when the inspector over-specifies the inspection (inspects to a higher level than your customer will pay for) and causes expensive 'rework' that eats into profits but benefits no one. Rework is of the devil and acceptance of it is devil-worship! Best of luck, Clark K.

reply »

ianchan

#19445

Process control, SMT inspection and rework | 9 April, 2002

We are a local enterprise with limited resources, just like you described, so hope this experience sharing helps :

we place a production Visual Mechanical Inspection (VMI) inspector at the end of each Reflow Oven area, complete with magnify scope 30x and a cutie spacious table, with instructions to nothing but inspect each and every pcs of PCBA output from the Reflow Oven. We do this 100% in-process(aka.= IPQC) inspection for both 1st and 2nd SMT processes' output pcs.

All the IPQC data is entered into a excel spreadsheet, by SMT supervisor on daily basis (yesterday output, today we data entry into excel), as this keeps the SMT guys accountable for daily monitor of quality performance and blowing the whistle to QA, Process, Technicians for help instead of running like nobody's business and then asking for conditional quality waive just before shipment.

The data is analysis by QA Engineer, who will ask the Technicians and Process Engineer for immediate action on critical defects (eg. crack components). This portion is tricky coz different companies define crtical defects differently. I suggest anything that is potential "market claims" or "difficult/intermedient to ICT/FT detect", becomes a critical defect. Finetune your inspection and testing requirements along customer contractual requirements. If the customer is doing the testing and all you guys is build the PCBA boards, then just stick to mechanical visual inspection processes.

More acceptable "inherently design/machine" process defects (eg. insufficient solder, missing components) will have engineering root cause study, proposed remedy action to either customer(design/materials fault) and internal management(machines/manpower/method fault). give your management and customers a solution together with the fault analysis, else you will get F' big time...

Inspection data must be collected (initial stage at the very least) to know what defects are happening on each (1st/2nd SMT) sides of the PCBA, and at what severity rate?

Only after knowing what defect location, types of defect, and frequency, y'all looking at can you form an effective "counterstrike" program against output defects. (that program however is another story).

On a personal level, agree that rework is not supposed to be inherent for IDEAL cases. unfortunately we do not live in IDEAL world(s). Therefore effort must be put into studying and getting appropriate parties to do their respective jobs well to minimize the screwups to SMT processes. (follow the fish-bone diagram basis of machine, manpower, method, materials, environment).

Its a tough job, yeah I know, thats why you were hired to be the smuck... take pride in being that "great fella who's gonna make it all right"... forget about knocking heads with the opposition party, either :

1) assign them out of the team (if you have that authority) and stick with your cheerleader (similar-mindset) folks,

2) else you can force feed them to accept and just do it(again, if you have the authority),

3) else you can ignore them and go ahead without them full steam (if you have no say and authority) eventually you can show em' you did it with your supporters (lonely though),

4) else you will have to bribe, lie, cheat, smooch your way into their hearts to accept and do it your way.

Been there, done it all, and heck proud of what we achieved as a TEAM. What a interesting job we do, eh?

reply »

#19448

Process control, SMT inspection and rework | 10 April, 2002

Ask your compadres how you can control a process when it's output has already been completed? If you do second side assembly before you inspect the first side, you can't possibly apply real time corrective action. I know what I'm talking about in this sense (which is a rarity) because that's what we do here, and lobbying against it so far has been futile. Well, everyone agrees that real time inspection (and data collection and tracking) is the best way to go, but no one is willing to commit the resources to make the change (man-power requirement assessment, a new layout, separate inspection and touch-up stations, etc.)

You on the other hand have the opportunity to set your process up the right way from the get-go.

One of the most frustrating characteristics of this inspect/touchup process is that, with two shifts operating, one shift is always inspecting and touching up the previous shift's output. Not only is it not real time feedback, but there's never anyone around to answer the most important question "What 'n tarnation happened to cause THIS???"

reply »

#19491

Process control, SMT inspection and rework | 12 April, 2002

SPC for each assembly(I use C(for board defects)NP(for unit defects) at the end of reflow to determine process capabilities determine Control limits. Once control limits are set, inspectors stop the process until out of control limit is addressed and signed of(by operator or tech). Once your process is under control for any given assembly you progress to random sampling... gluck

reply »

#19511

Process control, SMT inspection and rework | 15 April, 2002

Use the same size samples all the time, eh?

reply »

ianchan

#19519

Process control, SMT inspection and rework | 15 April, 2002

Hi Clark,

Out of curiousity and in the name of learning new stuff, can you share with us how does your team, calculate the cost of inspection? Is inspection an additional chargable item in the quotation sheet?

To quote your posting : "Just a side note, one of the most expensive downsides of the 'mass inspection' is when the inspector over-specifies the inspection (inspects to a higher level than your customer will pay for) and causes expensive 'rework' that eats into profits but benefits no one."

Am very interested to learn how to define and establish the quality level (in terms of $ and cents) that the customer pays? Any formulae or calculation basis to determine this?

For our side quotations, we only have a fixed costing based on processes (machine/manual) performed to produce the product. Dun think inspection cost (apart from ICT and X-RAY) is factored in? I believe majority of customers these days take it that QC inspection is a inherent factor into any process run for their product (therefore should not be a seperate chargeable cost?).

What do the other folks, in the shoes of a customer, think? Kindly share!

reply »

vacuum tweezers - manual pcb assembly