Rework unfortunately is a necessary evil. However rework can be minimized in the following ways.

* An optimized print process can eliminate 80% of SMT defects -- this becomes even more important for smaller apertures with sub-0.6 area ratios

• Use an optimized solder paste to maximize paste transfer efficiency (from the stencil to the board) and minimize deviation
• Optimize the print parameters (print speed, separation speed, board support, clamping...)

* If a pin-in-paste process is used and incomplete hole fill is being observed, post-rework can be avoided by using solder preforms in conjunction with solder paste and achieving 100% hole fill

* Also during rework, fluxes that pass SIR unactivated should be used thus eliminating corrosion concerns with unactivated flux. Rework operators typically like to use a lot of flux and given the short heat cycle of a solder iron, the flux may not get activated and therefore pose a corrosion concern.

In my opinion, rework can be acceptable for products which perform non-critical functions, or are part of a non-critical system. Would you want a pacemaker implanted inside yourself, to be one which exceeds all test standards flawlessly or one which was reworked and then "passed" the tests?

Reworked parts for an office printer connote less potential risk than those used in the NASA space shuttle. Clearly cost is also an issue. One is reluctant to discard an assembly which costs thousands of dollars...yet, is a reworked system "the same as" one which was qualified initially, flawlessly.

Does the rework operation run the risk of introducing other flaws, or affect other performance criteria, by the very act of reworking outside of the standard process flow?

Again, in my view, it comes down to risk assessment. As my mother used to say, "New is new...patched isn't new." Risk should be factored explicitly into the practicality calculus.

In case rework is required this also means a deviation from standard SMT assembly technology. Consider for example removing defective components (local heating of PCB), preparing the mounting position (cleaning, paste dispensing or fluxing), manual assembly of the component to be replaced (chance of misplacement), local heating again (selective soldering).

All these processes will disturb the regular SMT process and will be potential future risk areas on the board assembly. So your customers requirement is quite acceptable in case of manufacturing a premium brand OEM product!

To avoid rework I would recommend a pick & place machine based on the parallel placement principle. These machines will give you typical DPM figures below 10 DPM!

Rework technology (equipment, processes, and materials) have come a long way in the last two decades. There are significant data showing fantastic reliability for all types of rework, including high performance devices in aerospace, harsh environments, and mission critical applications.

I believe the only easy answer as to when rework should be avoided is when it is not cost effective (i.e., when the cost of rework exceeds the cost of replacement).

Done properly, a reworked assembly should have nearly identical performance and reliability to a new assembly. That said, I am sure your customer has considered the potential downside to rework.

Performing rework requires exposing the assembly to additional thermal cycles. It is very difficult to predict what adverse affect these cycles will have on the integrity of laminates, solder joints, and active components.

It all comes down to managing the risk, no matter how small.If you are building pacemakers or satellites, your customer may be right. If you are building computers or engine controllers, you may want to present your case.

This is a little unrealistic given the nature of the processes involved and the variables that we work with every day.

The only time it would be possible to completely exclude any degree of rework would be a low cost consumer product where the cost of reworking a pcb is as high or exceeds the profit margin of the product.

It is often held that a reworked part has less integrity than a non reworked part -- an argument with camps on both sides, the reality is that stuff that does not work properly can be repaired and reworked with exceptionally high levels of success and exceed the criteria laid out by industry bodies for quality.

Without knowing the end use application, I would suggest that this is quite unusual and impractical. Reworked boards can be just as reliable as non-reworked boards, assuming that the rework processes are appropriate and well-controlled.

It sounds like your customer has been "burned" by reworked boards in the past, and therefore has a predisposition against any sort of repair operation. However, when processed correctly, repaired boards pose absolutely no risk.

So my suggestion to you would be to investigate the reasons behind this (irrational?) fear on the part of your customer, and demonstrate that your company has manufacturing practices and controls in place to mitigate those potential risks.

I will also add that improperly repaired boards (too much flux, flux running underneath components, not enough heat, etc.) CAN cause reliability issues. The "over-apply & under-heat" issue has real reliability ramifications.

Additionally, the selection of appropriate materials for the operation (cored wire, liquid assist flux) is also important. The use of tacky fluxes (or "gel fluxes") is generally not recommended for most applications but may be appropriate for some components and heating processes (such as a BGA repair machine).

I would suggest to speak to your supplier to make sure that you have selected the proper materials for rework applications.

At Fuji Machine we have customers that build to the "0" Defect standard.

Using great equipment and tools like electrical set-up confirmation, Auto Back-up Support, Intelligent Part Sensing and strict process controls, quality levels of well under 1 DPM are attainable.

The answer to this is depends, they are after all the customer and so it is their choice, you just need to make sure that you can both agree on the cost of not reworking and scraping the boards that they will not accept.

There will always be boards that do not function for one reason or another some components arrive broken and so you will never achieve 100% first time pass unless your process is relatively simple or you have really good incoming inspection and test 100% of all incoming parts. So some level of failure is to be expected

There is a risk associated with rework, when you heat up a local area of PWB to remove, repair or replace a damaged component, damage could occur to surrounding components by heat or by physically disrupting the component during rework. These can be minimized with good training, but a risk will remain.

If there are any potentially delicate components present, then a rework ban can occur. This is normal for some safety critical devices on automotive or aerospace parts where long term reliability is essential and any potential damage caused by rework cannot be accounted for during standard design and test as the rework process puts different stresses on the finished product.

So the answer is that it may not be practical, but if the customers wants it, then you both need to agree on the cost that this will generate.

Well sure, if the assembly does not have any components, or else you have perfect processes and perfect operators. If you do, you could sell tickets for a lot of money, and folks from all over the world would come in droves to see it.

Realistically, it depends on a couple of things. The number of components on the board and the type of parts. Is touch-up allowed, or do they mean only no hot gas rework? Is the board a low-dollar item?

It could be that the cost of the assembly is so low that the customer may not wish to have literally any rework done, but to scrap anything not deemed perfect.

If so, you would need to factor in a higher estimated scrap cost than you normally would in your bid.

Without knowing if there is a particular reason for your customer's position, it is impossible to say if it is practical.

Inexpensive assemblies can often cost more to rework or process as returns than if they are simply scrapped.

If the boards are more costly, there should be some other logic behind the decision. If your customer has a problem with soldering irons in particular as opposed to all rework, that is understandable but could be better defined as to why. Some board materials and components are vulnerable to inappropriate heat cycling, and rework processes are the most likely sources of such problems.

If it were my customer, I would initiate a more detailed discussion to understand the whole situation.