Volume 2, Number 2, April 2002
A major difference between the high-mix environment and a volume assembly environment is the approach to process capability. A volume assembly does not usually move between process lines, allowing the statistical data collected for each run to be tied to a unique assembly line. “Tweaks” made to the process can be directly correlated to the results measured on the end product. In the high-mix environment, the product may move between process lines on successive runs or may be built only once. In either case, there is no direct correlation between a defect noted on the end assembly and the necessary corrections to the process.
To effectively link the process corrections
to the end product, the high-mix assembler must assess the process capability
for each assembly line and then match the requirements of a given job to the
appropriate process line. Further, assemblers must develop a generic process
profile broad enough to encompass the majority of the products they assemble.
Even a beginning Algebra student knows that you must have as many equations
as variables to achieve a unique solution. The manufacturing corollary to this
math fact is that if you wish to derive meaningful data from a process test,
you must vary only one factor at a time. There are no shortcuts.
For a basic SMT process, there are
three metrics that should be quantified:
There is a common misconception that
absolute accuracy of placement is not critical since the parts will move once
the solder paste is molten. There is some logic in this assumption: once the
solder paste is molten, the surface tension of the solder pulls on the part
until (if possible) the sum of these forces is zero.
With a perfectly symmetrical pattern
and equal “wetting” of the pads and the SMT part, it is possible for
the part to center itself on the pattern.
However, if we consider the conditions that must be met, it becomes apparent
that we cannot depend upon this feat of magic. A simple mismatch in the volume
of solder paste applied to the pads creates unbalanced forces, but the real
variable is the mass of the SMT part in relation to the volume (mass) of solder.
For a 1206 chip, the ratio is nearly 1:1, but the mass for QFPs and fine-pitch
QFPs overwhelms the solder paste mass. You must accurately place the part for
precision positioning.
|
||||
Several commercial systems are available for evaluating placement accuracy. Most use glass plates with graduations that can be read with a 10 or 20 power loop. The plate is sprayed with an adhesive. Then the placement system is directed to position a part in a mathematical location with a programmed rotation.
This process is repeated 10 times and the results are plotted. The resulting chart will give an X error, a Y error, and a Theta error, which is calculated from X and Y measurements. For fine pitch work, the placement system must be able to put at least 75 percent of the lead on the smallest pattern to be placed in production. This means that to place a 20 mil pitch QFP, the maximum placement error is .0021 inches in either axis.
|
||||
To complicate this problem, the error is not usually the same for the two axes because many systems use a lead screw driven Y-axis and a belt driven X-axis. The differing servo systems, drive mechanisms, and encoding schemes result in a mismatch. Lastly, remember the Pythagorean theorem: A one mil error in X and a one mil error in Y will result in a 1.4 mil error on the diagonal. So, even if the head rotation is perfect, a part placed at an angle magnifies the axis error.
There is no universal solder paste that can be applied in all situations. Medical companies typically require the use of an aqueous clean paste while automotive companies lean towards no-clean pastes. Further, the requirements of a high-mix environment are different from those of a volume production line.
The following evaluation process was designed to select an aqueous solder paste for a high-mix application. A test pattern on a 5 inch by 7 inch board was designed to incorporate the patterns shown in the table below.
| Package |
No.
of Placements
|
No.
of Rotations
|
| TO-252 |
4
|
2
@ 90°
|
| SOT-25 |
10
|
1
|
| SOT-23 |
48
|
2
@ 90°
|
| SO-16 |
2
|
1
|
| 304 QFP |
1
|
1
@ 45°
|
| 168 QFP |
1
|
1
@ 45°
|
| 120 QFP |
1
|
1
@ 45°
|
| 84 PLCC |
1
|
1
|
| 169 BGA |
1
|
1
|
| 46 µBGA |
1
|
1
|
| 1206 |
100
|
2
@ 90°
|
| 0805 |
100
|
2
@ 90°
|
| 0603 |
100
|
2
@ 90°
|
| 0402 |
100
|
2
@ 90°
|
The evaluation criteria was divided into three major areas. Within each area, specific inspection criteria was established. Either a 1 to 10 "rating" with 10 equal to the highest) or an "amount" equal to the number of occurrences was given.
The IPC-A-610 acceptance criteria should be used as the baseline, and all boards should be inspected and ranked by the same IPC-610 certified inspector. To make the test statistically valid, the test should be run three times with a one-hour wait between runs. Use eight boards in each run (for a total of 24 boards) and process the boards at five-minute intervals. The stencil should not be cleaned between or during the three runs, and handlers should avoid "kneading" the paste while on the stencil. The temperature and humidity must be noted and held constant. The recommended temperature is 70 °F and the relative humidity is 50%.
|
|
|
| Printing Process | |
| Print definition | Rating |
| Aperture release | Rating |
| Wet bridging | Amount |
| "Roll" of paste | Rating |
|
|
|
|
|
|
| Placement Process | |
| Component loss | Rating |
| Component "X" movement | Yes/No |
| Component "Y" movement | Yes/No |
| Component "theta" movement | Yes/No |
| Upside down component loss | Amount |
|
|
|
|
|
|
| Reflow Process | |
| Solder balls | Amount* |
| Solder bridges | Amount |
| Insufficient solder | Amount |
| Flux residue | Rating |
| Voiding on BGA | Amount |
| Solder joint wetting | Rating |
| Appearance of solder joint (shiny) | Rating |
|
|
|
* Use 30x magnification
For component movement, it was decided that “noticeable movement” would be noted but not measured during the paste portion of the evaluation.
Paste and hold for 30 minutes.
Populate and hold for 30 minutes.
Flip the first board for 20 minutes.
Reflow all boards.
Vacuum seal the first
board for solder ball inspection.
Paste and hold for 30 minutes.
Populate and hold for 30 minutes.
Flip the ninth board for
20 minutes.
Reflow all boards.
Vacuum seal the ninth board
for solder ball inspection.
Paste and hold for 30 minutes.
Populate and hold for 30 minutes.
Flip the seventeenth board for 20 minutes.
Reflow all boards.
Vacuum seal the seventeenth board for solder ball inspection.
Upon the completion of this test,
19 pieces of inspection data will have been collected on each of the 24 test
boards. If this data is entered into a spreadsheet, an easy evaluation of prospective
pastes can be performed.
Too often, the role of the adhesive
in the proper attachment of bottom side SMT parts is overlooked. If appropriately
designed, the same test pattern process is used for solder paste evaluation
can employed to evaluate prospective adhesives. This tool can also help to establish
the process limits for bottom side SMT.
The purpose of the adhesive is to retain the part in the correct orientation on the bottom side of the board until wave solder. Often, an intermediate assembly operation inserts through-hole components that will be soldered during the wave operation as well. In this event, the adhesive must not only maintain the part in proper orientation, it must also be resistant to handling as the board is moved through insertion operations.
Future articles will further explore how to establish the process limits for bottom side SMT using glue dot, placement, cure, and wave solder. We will also explore the additional difficulties that occur when adhesive glue dots are used with an aqueous clean wave solder flux.
 |
 |
 |
 |
 |
| http://www.divsys.com | ||||
| http://www.rapidproto.com | ||||
| Voice:
(317) 299-9547 Fax: (317) 298-2055 |
||||
| sales@divsys.com |
