IPC9850_Surface Mount Equipment Characterization.pdf - 第18页
IPC-9850 Official Proposal May 2001 18 4 Attribute Defect Rate and Reliability Performance Metric-- Form IPC-9850-F2 Complete evaluation of an SMT placement system must not only consider placement capability performance …
IPC-9850
Official Proposal
May 2001
17
If the component is placed so that the long sides are along the board’s Y-axis,
Maximum Lead Tip Error (MLTE) = )evsin(
2
),max(
d
SS
ydev
yx
θ
⋅+
Ball-to-Land (BTL) refers to the percentage of the BGA ball or column of an area array components that is placed
on its (round) land. BTL quantifies the error for the “ball” that experiences the most offset (of all the balls) due to
the joint x, y, and θ errors.
BTL= 100
R
A
1
2
1
•
−
π
, where
( ) ( ) ( )
+−−−
∆−
+∆−−
∆−
−
−−−−
2
1
2
2
22
2
1
2
2
1
1
2
1
2
2
1
1
2
1
R
c
sin
2
R
cR
2
c
1sin
2
R
2
R
rc
sin
2
R
rcR
2
rc
1Sin
2
R
2 =A
where:
∆X = |Xdev| + |
2
x
S
- (
2
x
S
· cos (θdev) -
2
y
S
· sin (θdev)) |
∆Y = |Ydev| + |
2
y
S
- (
2
y
S
· cos (θdev) +
2
x
S
· sin (θdev)) |
22
YX= ∆+∆∆r
c =
r
2
RrR
2
B
22
L
∆
−∆+
R
L
= radius of the land
R
B
= radius of the ball
R
1
= min(Radius of Ball, Radius of Pad)
R
2
= max(Radius of Ball, Radius of Pad)
Xdev = Offset of component along the X-axis from the target placement
Ydev = Offset of component along the Y-axis from the target placement
θdev = Rotational offset of component from the target placement
∆X = Maximum offset along the X-axis of the ball from its target placement
∆Y = Maximum offset along the Y-axis of the ball from the target placement
∆r = Radial offset of ball center to pad center. Synonymous with MBE in this document.
Computational note: the values of
Sin
1−
(x) need to be expressed in radians (rather than degrees).
Example of Computing Cpk limits for Termination-to-Land Coverage: After the x,y, and theta errors have been
determined by the CMM (for a four board run using one particular component type), and the individual components’
Termination-to-Land computations have been made as well (using the equations above), the next step is to calculate
the mean and standard deviation of this group of Termination-to-Land values. The Cpk equation should then be
applied, using first the 50% specification limit and then the 75% limit. These 2 Cpk values then go in the
Performance Reporting Form. Suppose the mean of the 144 Termination-to-Land values in a QFP100 run is 85%
and the standard deviation of these values is 5%. Then Cpk versus 50% requirement is 33.2
5
3
5085
=
•
−
.
IPC-9850
Official Proposal
May 2001
18
4 Attribute Defect Rate and Reliability Performance Metric-- Form IPC-9850-F2
Complete evaluation of an SMT placement system must not only consider placement capability performance
covered by Form IPC-9850-F1, but also include information about the yield and reliability of the system. Like the
general performance use of Form IPC-9850-F1, Form IPC-9850-F2 shall be used to present the expected attribute
defect rate and reliability of a specific machine model. However, unlike Form IPC-9850-F1 there shall be no
requirement for a validation for a specific individual machine and there shall be no guarantee for the warranty
period associated with IPC-9850-F2.
The selected reliability metric parameters for this standard fall into three categories -- reliability, availability, and
maintainability. With the exception of Attribute Defect Rate, parameters are based on the SEMI E10-0699E
standard.
The information presented in this form shall be gathered and reported according to this standard, and shall reflect
the performance of the same machine model as in the Placement Performance Metric. The following information is
reported on this form:
1. Manufacturer Name, Machine Model
2. Attribute Defect Rate (ppm)
3. Mean Placements Between Assists (MPBA)
4. Mean Time To Recover from assists (MTTR
a
)
5. Mean Placements Between Failures (MPBF)
6. Mean Time To Repair failures (MTTR
f
)
7. Equipment Dependent Uptime
8. Amount of Preventative Maintenance per 6000 Hours
9. Mispick Rate (ppm)
10. Data Collected From:
a. Number of Factories
b. Number of Machines
c. Total Number of Placements
4.1 Attribute Defect Rate
Although production yield depends on the solder paste printing quality of the PWBs, sticky media is used for this
procedure instead solder paste. It is recognized that the attribute defect level from placement operations on sticky
tape are likely to be lower than defect levels from placement on solder paste or conductive adhesives in production
runs. However use of solder paste introduces many additional variables such as paste viscosity, tackiness and height.
The use of sticky tape provides a method to compare machines with a minimum number of variables. In addition, it
eliminates the need for vendors to have costly screen-printing equipment and expertise in utilizing stencil printing,
board cleaning processes, and controlled disposal of residual solder paste.
Since products assembled by surface mount equipment vary greatly by the type of component mix and size of the
PWB, a special kit is provided. This kit is designed to be representative of the product of a ‘typical user’. This
‘typical user’ kit is made of 4400 components and shall be mounted onto the PWB with sticky media. Table 4-1
provides a list of the 4400 component types and quantities. Because some equipment is not designed to place all
SMT parts, it may not capable of placing this specific set of components. If it is necessary to change the mix of
components, the test shall be run using a similar mix of component types that are within the machine’s capability.
The selected mix of components shall be clearly stated on form IPC-9850-F2. The components shall be presented to
the machine using standard feeders or trays manufactured by a vendor-approved supplier.
IPC-9850
Official Proposal
May 2001
19
Table 4-1 Component Types and Quantities for ‘Typical kit’
Quantity Component
Type
440 SOT-23
440 SOIC-8
880 1608C
880 1608R
880 1005C
880 1005R
The components are to be placed onto an adhesive laminated PWB. The land pads for the PWB shall meet IPC-SM-
782. (Specific board size and layout are not defined by this standard.) After the parts have been placed on the sticky
PWB, a manual visual inspection is made to determine how many placement defects have occurred. While manual
inspection is less than ideal due to its subjectivity, the investment in Automated Optical Inspection (AOI) equipment
for such a limited application cannot be justified.
In this standard, attribute defect rate information differs significantly from the performance metric defined by the
repeatability and accuracy parameters. Whereas the repeatability and accuracy performance parameters are defined
through continuous data -i.e. numerically how much is the component placement error-, the attribute defect rate
parameter is only a function of discrete defects i.e. components which are known to be placed such that a proper
solder joint will not form. Generally these discrete defects occur at such small rates that they are not measured in
percentages of the total number of placements but in defective parts per million (PPM) placements. These discrete
defects are referred to as Attribute Defects, and the anticipated frequency of their appearance is referred to as the
Attribute Defect Rate.
Attribute defects are defined as components placed upside down, tombstone, on side, missing, extra part, damaged
lead(s), damaged part, completely off land, and wrong polarity. An exception is that upside down components are
not to be counted as defects when using bulk feeders. Partially off-land and askew type placement defects are not
included, since the performance level for these types of defects is accounted for by the repeatability and accuracy
parameters. For this parameter evaluation, each component equals one opportunity for a defect and can have a
maximum of only one defect, regardless of the associated anomalies –such as the number of leads, etc. Things other
than the placement machine can cause attribute defects, but they shall be attributed to the machine in the absence of
any other clear cause, such as poor board fiducials, etc.
Estimating the average number of defects that will occur in each million placements is no small task. Estimating
defect rates less than 50 PPM requires very large sample sizes. Different sampling plans exist, each with it's own set
of characteristics and sample sizes.) This standard requires the placement of 88,000 parts. Due to the time and cost
associated with a machine supplier placing this many actual components with each machine manufactured, the
88,000 placements may be spread out over 20 consecutive machines in builds of 4,400 components. As a result, the
confidence level applies only to that group of machines rather than any specific member of that group of machines.
The data may be used as an estimate of the performance of the model type. When a new model machine is launched,
where only a single machine of its kind is available for testing, this single machine shall be utilized for the 88,000
placements.
Calculation Method The PPM level is the total number of observed attribute defects during 88,000 placements of
various component types, divided by 88,000, multiplied by 1,000,000. A minimum total of 88,000 components is to
be placed over 20 tests, where each test run places a component mix of at least 4400 components. The number of
defects is then calculated as a moving average of the last 20 tests.
Attribute Defect Rate = (1,000,000/88,000) *
∑
+
=
20i
itest
i
efectsAttributeD
The following example refers to Table 4-3 where two defects over the 88K parts yields Attribute Defect Rate of 23
ppm. For the first and second machines, the attribute defect rate is obtained by dividing the three attribute defects –
found during run series 2, 7, and 17-- divided by 88K parts, for a 34 PPM. Once the third machine is shipped, the
attribute defect level is reduced to 23 PPM, since the single attribute defect of run series 2 is replaced by the zero
attribute defect of machine 3, run series 22.