SMT in Line / Feature Article: Placement Machine Cpk Verification

Placement Machine Cpk Verification
A new method measures long-term accuracy and reliability.

By Dave Gunster

The challenge in developing a quality acceptance test (QAT) for placement machines is to ensure that the parameters being measured will accurately represent the long-term performance of the machine. Measurements must quantify and verify X-axis, Y-axis and q rotational deviations from the ideal placement. One method for verifying placement accuracy involves a glass slug imprinted with the footprint of a “perfect” high-pin-count QFP that must be placed by the machine (see lead photo). By placing an ideal component, in this case a 140-pin QFP with 0.025” spacing, both camera and spindle-placement accuracy can be consistently measured. In addition to specific machine performance data, estimates of intrinsic availability, throughput and meassures of reliability should be provided based on historical data accumulated over proliminary dry cycle and set-up procedures, including mapping and calibration, the quality acceptance criteria (QAC) process begins.

Eight-stage process

QAC are precise performance parameters that must be met by a placement machine. The first step in the eight-stage QAC process involves an initial 24-hr dry cycle run during which the machine must function continuously with no failures.

The second stage requires the accurate placement of components on two boards containing 32 of the 140-pin glass slug components per board. The master board contains six global fiducials that are referenced by the machine prior to placement and by a vision measuring system to verify accurate component placement. The tatal number of boards that are populated depends on the specific head and camera configuration of the machine being tested. For example, if the machine has two placement heads and two cameras, then eight boards will have to be populated with a total 256 components (35,840 leads). This covers all of the possible head and camera combinations.

The components are placed at all four orientations: 0° , 90° , 180° , and 270° using all four spindles. Following this procedure, each board is scanned by the measuring system and a complete listing of any debiations is provided. Each 140-pin glass slug contains two round fiducials held to ± 0.0001” relative to the lead layout at opposite corners of the component, and used to calculate X, Y and q rotational deviations. All 32 placements are measured by the system and the deviations calculated for each placement. The predetermined specification, which must be held by the machine for each individual placement, is ± 0.003” in the X and Y direction and ± 0.2 in q rotation.

In order to pass the initial “slug run,” each of the 32 “components” placed at the various sites on the board must meet four test criteria: no individual placement can be outside the ± 0.003” or ± 0.2° specifications during the run. In addition, the mean for the X and Y deviations cannot exceed ± 0.0015” or less and their standard deviations must be within 0.0006”, standard deviation for q must be less than or equal to 0.047° and the mean deiation less than ± 0.06° , and the Cpk (process capability index) to be greater than 1.50 in all three quantified areas. This translates into a minimum of 4.5 s or approximately 3.4 defects per million (dpm) allowable at the extreme.

Generally, the performance numbers realized to date exceed a process capability index of 2.0 or approximately two defects per billion (6 s performance). This measurement procedure allows the manufacturer to gauge how well production requirements are being met.

Once accumulated, the individual performance information is utilized to figure the mean and standard deviation for all components placed on the board and to determine Cpk. A final QAC summary should be provided by the measuring system, which lists target locations, deviations from target and calculated lead-to-pad coverage for various pitches in thousands of an inche (Figure 1).

Name: David Gunster							Pattern: STD Dual			Cal#: 3
Ambient temp: 67.8							Board ID: 2			S/W level: Spec
Time: 16:03:02							Comp: 356			Spindles: All
Date: 08/08/95							Pitch: 10			Diag: off
Mach Type: GSM1							Head Type: HFH			Head Loc: Frt
PEC Information
Global Fiducials: 6 point								Specifications
								UIC Tolerance		% Voverage
								X, Y + - 0.0030		75% @
								Theta + - 0.20		LW = Pitch* .5
										PW = LW*1.25
										Span = 1.5
Correction Matrix
	0.999986	0.000592	-1.49735
	-0.00055	0.999952	22.76718
Board #2 Data									UIC P/F	EST. % COV
VW	PL	SP	XNM	YNM	TNM	XDEV	YDEV	TDEV	X Y T MLTE	20	25	30
1	1	1	4	1	0	0	-0.0001	0.003	0.0001	100	100	100
2	5	2	6	1	0	0.0004	-0.0001	0.0357	0.0009	100	100	100
3	9	3	7	1	0	0	0.0002	0.0009	0.0002	100	100	100
4	13	4	9	1	0	0.0003	0.0000	0.0011	0.0003	100	100	100
5	3	1	4	4	180	-0.0002	-0.0003	0.0168	0.0005	100	100	100
6	7	2	6	4	180	0.0004	-0.0007	0.0666	0.0016	97	100	100
7	11	3	7	4	180	0.0001	-0.0002	0.013	0.0004	100	100	100
8	15	4	9	4	180	0.0001	-0.0002	0.0169	0.0004	100	100	100
9	2	1	4	2	90	-0.0003	0.0005	0.0562	0.0012	100	100	100
10	6	2	6	2	90	-0.0002	-0.0002	0.0709	0.0011	100	100	100
11	10	3	7	2	90	0.0001	0.0000	0.001	0.0001	100	100	100
12	14	4	9	2	90	-0.0001	0.0001	0.0061	0.0002	100	100	100
13	4	1	4	6	270	0.0001	-0.0005	0.0316	0.0009	100	100	100
14	8	2	6	6	270	0.0005	-0.0005	0.0521	0.0012	100	100	100
15	12	3	7	6	270	0.0001	-0.0001	-0.0031	0.0001	100	100	100
16	16	4	9	6	270	0.0002	-0.0001	-0.0167	0.0004	100	100	100
17	17	4	14	2	0	-0.0002	0.0001	0.0301	0.0006	100	100	100
18	18	3	1	2	0	-0.0001	0.0001	-0.0195	0.0004	100	100	100
19	19	2	1	4	0	0	-0.0001	0.0432	0.0007	100	100	100
20	20	1	14	4	0	-0.0003	-0.0001	0.014	0.0005	100	100	100
21	21	4	2	1	90	-0.0003	-0.0001	0.0243	0.0006	100	100	100
22	22	3	2	2	90	-0.0003	-0.0003	0.0214	0.0006	100	100	100
23	23	2	2	4	90	-0.0003	-0.0004	0.0637	0.0012	100	100	100
24	24	1	2	6	90	-0.0003	-0.0005	0.0116	0.0007	100	100	100
25	25	4	11	1	180	0.0002	0.0000	-0.0164	0.0004	100	100	100
26	26	3	11	2	180	0	-0.0001	0.0317	0.0005	100	100	100
27	27	2	11	4	180	-0.0002	-0.0003	0.0139	0.0005	100	100	100
28	28	1	11	6	180	-0.0003	-0.0005	0.0283	0.0009	100	100	100
29	29	4	12	1	270	0	0.0000	0.0225	0.0003	100	100	100
30	30	3	12	2	270	-0.0005	0.0000	0.011	0.0006	100	100	100
31	31	2	12	4	270	0.0001	-0.0004	0.0473	0.001	100	100	100
32	32	1	12	6	270	0	-0.0003	0.0207	0.0006	100	100	100
Summary Board #2
Parameter		N				Mean	STD DEV	MIN	MAX	CPK
X		32				-0.00003	0.00025	-0.00048	0.00053	4.00477
Y		32				-0.00016		-0.00071	0.0005	3.79266
Theta		32				0.02188		-0.01949	0.07086	2.49679
Summary run #101
Parameter		N				Mean	STD DEV	MIN	MAX	CPK
X		64				-0.00003	0.00023	-0.00048	0.00053	4.32489
Y		64				-0.00004	0.00026	-0.00071	0.0006	3.84329
Theta		64				0.02191	0.02174	-0.01949	0.07086	2.7301
Total Boards in Run: 2

Figure 1.Final run QAC

Machine Attributes Run

After the machine passes these tests, a 3,000-component machine attributes run is initiated, placing a full range of standard SMDs including: 0603, 0805, and 1206 chips, SOT 23s, eight-pin SOICs, 33-pin PLCCs and 100-I/O QFPs. Essentially a set-up run with no pass/fail criteria, it allows any feeder or other mechanical concerns to be resolved. Errors tracked may include upside-down components, devices on their side or on end, skewed components and any parts with pad coverage below 75 percent.

The next step in the evaluation process involves another nonstop dry-cycle run of 12 hr.

This test is followed by a second glass slug component run requiring on board with 32 glass components, camera and head and follows the same tracking process as in stage 2. (The number of boards to be populated is determined by the total of head and camera combinations.) Once again, the requirements established with the first glass components must be met for the machine to continue the QAC.

Upon completion of the preliminray QAC process, the machine must undergo a final run in which 3,000 components are placed with only one error permitted. If there are on or more placement errors, the root cause must be identified by the team and corrected. During this segment, intrinsic availability of the machine cannot fall below 98 percent.

A final 12-hr dry cycle run in which no failures are allowed is conducted. One last placement series requiring the population of two boards with the 140-pin QFP glass-slug devices, camera and head is performed and all placement locations scanned by the measuring system and data recorded and charted.

Process Capability Index

Board design geometries and desired quality levels determine the capability (accuracy and repeatability) requirements for the assembly equipment in a given application. For example, consider the following case in which a component lead is to be placed on a board pad (Figure 2): Based on the geometries, the specification limits for the placement process (in the X dimension only) are ± {(Wp –Wl)/2+c(where c is expressed in 0.001”)} around the ideal of a perfectly centered lead on the pad. That is, whenever placement of the lead on the pad exceeds the specified limit in either +X(Upper Specification Limit) or –X(Lower Specification Limit) direction, the board is rejected.

Figure 2. Lead-to-pad coverage and deviation permitted.

Once the worst-case specification limits are determined for a board, statistical data provided by the equipment supplier can be put into the following form:

Cpk = minimum of {(USL-m )/3s , (m -LSL)/3s }

Where m and s are the mean and standard deviation, respectively, achieved by the equipment over many of program-driven trails. These data are supplied with a detailed description of the test method by which the data were collected to ensure that machine performance meets capability requirements as specified. The resulting value of the Cpk is proportional to the number of standard deviations that fall between the mean of the distribution and the closest upper and lower specification limits. The Cpk is important because it coneys a specific defect rate for placement accuracy (Fifure 3).

Figure 3. The distribution of deviations from normal for X, Y and Theta are provided as part of the placement machine data.

Support Services

Providing support services to customers is necessary for equipment manufacturers together with the responsibility to keep communication lines open with accurate, understandable information about the machine. In order to meet the need to know, a complete machine performance package should be supplied upon delivery and should include a sheet containing all of the machine specific data collected during the QAC testing process as well as documentation explaining terminology and the types of information collected.

To assist the customer in understanding the QAC process and to provide contacts specific machine, the documentation package also should include the names and signatures of those directly involved in its QAC evaluation.

In addition to this information, a complete disclosure of the operating, interrupt, failure and total repair times plus total number of placements, quantity of placement errors, intrinsic availability, intrinsic throughput and placement performance in terms of ppm levels also fo in the machine performance documentation package together with a full glossary of terms and definitions of all of the statistical relationships that are quoted in the document.

Finally, to provide a better perspective on the long-term process performance of the equipment, as well as the ability of the equipment manufacturer to maintain control of the internal manufacturing process, the historical data for the last 70 machines should also be presented as part of the package. This information allows the calculationof the mean times between failure, between interrupt and between stoppages as an average over an extended total period. Data from a field of 70 machines enable total placement counts to move from the 10,000+ to the 500,000+ placement realm, yielding the associated cumulative benefits that an increased data pool provides. In effect, these provide a long-term tracking process for both machine performance and the internal measurement process of the equipment manufacturer.

Conclusion

The demands associated with the assembly of PCBs are becoming more intense. Driven by the increasing complexity of the process and an expanding range of component types, placement machine performance must become more precise in all aspects. Collecting and supplying machine performance data in an intelligible manner is a critical part of the partnership process. Measurement by itself only answers part of the question; properly interpreting and applying the results completes the understanding.

DAVE GUNSTER is a quality engineering supervisor for Universal Instruments Corp., P.O. Box 825, Binghamton, N.Y. 13902-0825; (607) 779-7522; fax: (607) 772-1878.

Information & Suggestions: webmaster@smtinline.com

SMT in Line
www.smtinline.com

Surface Mount Technology in Electronics Manufacturing

SMT in Line www.smtinline.com

Surface Mount Technology in Electronics Manufacturing

SMT in Line
www.smtinline.com