STATISTICAL PROCESS CONTROL FOR · PDF fileCuvinte cheie: lipire, controlul statistic al...

MODELAREA ŞI PROIECTAREA FILTRELOR DE REŢEA

Buletinul AGIR nr. 2-3/2010 ● aprilie-septembrie 103

STATISTICAL PROCESS CONTROL FOR SOLDERING

Conf. dr. ing. Ion GROZAV Universitatea “Politehnica” din Timişoara

Este absolvent al Facultăţii de Mecanică, promoţia 1969 a Institutul Politehnic „Traian Vuia” din Timişoara,

actualmente Universitatea “Politehnica” din Timişoara. A urmat cursuri de specializare privind metode de îmbunătăţire a calităţii, organizate în ţară (Timişoara şi Cluj-Napoca), la Universitatea din Knoxwille – Tenneessee, S.U.A. şi la Universitatea Tehnică din Magdeburg – R.F.G. În prezent este conferenţiar la

catedra de Tehnologia construcţiilor de maşini, a Facultăţii de Mecanică din Timişoara, unde predă cursuri la disciplinele Dispozitive, Utilaje de asamblare şi ambalare şi Metode de îmbunătăţire a calităţii produselor sau

proceselor de fabricaţie. Este consultantul unor firme din oraşul Timişoara în probleme de calitate a produselor sau a proceselor de fabricaţie.

Ing. T. NICOLI Grupului Şcolar Industrial „Electrotimiş“ A absolvit Facultatea de Mecanică din Timişoara, secţia T.C.M, specializarea de Mecatronică. Face parte din corpul cadrelor didactice de la Grupul Şcolar Industrial „Electrotimiş“. Are preocupări privind calitatea proceselor de fabricaţie .

Ing. P. MILOSTEAN

Este absolvent al Facultăţii de Mecanică din Timişoara, precum şi a cursurilor de master, specializarea Inginerie integrată, din cadrul aceleiaşi facultăţi. Are preocupări legate de calitatea produselor şi proceselor de fabricaţie.

REZUMAT. Procesele de fabricaţie se desfăşoară sub acţiunea unor aşa numite cauze interne, care vor provoca dispersia internă a proceselor de fabricaţie. Asupra acestor procese vor acţiona şi diferiţi factori perturbatori, care constituie aşa numitele cauze externe, care generează dispersia accidentală a proceselor de fabricaţie. Acţiunea acestor cauze externe trebuie sesizată, pentru ca ele să fie eliminate, sau în cel mai rău caz limitate. În lucrare se prezintă cazul lipirii unor componente electronice SMD (Surface Mounted Devices)şi a unora convenţionale. Prin utilizarea unor fise de control s-a urmărit eliminarea surselor externe care provoacă defecte. Efectele utilizării acestor metode statistice sunt descrise în lucrare. Cuvinte cheie: lipire, controlul statistic al proceselor, dispersie, SMD ABSTRACT. The manufacturing processes run upon so called internal causes. These are the cause of the internal dispersion of the manufacturing processes. Upon manufacturing processes act also some uncontrolled factors. They are the causes of the undesired accidental variation of the manufacturing processes. The action of the external causes must be perception and they must be eliminated, or at least limited. In the paper is presented the case of soldering of electronically components SMD (Surface Mounted Devices) and some conventional components. By using control charts was eliminated the external causes that generate defects. The effect of these statistical methods is described in the paper. Keywords: soldering, statistical process control, variation, SMD.

1. INTRODUCTION

Statistical methods, for monitoring the manu-facturing processes, offer great possibilities for finding the causes that made the processes to have non-normal behavior. When the processes have a normal behavior, there is only so called internal causes, these causes will generate the natural dispersion of the processes, which usually is smaller then allowed tolerance field [1], [2].

If upon manufacturing processes will act and so called external causes, then will appear accidental dispersions which always are greater then allowed tolerance field. In these conditions the products will not satisfy the quality criterions.

The statistical methods give us the tools for finding the moments when the external causes will act upon the processes. So is the possibility to look for the external causes which have generated the undesired dispersions.

CALITATE ŞI SECURITATE AMBIENTALĂ


Next these external causes will be eliminated or in the worse case them action will be restricted.

2. CASE STUDY

In this case study is taken an electronic board, pre-sented in figure 1. This board has SMD (Surface Mounted Devices) and conventional electronic com-ponents. The SMD components are inserting using SMT (Surface Mounting Technology) machine. The conventional components can be insert using machine or manual.

Fig. 1. Electronic board.

2.1. Describing the manufacturing process

A conceptual manufacturing process for electronic board is presented in figure 2.

SMD inserting and soldering include the phases presented in figure 3 .

2.2. Paste printing of SMD

The boards are manual load in an automatic hooper. From here they are taken on a conveyor to a printing machine. Here on the pads are putting the paste. These are making with the help of some knifes and a perforated screen. The perforated screen has apertures corres-pondent with the pads where will be inserting SMD components. In figure 4,a is presented the perforated screen and in figure 4,b are presented the apertures.

Over the board is putting the perforated screen and over it is putting a layer of paste with the help of knifes. In figure 5 is presented the printing head where can be seen knifes and perforated screen [3].

In the printing process the board’s pads are covert with a thin layer of paste. In figure 6 is presented in detail this printing process.

The high of the paste layer is critical for quality. It has a great correlation with some defects which can appear in soldering process. The high of the paste is inspected in a special place on the working line.

Fig. 2. Conceptual manufacturing process.

Fig. 3. Phases of the SMD inserting and soldering.



a) b)

Fig. 4. Perforated screen with apertures.

Fig. 5. Board printing .

Fig. 6. Paste printing.

2.3. Inserting and soldering of the SMD components

After on the board’s pads are putting the paste layer follow the phase of inserting of electronically com-ponents. These are inserting with the help of inserting machine. The inserting of components is made succes-sively, in correlation with the size of the components.

The SMD inserting line has three types of inserting machine. This are explain by diversity of the SMD com-ponents. Several types of electronically components need several precision classes for insertion.

The electronically components arrived from vendors in rolls or matrix. These are load in automatic feeding devices (AFD). In figure 7 is presented AFD of the inserting machine.

Rolă cu componente

SMD

Maşină de plantat componente SMD

DAA cu rolă

Fig. 7. The automatic feeder device.

The last is the reflow phase. In this phase the boards are passing through a oven. In this oven due to the temperature the paste is molten and so it is soldering the components. The reflow oven is divided in 5 zone [4]. Three zones are for preheating, one for reflow and one for cooling. The zones of the reflow oven are presented in figure 8.

Fig. 8. Schematic view of a reflow oven.



2.4. Inserting and soldering of the conventional electronically components

After the inserting and soldering of the SMD com-ponents, on the boards are putting the cod bar label and will be manually inserting the conventional electronic-ally components. The next step is the wave soldering of these components. These soldering is made by a wave soldering machine. The machine has several zone that are presented in figure 9 [4].

Fig. 9. Wave soldering machine zone.

The boards are putting in some soldering frame. These frames are used to reduce the time for setting when the product is change and the size of the board is different. The frame is load on a conveyor who transports it in the tin bath. The conveyor angle relative to the tin bath must be between 40 and 70. This is setting function of the different type of products. Usually the angle is 70, so is presented in figure 10.

Fig. 10. the conveyor angle.

The fist wave called is call tumultuous wave an it has the function to solder the SMD components (see figure 11). The second wave has function to solder conventional components on the board. All this process is made in nitrogen environment. The absence of the nitrogen create some defects in soldering.

2.5. Board inspection

The boards are wave soldering and are visual inspected. This inspection is made for conventional

components. So is inspected the polarity, solder shape, and is inspected if the components are arise (stone tomb). The acceptance criteria are normalized. The incorrect soldering an the defects will be immediately corrected, if its possible, or the boards will be sent in the reaper shop.

Pompe de antrenare cositor lichid

Valul turbulent Valul lin Baie de cositor

Sens de circulare a cositorului lichid

Fig. 11. Soldering bath.

The boards are also tested from electric point of view. The values are measured and they will be compared with the specifications values. For do this are use a testing platform which allow testing a great number of the components from the board.

3. PROCESS MONITORING USING CONTROL CHARTS

Like improving target has been chose the increase of the passing rate of the testing boards. This measure define how many boards are declare good after testing stage. In the first stage of improving we say that is good a rate of acceptance equal with 94%.

Have been use an Individual Moving Range chart (I-MR) and the process has been monitoring during 10 commands. The control chars has been built using statistical software MINITAB [5], the charts are pre-sented in figure 12.

The chars shows that the process is not in statistical process control. The pattern of the chart point shows that there are some special cause. The mean of the process is 88.12 %, under the establishing level of 94%. Also can be seen a great mean of the dispersion 7.62%. For searching the special cause has been built the Pareto diagram. The result of this analyze is presented in figure 13.



AB238808AB238584AB238575AB238353AB238317AB238209AB237986AB237820AB236990AB236824

110

100

90

80

70

Cod comanda

Ind

ivid

ua

l V

alu

e

_X=88,12

UCL=108,39

LCL=67,85

AB238808AB238584AB238575AB238353AB238317AB238209AB237986AB237820AB236990AB236824

24

18

12

6

0

Cod comanda

Mo

vin

g R

an

ge

__MR=7,62

UCL=24,90

LCL=0

1

I-MR Chart of Rata de trecere

Fig. 12. Initial control charts.

Count 6 6 8572 70 59 14 9 8 7 7Percent 1.7 1.7 24.821.0 20.4 17.2 4.1 2.6 2.3 2.0 2.0Cum % 73.5 75.2 100.021.0 41.4 58.6 62.7 65.3 67.6 69.7 71.7

Page NameOt

her

L14_

IND

D21_

FS8

R173

_LR

D12_

FS5

L13_

IND

L2_IN

D

D16_

FS11

C2-O

THER

S

SHORT

S

Learn

APC

350

300

250

200

150

100

50

0

100

80

60

40

20

0

Coun

t

Perc

ent

Pareto Chart of Page Name

Fig. 13. Pareto diagram for initial case.

It can be seen that 58.6% of the defects are re-presented of three defects, Lean APC, Shorts and C2 –others. The testing team has inspected the platform test , the adaptor and the products. Has been underline the following aspects:

For “Learn APC” defect the boards have in the via pass some mark of the protector mask. This think ovoid the contact of the testing nail and so was generated a false declaration, that the board is not acceptable. This defect was corrected by removing of the protection at wave soldering step, so all via passing have been filled with lead-tin solder.

For “Short” defect has been seen some shorts among the pins of integrated circuits that was planted by SMD method. The cause was the excess of the soldering paste in the components zone that have a distance between terminals of 0.5 mm. To avoid this defect has been command a new printing screen with less thickness and a reduction for components with small steps.

For C2-others” defect the cause was the advance wear of the test nail of testing device. This has been change with a new one.

The capability in this stage can be seen in figure 15.

a) b)

Fig.14. Pasing exemple: a – blind; b – good.

110100908070

LSL UB

LSL 90Target *UB 100Sample Mean 88,121Sample N 10StDev (Within) 6,75729StDev (O v erall) 10,4827

Process Data

C p *C PL -0,09C PU *C pk -0,09

Pp *PPL -0,06PPU *Ppk -0,06C pm *

O v erall C apability

Potential (Within) C apability

PPM < LSL 500000,00PPM > UB 0,00PPM Total 500000,00

O bserv ed PerformancePPM < LSL 609520,72PPM > UB *PPM Total 609520,72

Exp. Within PerformancePPM < LSL 571128,47PPM > UB *PPM Total 571128,47

Exp. O v erall Performance

WithinOverall

Process Capability Rata de trecere 10 comenzi starea initiala

Fig. 15 Initial Capability.

Can bee see that process is not capable. The number of the defects, in the case of the million of opportunities is 571128.47 and the capability index is Cpk = -0.09, that is totally improper. [6].

The previous graphs shows that the process is not in statistic control, is not capable and generates a great number of the defects. This situation must be improved.

After the actions of improvement have been applied the process was again monitored. The control charts in this second stage are presented in figure 16. In figure 16,a are presented the initial control charts, they are compare with the charts from figure 16,b, that are after application of the first trial of the improving actions.

The process capability increase to Cpk =0.47, which is also to low. The process is well centered, but the dispersion is to high, it make the capability to be so small. A new Pareto analyses has been made, the result is presented in figure 18.

The Pareto analyses show that main defects are: The cause of APC defect is vacuum transducer.

This is a false defect and it was removed. SPX23 is a defect due to wear of the nail test, use

for measure this component. Becauze this defect was



meet and in the first stage, has been decided to change all testing nail after 5000 testing, according with pro-duct specifications.

The effect with code 74FCT162245 [D10] has like cause the interface destruction from the test adaptor. This interface has been change.

AB242162AB240820AB240299AB240014AB239636AB239270AB238584AB238353AB238209AB237820AB236824

110

100

90

80

70

Cod comanda

Ind

ivid

ua

l V

alu

e

_X=94,00

UC L=103,03

LC L=84,97

A B236824 A B238808

AB242162AB240820AB240299AB240014AB239636AB239270AB238584AB238353AB238209AB237820AB236824

24

18

12

6

0

Cod comanda

Mo

vin

g R

an

ge

__MR=3,40

UC L=11,09

LC L=0

A B236824 A B238808

I-MR Chart of Rata de trecere by Cod comanda

a) b)

Fig. 16. Control chars in second stage.

100989694929088

LSL UB

LSL 90Target *UB 100Sample Mean 94,1203Sample N 12StDev(Within) 2,89209StDev(Overall) 2,94876

Process Data

Cp *CPL 0,47CPU *Cpk 0,47

Pp *PPL 0,47PPU *Ppk 0,47Cpm *

Overall Capability

Potential (Within) Capability


Observed PerformancePPM < LSL 77124,91PPM > UB *PPM Total 77124,91


Exp. Overall Performance

WithinOverall

Process Capability Rata de trecere

Fig. 17. Capability in second stage.

Count 21 17 9 6 6 5 34Percent 21,4 17,3 9,2 6,1 6,1 5,1 34,7Cum % 21,4 38,8 48,0 54,1 60,2 65,3 100,0

Page NameOthe

rF4

-1F1

16C5

52_P

LCC6

8[D1

5]

74FC

T162

245[

D10]

_1

SP_X

23APC

100

80

60

40

20

0

100

80

60

40

20

0

Coun

t

Perc

ent

Pareto Chart of Page Name

Fig. 18. Pareto diagram for second stage.

After application of these all improving actions, the control charts of the process are presented in figure 19c. In figure 19 are presented all the control charts for them comparing. From the charts, can be observing a great improving for dispersion and for individual values of the boards passing rate. This thing has generated a substantial increasing of the process capability; it is presented in figure 20. The mean of passing rate has the value 19.18% and the process capability increase to Cpk = 2.82, a good value.

3128252219161310741

110

100

90

80

70

Observation

Ind

ivid

ua

l V

alu

e _X=98,18UCL=101,08

LCL=95,27

AB236824 AB238832 AB242756

3128252219161310741

24

18

12

6

0

Observation

Mo

vin

g R

an

ge

__MR=1,09UCL=3,57

LCL=0

AB236824 AB238832 AB242756

1

I-MR Chart of Rata de trecere by Cod comanda

a) b) c)

Fig. 19. Control charts in third stage.

100,599,097,596,094,593,091,590,0

LSL UB

LSL 90Target *UB 100Sample Mean 98,1771Sample N 8StDev(Within) 0,967503StDev(Overall) 1,23078

Process Data

Cp *CPL 2,82CPU *Cpk 2,82

Pp *PPL 2,21PPU *Ppk 2,21Cpm *

Overall Capability

Potential (Within) Capability


Observed PerformancePPM < LSL 0,00PPM > UB *PPM Total 0,00


Exp. Overall Performance

WithinOverall

Process Capability of Rata de trecere

Fig. 20. Capability in third stage.

In these conditions a new attempt to improve the passing rate, over 98%, will increase to much the production costs with small increasing of passing rate (insignificant).



4. CONCLUSIONS

During manufacturing process appear great dis-persions from different external special causes that act upon the process. In order to find the moment when this special causes act, the process should be monitor-ed with control charts. These are statistical tools that help to improve the processes. The charts mark the moment when external special causes act upon the process, but the carts cannot show witch cause are. The working team has the task to find the external special causes and to eliminate them action or in the worst case to limited them actions. For finding the most probable defects are use Pareto diagram and analyses. Pareto analyses can be emphasizing the mean defects of the manufacturing process. Then by using the cause-effect analyses can be find the special

cause that act upon process and cause the mean de-fects. By eliminate external special cause, the process performance can bee more improved, as has been sowing in this paper.

BIBLIOGRAPHY

[1] Thomas Pizdek The Six Sigma Handbook, 2003 The McGraw-Hill Companies, Inc

[2] Vardeman, S.B., Statistics for Engineering Problem Solving, PWS Publishing Company, Boston, 1998.

[3] Martin Tarr, Reflow soldering http://www.ami.ac.uk/courses /topics/0132_rs/index.html#top#top

[4] Martin Tarr, Wave Soldering http://www.ami.ac.uk/courses/ topics/0225_wave/index.html

[5] * * * MINITAB 15, http://www.minitab.com [6] * * *, Engineering Statistics Handbook, http:// www.itl.nist.gov/

div898/handbook/pmc/pmc.htm

STATISTICAL PROCESS CONTROL FOR · PDF fileCuvinte cheie: lipire, controlul statistic al...

Documents

Transcript of STATISTICAL PROCESS CONTROL FOR · PDF fileCuvinte cheie: lipire, controlul statistic al...