Presentation is loading. Please wait.

Presentation is loading. Please wait.

1 ALMA 2003 Winter Symposium World Cone Symposium World Cone Symposium A Simple Approach to Improve Sensitivity of Quality Control Procedures of Loudspeakers.

Similar presentations


Presentation on theme: "1 ALMA 2003 Winter Symposium World Cone Symposium World Cone Symposium A Simple Approach to Improve Sensitivity of Quality Control Procedures of Loudspeakers."— Presentation transcript:

1 1 ALMA 2003 Winter Symposium World Cone Symposium World Cone Symposium A Simple Approach to Improve Sensitivity of Quality Control Procedures of Loudspeakers Presented by Mauro Bigi Maurizio Jacchia Audiomatica Srl - Italy Imperial Palace, Las Vegas - January 8, 2003

2 2 THE AUTHORS INVESTIGATE THE APPLICATIONS OF ONE CAPABILITY OF THE CLIO SYSTEM: ADDING A DC COMPONENT TO THE EXCITATION SIGNAL. WHILE THE IMPLICATIONS SPAN FROM R&D TO PERFORMANCE ASSESSMENT IN GENERAL, THE AUTHORS FOCUS ON QUALITY CONTROL ENVIRONMENTS. AN ENGINEERING REPORT COMPARES SEVERAL QUALITY CONTROL TEST METHODS WITH AND WITHOUT THIS STRESSING TECHNIQUE. ABSTRACT

3 3 INTRODUCTION WE WOULD LIKE TO GIVE FIRST THE GUIDELINES WE FOLLOWED WHILE DEVELOPING THE QUALITY CONTROL PART OF OUR MEASURING SYSTEM, AS THESE ARE CONSISTENT WITH OUR PRESENTATION HERE

4 4 INTRODUCTION HOW IS AN EFFECTIVE QC MEASUREMENT SYSTEM DEVELOPED? MERGING WELL ESTABLISHED AND DOCUMENTED MEASURING TECHNIQUES WITH REQUESTS COMING FROM EXPERIENCED QC ENGINEERS 10 YEARS OF EXPERIENCE TELL US THAT THE QUALITY OF A QUALITY CONTROL ENVIRONMENT IS MORE DEPENDENT ON THE SKILL OF THE QC ENGINEER(S) RATHER THAN ON THE MEASUREMENT SYSTEM ITSELF

5 5 THE WORD QUALITY STAYS FOR THE FINAL RESULT, I.E. AVOIDING SHIPPING DEFECTIVE LOUDSPEAKERS TO CUSTOMERS; THEY SHOULD ALSO BE AS CLOSE TO THE REFERENCE AS POSSIBLE INTRODUCTION THE MEASUREMENT SYSTEM MUST HOWEVER BE FLEXIBLE, ACCURATE AND RELIABLE; IT SHOULD SPEED UP, AS MUCH AS POSSIBLE, ANY OPERATION INVOLVED WE ADDRESS PROBLEMS GIVING POWERFUL TOOLS TO ENGINEERS, NO WIZARDS OR UNIQUE MAGIC NUMBERS WHAT WE PRESENT HERE IS JUST ONE MORE TOOL

6 6 R&D ENGINEERS PRODUCTION LINES QC ENGINEERS CRITICAL TIME CONSUMING ITEMS ARE: A) USERS LEARNING CURVE B) QC PROCEDURES DEVELOPMENT C) EASY FITTING IN EXISTING PRODUCTION LINES D) STATISTICS AND RESULT REPORTING E) FEEDBACK TO TOLERANCE MASKS INTRODUCTION YOUR COMPANY

7 7 THE STRESSING METHOD WE ARE PRESENTING APPLIES, AS AN ADD-ON, TO VIRTUALLY ANY WELL ESTABLISHED MEASURING TECHNIQUE AUGMENTING ITS GOOD-TO-BAD SPREAD OUR ENGINEERING REPORT WILL COVER TWO DIFFERENT CASES ADDRESSING RUB&BUZZ DETECTION: INTRODUCTION SWEEP WITH RESPONSE AND DISTORTION VS. FREQUENCY ANALYSIS SINGLE FREQUENCY THD

8 8 GENERIC TERM MEANING AN ACOUSTIC DEFECT WITH IMPULSIVE (LOW ENERGY) CHARACTERISTICS MORE EASILY DETECTED BY THE END USER THAN LINEAR AND NON LINEAR CONVENTIONAL DISTORTIONS IT IS THEREFORE THE BLACK BEAST OF ANY QC ENGINEER, AS LACK OF ITS DETECTION HAS GREAT IMPACT ON COMPANYS IMAGE RUB&BUZZ IT IS CAUSED, DURING PARTS ASSEMBLY, BY SEVERAL DIFFERENT REASONS RANGING FROM DEFECTIVE PARTS TO PRODUCTION TOLERANCES IT CAN BE INTERMITTENT

9 9 STILL IN MANY ENVIRONMENTS AN HEARING TEST PRECEDES AN AUTOMATED QC TEST ITS INTENSITY RANGES FROM MACROSCOPIC - DETECTABLE BY ANY MEASUREMENT, I.E. LEVEL - TO VERY LITTLE - REQUIRING A QUITE ENVIRONMENT TO BE HEARD RUB&BUZZ THE FIRST CASE OF STUDY ADDRESSES THE LATTER SITUATION

10 10 BLOCK DIAGRAM AC OUTPUT WITH DC CAPABILITY (± 2.5V) CURRENT SENSING OUTPUT DC COUPLED AMPLIFIER (WITH 10 dB GAIN)

11 11 8 P.A. WOOFER EXAMPLE A ACOUSTICALLY (BY EAR) THE BAD UNIT EMITS A VERY LOW LEVEL CLICKING NOISE WHEN THE SWEEP REACHES THE RESONANCE REGION WHICH IS BETTER HEARD BELOW RESONANCE Fs 80 Hz Re 5.2 Ohm Bl Tm Cms 0.15 mm/N Qes0.45 Qms3.51

12 12 EXAMPLE A BELOW RESONANCE THE PROPER SOUND EMISSION DECREASES WHILE DISPLACEMENT CONTINUES TO INCREASE OR REMAINS THE SAME UNTIL DC (0 Hz) OBTAINING DISPLACEMENT IS THE FIRST REQUIREMENT TO UNDERLINE RUB & BUZZ IN RARE CASES DISPLACEMENT ALONE IS NOT SUFFICIENT AND A CONTEMPORANEOUS HIGHER FREQUENCY TONE IS REQUIRED

13 13 EXAMPLE A THE LATTER OBSERVATION SUGGESTS THAT WHEN THE CONE BEHAVES AS A RIGID PISTON (LOW FREQUENCY - HIGH DISPLACEMENT) IS ABLE TO REDUCE SOME HOW THE RUB & BUZZ ADDING A HIGHER FREQUENCY TONE BREAKS CONE PISTON BEHAVIOUR MAKING RUB & BUZZ MORE EVIDENT ALL THIS YIELDS TO THE POSSIBLE USE OF A TWO TONE SIGNAL FOR THE TEST LOOKING FOR THD AND IMD COMPONENTS BY FFT ANALYSIS

14 14 EXAMPLE A UNFORTUNATELY THE CHOICE OF THE FREQUENCY OF THE ADDED TONE IS NOT EASY TO BE DETERMINED ASIDE FROM EMPIRICAL BASIS WHILE THESE KIND OF TESTS ARE VERY FAST AND CAN THEREFORE BE REPEATED FOR MANY SECOND TONE FREQUENCIES, THEY DO NOT ALLOW SIMULTANEOUS FREQUENCY RESPONSE CHECK

15 15 EXAMPLE A THE BASIC IDEA FOR THIS EXAMPLE IS TO PERFORM A SWEEP TEST WITH FREQUENCY RESPONSE AND HARMONIC CHECK AND OBTAINING DISPLACEMENT AT ANY FREQUENCY DURING THE SWEEP BY ADDING A DC COMPONENT TO THE TEST SIGNAL. THE TEST SHOULD BE REPEATED TWICE, TO ADDRESS BOTH DISPLACEMENT DIRECTIONS

16 16 EXAMPLE A CRITICAL CHOICES FOR THE TEST ARE: AC SIGNAL LEVEL FREQUENCY RANGE DC SIGNAL LEVEL

17 17 EXAMPLE A TEST LEVELS, AC AND DC, AFFECT BOTH VOICE COIL DISSIPATION AND DISPLACEMENT FOCUS SHOULD BE ON DISPLACEMENT AS THIS IS THE LIMITING FACTOR; DISSIPATION LIMIT IS NOT AN ISSUE HERE THE FREQUENCY RANGE SHOULD BE EXTENDED AROUND 1 OCTAVE BELOW RESONANCE

18 18 NORMALIZED EXCURSION VS. FREQUENCY FOR Q FROM 0.2 TO 1 IN 0.2 STEP FOR CONSTANT VOLTAGE DRIVE EXAMPLE A

19 19 EXAMPLE A THIS GRAPH, DERIVED FROM SMALLS EQUATIONS [1], HOLDS FOR SMALL SIGNALS IT SHOWS THAT FOR VOLTAGE DRIVE, AS WELL AS FOR Q<0.7, WHICH IS THE CASE HERE AND IN MOST REAL LIFE APPLICATION, MAXIMUM DISPLACEMENT IS REACHED WELL BELOW RESONANCE AND APPROACHES THE DC ONE THIS STATEMENT SHOULD BE TAKEN WITH CARE

20 20 EXAMPLE A KLIPPEL [2], IN ASSESSING VOICE COIL PEAK DISPLACEMENT, SHOWED INCREASED DISPLACEMENT AT RESONANCE DUE TO BL NON LINEARITY BEING DOMINANT OVER SUSPENSION LIMIT GANDER [3] GIVES THE FOLLOWING RELATIONSHIP FOR FORCE VS. VOLTAGE DRIVE AT RESONANCE

21 21 EXAMPLE A THE BL IN THE DENOMINATOR SHOWS THAT BL REDUCTION, DUE TO HIGH DISPLACEMENT, INCREASES FORCE AND GIVES REASON FOR THE EXPANSION EFFECT AT RESONANCE HOWEVER DURING QC TESTS GREAT OVERLOAD, I.E. GOING IN STRONG NON LINEAR AREA, SHOULD BE CONSIDERED WITH CARE FOR THE FOLLOWING REASONS:

22 22 EXAMPLE A QC HAPPENS, MOST OFTEN, JUST AT THE END OF THE ASSEMBLY PROCESS BEFORE PACKAGING; IN THIS PHASE, GLUES AND CONE TREATMENTS SETTING MIGHT NOT BE COMPLETE QC DEFINES ITS PASS/FAIL LIMITS UPON A REFERENCE OR AN AVERAGE OF PRODUCTION. STRONG NON LINEARITY MAKES LIMITS AMBIGUOUS AND THEREFORE DIFFICULT TO HANDLE NOISE POLLUTION IN GENERAL IS AN ISSUE BOTH FOR OPERATORS HEALTH AND INTERFERENCE BETWEEN LINES

23 23 EXAMPLE A THE FOUNDATION OF THIS APPROACH IS TO ACHIEVE COMPLETE KNOWLEDGE OF CONES DISPLACEMENT WE CHOSE AN AC LEVEL FOR A MAXIMUM DISPLACEMENT CORRESPONDING TO 3÷5% OF 3RD HARMONIC DISTORTION [3] WHICH HAPPENS AT THE LOWEST SWEEP FREQUENCY (45Hz)

24 24 EXAMPLE A FROM THE FOLLOWING RELATIONSHIP [3],[4] WE OBTAINED Xpeak=1.35mm=Xmax

25 25 EXAMPLE A WE GAVE A QUICK METHOD TO DETERMINE MAXIMUM EXCURSION; SHOULD YOU KNOW IT AS A TECHNICAL SPECIFICATION YOU MAY EASILY REVERT THE FORMULA TO OBTAIN THE DESIRED SPL IF THE TECHNICAL SPECIFICATION COMES FROM A COMMERCIAL BROCHURE, WE WARMLY SUGGEST THE FIRST APPROACH!

26 26 EXAMPLE A WE CHOSE A DC VALUE CORRESPONDING TO 1/3*Xmax=0.45mm TO MAINTAIN THE MAXIMUM DISPLACEMENT WITH DC, ON THE DC SIDE, AROUND THE SAME, THE AC LEVEL HAS BEEN REDUCED BY 2dB WHILE THE RELATIONSHIP BETWEEN DC CURRENT AND DISPLACEMENT IS FAR FROM BEING LINEAR [5], WE CONSIDER THE ERROR ACCEPTABLE FOR OUR PURPOSES

27 27 EXAMPLE A GRAPHICAL REPRESENTATION OF LEVELS AT VERY LOW FREQUENCIES AC+VDC AC ONLY AC-VDC

28 28 EXAMPLE A CONSIDERATIONS ON THE RESULTS: THE LEFT SIDE GRAPHS SHOW VERY LITTLE DIFFERENCE BETWEEN BAD AND GOOD UNITS IN THE LOWEST HARMONICS, NOT USABLE TO ESTABLISH RELIABLE QC LIMITS HIGHER HARMONICS, BELOW RESONANCE, SHOW AN APPRECIABLE BUT STILL SMALL DIFFERENCE, WHICH DOESNT CORRELATE WITH THE HEARD PHENOMENON THE RIGHT SIDE GRAPHS SHOW A MUCH INCREASED DIFFERENCE BETWEEN BAD AND GOOD, EVEN IN LOW HARMONICS, FOR A WELL EXTENDED FREQUENCY RANGE

29 29 2ND HARMONIC - NO DC2ND HARMONIC - WITH DC EXAMPLE A 3RD HARMONIC - NO DC3RD HARMONIC - WITH DC FUNDAMENTAL GOOD UNIT BAD UNIT

30 30 4TH HARMONIC - NO DC4TH HARMONIC - WITH DC EXAMPLE A 5TH HARMONIC - NO DC5TH HARMONIC - WITH DC FUNDAMENTAL GOOD UNIT BAD UNIT

31 31 8 P.A. WOOFER EXAMPLE B THE SAME MODEL OF THE PREVIOUS EXAMPLE WITH A BAD SAMPLE SHOWING MORE EVIDENT RUB&BUZZ DEPENDENT ON DISPLACEMENT ONLY Fs 80 Hz Re 5.2 Ohm Bl Tm Cms 0.15 mm/N Qes0.45 Qms3.51

32 32 EXAMPLE B AS CLEARLY POINTED OUT BY TEMME [6], THD IS NOT ABLE TO RELIABLY DETECT RUB&BUZZ THIS MAINLY HAS TWO REASONS; RUB&BUZZ CREATES IMPULSIVE LOW ENERGY DISTORTION WHILE LOUDSPEAKERS SHOW RELATIVELY HIGH ENERGY LOW ORDER HARMONICS NEVERTHELESS CLASSICAL THD MEASUREMENT IS VERY ATTRACTIVE DUE TO ITS ONE VALUE RESULT AND, WHEN PERFORMED WITH ACTUAL DSP SYSTEMS, TAKES A FRACTION OF A SECOND TO EXECUTE

33 33 EXAMPLE B OUR APPROACH HERE IS TO REDUCE SUBSTANTIALLY THE AC TEST LEVEL DEALING THEREFORE WITH A MUCH MORE LINEAR DEVICE AND, MOST IMPORTANT FOR THE QC ENGINEER, A SINGLE VALUE RESULT MEANS A SINGLE VALUE LIMIT MASK THE AC LEVEL WAS INTENTIONALLY OVER REDUCED, IN THE DC CASE, BY 15dB THE DC LEVEL WAS SET TO OBTAIN HALF XMAX AS DETERMINED IN PREVIOUS EXAMPLE

34 34 EXAMPLE B THE RESULTS ARE SPECTACULAR IN TERMS OF BAD TO GOOD DETECTION IN THE DC CASE DUE TO THE LOW AC LEVEL THE TEST FREQUENCY WAS SET AT THE DEVICE RESONANCE THE SPEED OF THE TEST PERMITS ITS REPETITION AT DIFFERENT DC LEVELS TO COVER FULL EXCURSION THE AC REDUCTION LEVEL TO THE NUMBER OF DC STEPS REQUIRED TO COVER THE ENTIRE EXCURSION IS LEFT TO QC ENGINEERS NEED

35 35 GOOD UNIT - NO DC EXAMPLE B BAD UNIT - NO DC THD 5.13% THD 3.65% GOOD UNIT - WITH DCBAD UNIT - WITH DC THD 23.7% THD 0.63%

36 36 CONCLUSIONS THE SIMPLE TECHNIQUE PROPOSED HAS PROVEN TO BE EFFECTIVE IN AUGMENTING GOOD-TO-BAD SPREAD IN QUALITY CONTROL TESTS IT CAN BE EASILY ADDED TO ANY MEASUREMENT SYSTEM AS AN EXTERNAL ADD-ON, TYPICALLY INCLUDED IN THE POWER AMPLIFIER.

37 37 [3] M.R.GANDER MOVING-COIL LOUDSPEAKER TOPOLOGY AS AN INDICATOR OF LINEAR EXCURSION CAPABILITY - JAES V29, No1/2, JAN/FEB 1981 BIBLIOGRAPHY [4] D.B.KEELE, JR. LOW FREQUENCY MEASUREMENT OF LOUDSPEAKER BY THE NEAR- FIELD SOUND PRESSURE TECHNIQUE - PREPRINT No909, 45TH AES CONVENTION, 1973 [6] S.F.TEMME ARE YOU SHIPPING DEFECTIVE LOUDSPEAKERS TO YOUR CUSTOMERS? - LISTEN, INC. [2] W.KLIPPEL ASSESSING VOICE COIL PEAK DISPLACEMENT - 112TH AES CONVENTION, 2002 [5] J.SCOTT, J.KELLY, G.LEEMBRUGGEN NEW METHOD OF CHARACTERIZING DRIVER LINEARITY - JAES V44, No4, APR 1996 [1] R.H.SMALL CLOSED BOX LOUDSPEAKERS SYSTEMS PART 1: ANALYSIS - AES LOUDSPEAKERS ANTHOLOGY V1-V25 ( )


Download ppt "1 ALMA 2003 Winter Symposium World Cone Symposium World Cone Symposium A Simple Approach to Improve Sensitivity of Quality Control Procedures of Loudspeakers."

Similar presentations


Ads by Google