From HALT Results to an Accurate Field MTBF - Presentation

From HALT Results From HALT Results To An Accurate To An Accurate

Field MTBF Field MTBF EstimateEstimate

Harry McLeanAdvanced Energy Industries

January, [email protected]

This material is protected by the copyright laws of USA.Any reproduction of this material is prohibited by law.

Mike SilvermanOps A La Carte

January, [email protected]

Speaker’s BiographyHarry McLeanMember of Technical StaffAdvanced Energy Industries, Inc.Over 20 years direct involvement with techniquesCredentials:

BSEE Northeastern University in Boston.Invented HASA process at HP for DeskJet Printers. Advances made at AT&T.Authored papers on reliability and facilities engineering.Republished book HALT, HASS & HASA Explained in 2009 through ASQ.Holds four reliability patents. This is the fifth one.Has taught HALT & HASS for last 15 years in US and Brazil in Portuguese.Held various positions during 25 year career at HP.Managed Reliability Engineering & Quality Systems at AT&T Wireless for 5+ years and 3 years at Xantrex Technology in British Columbia, Canada.

Have we ever wanted to use the HALT data to estimate AFR and we were…

• Told it couldn’t be done…• Frustrated by your lack of data…• Lacking the bandwidth to develop a model…• Other impediments?

A Few Questions

Historical• Began thinking about this in late 1990’s.• Began developing a model in 2000.• Stopped shortly thereafter.• Development restarted with new approach in 2005.• Demoed model to 3 engineers in 2008.• Obtained validation data from others late 2008.• Presentation at IEEE/ASTR October 2008.• Validation in 2009.• Available through Ops A La Carte (408) 472-3889.

We will discuss a high level concept of the mathematical model but will NOT discuss its details as it will be patent pending.

Some Background• Highly Accelerated Life Testing (HALT) is a

great reliability process used for quickly finding failure mechanisms in a hardware product.

• In many cases, there is a need to know the MTBF or Actual field Failure Rate (AFR) of a product in the field (customer requirement).

• When this is the situation, most people turn to Reliability Demonstration Test (RDT).

5

Is there a better way?

What is the AFR Estimator• It is a patent pending Excel-based mathe-

matical model that, when provided with the appropriate HALT and product information, will accurately estimate the product’s field AFR or Actual field Failure Rate.• Three acceleration models are used, linear,

exponential, and quadratic.• The model will also provide HASS or HASA

time to detect a shift in the desired outgoing failure rate.

• The AFR Estimator has been validated on almost thirty products from diverse design environments and manufacturers.

6

Recommendation:Obtain “HALT, HASS, & HASA Explained”

It is highly recommended that you have a copy of, “HALT, HASS, & HASA Explained”, and use it as a reference when performing HALT, HASS, or HASA. This is the best source when performing any of these on a product, and following the book will ensure that the stressing process will be properly done and that the calculator will provide excellent results.

7

“HALT, HASS, and HASA Explained – Harry McLean at ASQ.org or Amazon.com

• Complete at least one HALT observing:The need for a sample size of at least three, preferably four units. Model can accommodate 1 to 6.

Realize that HALT sample sizes of three or less will dramatically affect the ability to detect product defects and hence, the statistical confidence is likewise, impacted.

Performing HALT at each phase of Product Development Process.HALT Product Operational Response Limits are required.Ten minute (or greater but consistent) dwells for thermal and vibration.Including rapid thermal and combined environments. (Accounted for but not used as user inputs to model.)

Continued

To Maximize Use of the Model

To Maximize Use of the ModelTest product throughout. Include power cycling. Use a robust test protocol.All issues are corrected at least up to Guard Band Limits (beyond, preferably).

• Timely corrective actions are verified in HALT.• The units for HALT are the same configuration as field ones.

All interfaces, even if tested on another design, must be re HALT’d.

• The end use environment is the same as HALT tested one.Clarification: Load, thermal, product duty cycle, AC power, etc.are all to be considered when applying a HALT qualified unit in a different application.

• Access to MTBF estimate

Guard Band, Spec & End Use

New End Use

Prod Spec

End Use

Linear Acceleration

Two other acceleration models are used in the estimation equation.

Overview of EquationsHALT AFR:• AFR, ƒ (MTBF*Factor1, Thermal Range*Factor2, Vib-

ration*Factor3, Vibration Table*Factor4, Sample Size *Factor5)

Confidence Limits:• χ2 from Semi E10 (based on HALT AFR & HALT sample

size)

Days for Detectable Shift in AFR (HASS):• n = (Zα+Zβ)2*p*(1-p)/d2 and Days = n/Sample Size

Limitations of the Model• The model has not been validated on mechanical

designs.• The estimate is as good as the test protocol used in

HALT and other reliability tests. HALT does not capture every possible design defect, i.e., humidity related issues, field operation beyond Guard Band limits, some wear-out mechanisms, etc.• The units in HALT need to be tested with a protocol that

sufficiently tests the product in each stress environment. A recommended starting point is 75% test coverage.

Why Use the AFR Estimator• HALT takes a few days to run and to implement its

corrective action(s), and even if it’s a bit longer, this time would be far less than waiting for an RDT to run and to implement its corrective action(s).• It can be a huge time and cost saver. Consider life test for new

technologies or existing part/design in a different application but not to accurately estimate AFR.

• Higher HALT limits equate to lower AFR, this tool can accuratelyestimate the field AFR before launching the product.

• Stress levels in the Prod Env Table are highly recommended for HALT. These can assure the product should exceed customer expectations and for accurate forecasting of warranty expenses.

• Seven to ten simple data points (most coming from HALT), the AFR Estimator can provide accurate field AFR instantaneously and 90% statistical confidence limits based on HALT sample size and AFR.

14

Why Use the AFR Estimator• For HASS or HASA:• Will you perform HASS or HASA (Yes=1, No=0),

The Daily Sample Size is the number of units that will be subjected to the HASS or HASA process in a twenty-four hour shift.

If the HASS or HASA process control chart varies dramatically from shift to shift, then use an eight hour shift sample size until the control variables are under statistical control.

The Detectable Shift in AFR is the delta between the outgoing AFR and the detectable shift in outgoing quality (from HASS or HASA) that you wish to detect. For example, if the product baseline AFR is 4% and the worst case AFR is 10%, the Detectable Shift value is to be 6 (6%).

15

PublishedSpec, °C

Level Application Guard Band, °C

0 to +40 1 Consumer -30 to +80

0 to +50 2 Hi-end Consumer -30 to +100

-10 to +50 3 Hi Performance -40 to +110

-20 to +50 4 Critical Application -50 to +110

-25 to +65 5 Sheltered -50 to +110

-40 to +85 6 All Outdoor -65 to +110

Product Type & Guard BandProduct Environment & Level

HALT Chamber

• What was the chamber used for the HALT?• In HASS or HASA, one will need to normalize the

vibration data – you must be consistent.

Example for Vehicle Inverter/ChargerField Failure Rate Estimate - % of Failures/Year

Input Matrix Data VerifiyMTBF (in Hrs) = 56,800 OK Key

Product Thermal (Hot in °C) = 80 OK User inputProduct Thermal (Cold in °C) = -35 OK Calculated

Product Vibration (in Grms) = 17 OK SelectionProd Published Spec Level (see below) = 3 OK Data Validity

Vibration Table Technology = 2 OKNumber of HALT Samples = 4 OK

HASS or HASA (yes = 1, no = 0) = 0 OKIf HASS or HASA, Daily Sample Size = 1 OK

If HASS or HASA, Detectable Shift in AFR (in %) = 0 OK

Steady State AFR, % (HALT Only) = 3.40Steady State Field MTBF, Hrs (HALT Only) = 257,513

Lower 90% HALT Confidence Limit = 138,826Upper 90% HALT Confidence Limit = 529,297

Days to Detect Shift w/ HALT/HASS/HASA (Max) =

Published Spec Level # Guard Band Limits0 to +40 1 Consumer -30 to +800 to +50 2 Hi-end Consumer -30 to +100

-10 to +50 3 Hi Performance -40 to +110 -20 to +50 4 Critical Application -50 to +110 -25 to +65 5 Sheltered -50 to +110 -40 to +85 6 All Outdoor -65 to +110

Example of Office Product

Actual Field AFR = 0.8%

Field Failure Rate Estimate - % of Failures/Year

Input Matrix Data VerifiyMTBF (in Hrs) = 3,199,090 OK Key







Lower 90% HALT Confidence Limit = 349,429Upper 90% HALT Confidence Limit = 1,332,260




Two Estimator ExamplesHigh Power Charger Vehicle Power Inverter























Validation TableCalculated HALT Results Table AFR, % Return

Products: MTBF AFR, % Hot Cold Vib Level Tech HALT Calc Field Act Rate, %Display 415,000 2.1 130 -80 28 4 2 0.27 0.21Outdoor 175,800 5.0 100 -60 28 6 3 1.26 0.75 8.30Vehicle 143,600 6.1 102 -67 20 6 2 1.35 1.05 2.80Vehicle 342,100 2.6 100 -30 31 2 2 1.07 0.70 5.60Outdoor 275,000 3.2 110 -60 17 6 3 1.40 0.30 3.70Vehicle 157,100 5.6 90 -60 21 2 2 1.17 0.90 10.10Vehicle 192,500 4.6 90 -60 21 5 2 1.15 1.00 9.80Vehicle 106,800 8.2 100 -50 13 2 2 2.27 2.20 14.06Hi Perf 616,200 1.4 110 -42 19 3 3 1.28 1.00

Vehicle 56,800 15.4 80 -35 17 3 2 3.40 3.75Vehicle 109,800 8.0 105 -35 14 6 2 3.25 4.40 8.40

Office 3,199,090 0.3 80 -50 20 1 2 1.35 0.8 1.4Telecom (Out) 200,000 4.4 100 -80 28 6 1 0.83 0.5

Telecom 200,000 4.4 83 -82 31 4 1 0.88 0.5Telecom 200,000 4.4 85 -60 50 4 1 1.21 0.5Telecom 200,000 4.4 121 -54 21 4 1 1.06 0.5Telecom 200,000 4.4 102 -72 25 4 1 0.82 0.5

Consumer 70,000 12.5 100 -30 10 1 2 5.31 3.00Consumer 70,000 12.5 100 -30 16 1 2 3.13 3.00Consumer 70,000 12.5 90 -30 19 1 2 2.82 2.92

Avionics 17,000 51.6 120 -70 48 4 3 1.18 1.36 1.89Avionics 32,000 27.4 125 -100 48 4 3 0.39 0.78 1.61Avionics 14,000 62.6 120 -60 50 4 3 1.87 1.55 2.12Avionics 18,900 46.4 120 -80 62 4 3 0.89 1.78 2.49Avionics 20,600 42.6 120 -90 65 4 3 0.57 1.16 2.43Avionics 14,600 60.0 120 -90 55 4 3 0.81 0.16 0.61Avionics 71,000 12.3 120 -65 40 4 3 0.86 0.51 1.08Avionics 11,000 79.7 125 -70 25 4 3 1.34 0.36 2.08

Validation Table ClarificationProducts – Generic product type.Calculated MTBF – Estimated by Telcordia SR-332, Issue 1,

Parts Stress Method or equivalent tool.Calculated AFR – Actual field Failure Rate from MTBF.HALT Results – Product limitations from HALT.Level – Lookup value of the product’s published specs.HALT Calc AFR% – Calculated AFR from formula.Field Actual AFR% – Actual field AFR after HALT.Return Rate, % – Product return rate with failed units.

Observations on Validation Table

• Outdoor w/275K MTBF – HALT vibration is low.• Vehicle w/109.8K MTBF – The HALT results are

low.• Consumer – Let’s see why the large variation…

Comparison of Field Results to Estimate

Guard Band=20Grms 1st HALT Vibration was 10Grms, last HALT reached 19Grms, AFR was 2.82 and Delta -0.10.

‐1.00

0.00

1.00

2.00

3.00

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28

% Delta AFR Between Model & Field

Details of Large Variance

‐1.00

0.00

1.00

2.00

3.00

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28

% Delta AFR Between Model & Field

Parameter HALT Guard Band AFR HALT Result AFR

Thermal -30°C to +100°C ----- -30°C to +100°C -----Vibration 19Grms ----- 10GrmsSample Size 4 1.81 3 5.31Delta 3.50

Work Remaining• More data from others (would like 12 more):

HALT results (final values after corrective actions).Sample SizeCalculated (estimated) MTBF, if available.Actual field Failure Rate (AFR).Field return rate.Type of product, i.e., internet server, etc.Published thermal and vibration specs.If HASS is being done.

• Confidence limits. Completed!• Include effects of HASS. Completed!• Prepare for posting to website. Completed!• Patent submittal and grant. Progressing.

Conclusions• The methodology works very well.• Would like more data – would you like to help?

From HALT Results to an Accurate Field MTBF - Presentation

Documents