the reliable power supply – an oxymoron? can a power supply actually be reliable? is there such a...
TRANSCRIPT
The Reliable Power Supply – an Oxymoron?
Can a power supply actually be reliable? Is there such a thing as a Million Hour Power
Supply? Can you just specify the reliability and get it? Picking a good supplier is important, but is it
enough? How about MTBF calcs? --stress analysis?
–---What kind of testing?
In Search of Dependable Power
Pick a good supplier Specify the important stuff Trust but Verify Measure Field performance
You want a Good Power Supply?
No matter whether OEM or Custom:
Check out your supplier!
Important Specifications
Field MTBF Life Expectancy Require Failure Analysis of returned units MTBF Calculation?
Supplier check list
– ISO certification; Quality Manual
– Design (review and test)– Manufacturing (FA and CA process)– Material control (AVL)– Effective QUALITY Management– Reliability reporting and feedback
Stick to New Technology
Picking a good Supplier
You’re going to have problems: make sure you pick one that you can deal with– Accessibility of responsible personnel– Receptive attitude– Dialog
Make sure he understands that “Failure Analysis” is not the same as “Repair”
Proof of the Pudding
Don’t expect the Field to take care of itself– Real-time field tracking exposes problems as they
happen Design problems Revision incompatibility Customer specific Supplier process problems
Field Data Tracking
FACTORY CONTROL SYSTEM
INSTALLED SYSTEMS CONFIGURATION
DATABASE
FIELD SERVICE ACTIONS
DATABASE
HP NonStop Tracking Databases
How Good is Good?
Range from 300,000 to 1.6 Million hoursActual
Pwr Supply Watts Outputs Spec Field MTBFA1 750 5 1,000,000 1,500,000A2 750 5 1,000,000 300,000B1 1500 1 1,500,000 1,752,000B2 1500 1 1,500,000 584,000
D1.1 1200 1 1,500,000 324,444D1.2 1200 1 1,500,000 547,500D2 1200 1 1,500,000 673,846E1 276 1 1,000,000 1,135,000E2 276 1 1,000,000 1,135,000F1 460 1 750,000 585,000F2 460 1 750,000 585,000
Two Vendors, Same PS
One year’s Field Data3-month rolling average
1,464,570
314,755
0
500,000
1,000,000
1,500,000
2,000,000
2,500,000
Apr
May Jun
Jul
Aug Sep Oct
Nov
Dec Jan
Feb
Mar
Apr
May
MT
BF
(H
rs)
Vendor B
Vendor A
Early Wearout
Weibull Hazard Plot for System Power Supply
Slope = 1
0.01%
0.10%
1.00%
10.00%
100.00%
1 10 100 1000 10000 100000
Time to Replacement Since First Customer Ship
Cu
mu
lati
ve
Ha
zard
Slope > 1 : Increasing Failure RateSlope = 1 : Constant Failure RateSlope < 1 : Decreasing Failure Rate
(Hours)
Wearout began at ~2 years
Tales from the Crypt
AVL Control Customer environment Fan Selection Design problem Board layout Manufacturing process
Conclusion: In Search of Dependable Power
Pick a good supplier Specify the important stuff Trust but Verify Measure Field performance