1. PSU-Lab: A valuable expertise shared and open to CERN users
TE-EPC, CERN Switzerland – EDMS N°
Abstract This poster presents a specific service provided by TE-EPC (Power Converter Group) for its internal needs, but also open to TE Department members. The aim of this service is to help with the selection of Auxiliary Power Supplies (low power < 200 W) for the demanding physics accelerator environment. The provided analysis focuses on EMC &reliability, key factors for these units powering critical electronics systems.
The context & the needs
Almost any electronic system uses AC-DC Auxiliary Power Supplies. PSU-Lab help to choose & organize their tests.
A large scope of different domains to qualify a reference.
Qualifying an Auxiliary Power Supply requires to measure its compatibility with CERN environment. PSU-Lab focuses mainly on EMC, AC mains events (storms) and reliability.
PSU-lab Records
15 records in less than 3 years.
PSU-Lab Infrastructure
A specific lab, located in building 287, was designed for series test reception and type test qualification.
Type test - selecting
Reception Test - series
It requires to invest time  a week, testing several electrical features under specific stress. This test can disqualify a given unit if it is finally judged of poor quality.
These tests are required when receiving the series.
Two phases can be applied: - a short burn-in on all units: ex:  8 hours each - a longer period on some: ex:  : 6 months for 10 units.
A service adapted to different needs.
Domain
Field
Typical Measurements & Checks
EMC
Compatibility
Conductive Noise, Input and Output side.
Susceptibility
Burst Test, Input and Output side + Surge Test.
Electrical
Performances
Efficiency, Output Ripple, Starting sequence, short case.
Input Line Variations
AC phase loss, Over-Voltage conditions, Inrush current.
Output Load Variations
Dynamic Response to sudden load variations.
Fab. Quality
Process
Solder quality, PCB Layout, Component placement.
Design
Topology
Partial overview of pro / cons of topology reviewed.
Components
Checking critical components*: technology, ratings.
Reliability & IR Inspection
Infra-Red Inspection
A systematic Infra-Red inspection is performed.
Accelerated Testing
Evaluation of the reliability: lifetime* & MTBF*.
* Strongly dependent on number of samples and time allowed for providing results.
10 units under long reliability test.
PSU-Lab & TE-CRG Collaborations
A long history, with great collaboration and valuable achievements!
Chasing the unreliability, a real quest!
Not an easy task, but required for high availability systems.
A detailed & structure report, with one summary page
Report details each test, providing a clear global summary (1). 1
Collaboration-1
Which delivery?
Which result?
In 2009, a severe reliability issue on 200 units, each critical for LHC cryogenics systems, was discovered.
A complete survey was performed, at the level of the failed units but also the environment. All units were upgraded, tested on a dedicated tester designed by PSU-Lab & returned to operation.
Upgraded units were measured in operation, with a demonstrated MTBF higher than two Millions of hours, with only a single failure in four years on 200 units!
Collaboration-2
In 2012, TE-CRG decided to upgrade their 24 V.dc systems.
PSU-Lab helped them to design and maintain it, analysing every failure in operation. (see slide beside).
Reviews were organised, inviting an external EMC expert. Initial design was updated on EMC & on redundant solution being implemented.
Any selected units were individually evaluated and qualified, with even long reliability survey (8-months) on some PSU references.
A fruitful collaboration with a demonstrated redundant system with transparent failures to operation.
The system was carefully designed, following experts recommendation, and using only qualified units on several references.
Implicated from the beginning, PSU-Lab was later able to find the root cause on too low MTBF selected units.  
Context of failure happening on highly critical cryogenic system.
During [2013; 2015] operation, a unit reference selected by SU-Lab showed abnormally low MTBF, suspecting CERN related environment root cause.
PSU-Lab traced all failure history, identifying a common failure mode.
PSU-Lab evaluated each failed unit, and operating conditions on site as well.
Advanced technics (Weibull) were used to conclude issues were linked to an aging phenomena, with safe applied electrical conditions. After heavy and long measurement campaign, analyses, the root cause was found to be DUST combined with a high Issue was quickly solved unit sensitivity to it. once understood !  
Each test is explicated (2) in its method & goal, presenting clear results (3), and how to easily interpret them (4). 2 3
Select do not compress images in file > options > advanced 4
Can PSU-Lab answer if a reference is reliable enough?
- Visual inspection detects: poor quality assembly, poor components choice, chosen derating not sufficient, poor design - theoretical.
- Electrical tests + infra-red inspection detects: fragile or sensitive design (burst, surge, start on short), poor design - practical.
- What about MTBF and lifetime? Shouldn’t we try to estimate them? > Proving a MTBF (random failure) is at least hours requires a 0-failure test with 100 units operating 3.1 months (only at 90 % confidence). > Proving a lifetime of 10 years is much longer (years) & not realistic.