2. CERN benjamin.todd@cern.ch Converter Control Electronics Strategy for LHC Machine Electronics : Limitations & Risks v4benjamin.todd@cern.ch

3. CERN Converter Control Electronics Introduction 3 I am assuming that HL-LHC engineering implies all topics are open to discussion. This talk focusses on observations about our machine electronics strategy questions arise which need to be answered for the HL-LHC… ….this talk does not attempt to provide answers!

4. CERN benjamin.todd@cern.ch Converter Control Electronics WorldFIP

5. CERN benjamin.todd@cern.ch Converter Control Electronics Machine Control System 5 Industrial Computer Custom Hardware

6. CERN benjamin.todd@cern.ch Converter Control Electronics What this means… 6 optical fibre cable “WorldFIP” is a combination of field-bus related parts

7. CERN benjamin.todd@cern.ch Converter Control Electronics Re-engineering the fieldbus 7 BE/CO insourcing of Alstom industrial Products WorldFIP  cernFIP nanoFIPslave IP core [2012] goFIP repeater [2014] masterFIP [2015] approach is to re-engineer parts as they become obsolete ALSTOM repeater – sources owned… nanoFIPdiag – sources owned… E. Gousiou [BE/CO]

8. CERN benjamin.todd@cern.ch Converter Control Electronics Bandwidth and Impact 8 no access needed to repair Commercial solution very high performance access to repair commercial solution considerable performance significant intelligence in the custom hardware Severe bandwidth limitation! BE/CO  best case is only doubling to 5Mbps using worldFIP/cernFIP means complex electronics will always be next to the machine

9. CERN benjamin.todd@cern.ch Converter Control Electronics Radiation Level Predictions

10. CERN benjamin.todd@cern.ch Converter Control Electronics Surface Buildings machine electronics are installed in one of five areas with machine radiation risks: none Perpendicular galleriesnone - low Parallel gallerieslow Alcovesmedium - high LHC Tunnel high 10

11. CERN benjamin.todd@cern.ch Converter Control Electronics In the Tunnel - FGClite If HL-LHC radiation levels are below predictions the correct strategy depends on the predicted radiation level 11 Dispersion Suppressor = 5% Arc = 60% 2010-122015-182020-232024(+10) <1 2  4 24816  32 Dose per Year [Gy] <1 <6 <4 ≈18 <9 ≈42 ≈30-50 ≈200-370 Cumulative Dose [Gy] OK low end of predictionrotate 10% every LS high end of predictionrotate 10% every year, 20% spares above predictionsno solution Cumulative Dose 150Gy = limit for FPGA reprogramming

12. CERN benjamin.todd@cern.ch Converter Control Electronics Single-Source COTS in Radiation

13. CERN benjamin.todd@cern.ch Converter Control Electronics The FGClite FGClite – S. Uznanski et al. Processing Data Storage Analogue Digital Differential Current Loop Onewire Input/Output Field-Bus 13 WorldFIP / cernFIP

14. CERN benjamin.todd@cern.ch Converter Control Electronics COTS Processing Analogue Actel ProASIC 3 FPGA Texas Instruments ADCs These exact same parts are used in a lot of systems (QPS, RadMON, FGC…) some commercial parts are single-source solutions for electronics in radiation the Actel proASIC3 FPGA is the only viable solution for FPGAs in radiation 14 Above 150GY the FPGA cannot be re-programmed Alstom FielDrive & FieldTR Field-Bus the Alstom circuits are the only viable solution for the fieldbus electrical level the Texas ADCs are the only solution for ADCs in several systems

15. CERN benjamin.todd@cern.ch Converter Control Electronics PH/ESE Solutions for Machine Electronics

16. CERN benjamin.todd@cern.ch Converter Control Electronics Recipe for Electronic Systems processing data storage analogue digital differential current loop onewire Input/Output almost every machine electronic system I have seen at CERN… Configuration Program Variables µC / µP FPGA / CPLD DSP powering 16

17. CERN benjamin.todd@cern.ch Converter Control Electronics Recipe for Electronic Systems powering A very interesting example … when EPC/CCE began FGClite we spent a lot of time and effort qualifying regulators… initially settled on the MIC37302 Criticality + Testing + Long-Term Availability + Batch Variation + Process Variation + … = very high risk to our product Turned out to be justified, unusual failure mode observed in radiation  this part is not suitable for FGClite. 17 around one year, several radiation campaigns.

18. CERN benjamin.todd@cern.ch Converter Control Electronics Recipe for Electronic Systems powering A very interesting example … We finally chose LHC4913PDU Designed by PH/ESE  in the CERN stores SCEM 08.57.56.011.708.57.56.011.7 Criticality + Testing + Long-Term Availability + Batch Variation + Process Variation + … = risk minimised cost for risk mitigation… MIC37302 = 3 CHF  LHC4913PDU = 16 CHF 18 x6000 = 80kCHF PH/ESE solutions/approaches applied to machine electronics can give significant risk reduction

19. CERN benjamin.todd@cern.ch Converter Control Electronics Organisational Strategy

20. CERN benjamin.todd@cern.ch Converter Control Electronics A Very Personal Opinion 20 I would argue: We all have more to do with less …. * under-resourcing eventually means we need to be organised differenty * Knowledge management is becoming critical Maintenance must be easy to do, despite the changing workforce * Having fewer, common, more simple front-end electronic units would help in several ways * Electronics projects traversing the organisation An example which works extremely well – ex. TS/DEM “Outsourcing” electronics implementation and production to a centralised team of experts Coordinated by leaders within each equipment group/section Fewer staff, more students, more fellows, less long-term presence In addition, skill needs are evolving * Skill level needed in each electronics domain means one person can no longer master the whole common approaches are becoming strategically important

21. CERN benjamin.todd@cern.ch Converter Control Electronics Fundamental Questions for (HL-)LHC

22. CERN benjamin.todd@cern.ch Converter Control Electronics Observations  Questions 22 1. using worldFIP means complex electronics will always be next to the machine 3. some critical COTS components in radiation are single source 2. machine electronics strategy has a high sensitivity to radiation predictions 4. PH/ESE solutions / approaches applied to machine electronics can give significant risk reduction What does this mean for critical systems? Arguably risk varies from negligible to severe each shared COTS part presents a risk, the sum of these risks should not be ignored Session 1? what should we do to address this? Profit from this knowledge and help knowledge transfer? What does that mean for the infrastructure? 5. Common approaches are becoming strategically important

23. CERN benjamin.todd@cern.ch Converter Control Electronics Asking a Fundamental Question…

24. CERN benjamin.todd@cern.ch Converter Control Electronics FGClite – Hypothetical Evolutions 24 I asked S. Uznanski what would FGClite have looked like if we had no initial constraints FGC2 FGClite

25. CERN benjamin.todd@cern.ch Converter Control Electronics FGClite – Hypothetical Evolutions 25 analogue digital differential current loop onewire Input / Output almost every machine electronic system I have seen at CERN…