Mem. S.A.It. Vol. 86, 498c© SAIt 2015 Memorie della

Industrial solutions trends for the control ofHiRes spectrograph@E-ELT

P. Di Marcantonio, V. Baldini, G. Calderone, R. Cirami, I. Coretti, and S. Cristiani

Istituto Nazionale di Astrofisica – Osservatorio Astronomico di Trieste, Via Tiepolo 11,I-34143 Trieste, Italy, e-mail: dimarcan@oats.inaf.it

Abstract. Starting a few years ago, ESO initiated a number of projects aiming to ex-plore the possible adoption of industrial standards and commercial off-the-shelf compo-nents (COTS) for the control of future VLT and E-ELT instrumentations. In this context,ESPRESSO, the next generation high-stability spectrograph for the VLT and to a certainextent, a precursor of HiRes, has adopted since the preliminary design phase those solu-tions. Based on the ESPRESSO experience and taking into account the requirements in-ferred from the preliminary Hi-Res studies in terms of both high-level operations as well aslow-level control, I will present in this paper the current proposal for the HiRes hardwarearchitecture.

Key words. E-ELT instruments: control electronics – E-ELT instruments: control software– COTS – Industrial standards

1. Introduction

In the past, due to limited industrial solu-tions that could fit astronomical requirementsin terms of devices control, there was no otherchoice as to develop, ad-hoc, custom solutionsfor most of the hardware/software control re-lated aspects. This was the case for the ESOVLT Observatory and in the literature a lot ofexamples could be found that report the samehappening for other Observatories (Ghedina etal. 2014).

Even though requirements were in this wayfulfilled, this approach has inherently seve-ral drawbacks: custom solutions are expensiveboth to develop and implement as well as tomaintain (especially on a long term). Engineersthat contribute to the initial development notalways remain in the same working place andin course of the time the “painful” process

of know-how transfer could become an issue.Moreover, for more hardware related aspects,it is well known that technology is advancingvery fast, and the core components chosen inthe early design days may soon become ob-solete yielding problems for system maintain-ability.

For all these reasons and looking ahead to-wards the E-ELT era, ESO, with the contribu-tion of INAF-OATs, started to investigate thepossible adoption of widely accepted COTS in-dustrial standards components with the aim oflowering the system procurement and mainte-nance costs as well as of mitigating obsoles-cence by improving components interoperabil-ity.

In this paper, after describing the outcomeof these studies, I’ll present the results appliedto the control of the forthcoming HiRes spec-trograph for the E-ELT. As a real, working ex-

Di Marcantonio: Industrial solutions trends for E-ELT 499

ample, I will also introduce the ESPRESSOcase, the new high-resolution spectrograph forthe VLT (Megevand et al. 2014) which couldbe seen as a test bed for the future HiRes in-strument.

2. The VLT case

The current VLT standard dedicated to the con-trol of hardware sub-systems of all the instru-ments, including telescopes in Paranal, is hea-vily based on the concept of a VME-basedLocal Control Unit (shortly, LCU). The VMEbus (which is, by the way, an industrial interna-tional standards) interconnects (specific) CPUboard with others dedicated to perform particu-lar tasks such as motion control, analog/digitalsignal I/Os etc. composing what is known as“the VME crate”. The actual limitation is thatthere is a strong software/hardware coupling inthe way ESO has implemented this architec-ture which prevents to easily change the ob-solete boards without significant changes inthe associated software. Several limits in termsof deployment and distribution also came outwhen trying to adapt this architecture to theESPRESSO case (and in perspective to theHiRes) which is inherently distributed due toits capability to use any of the 4 UTs.

The VME-based real-time control architec-ture concept has been used by ESO also tocontrol the technical CCDs (shortly, TCCDs)needed for stabilization or for secondary guid-ing purposes at the instrument level. The ESOcompliant TCCDs, although fulfilling the as-tronomical requirements, are however expen-sive in terms of costs, manufacturing, integra-tion and maintenance especially if comparedto available COTS devices one can find on themarket which led to seek for other solutions al-ready for the ESPRESSO case.

To overcome all these issues, ESO andINAF-OATs initiated several projects aiming atmodernize the VLT hardware/software controlsystem. The eventually identified technologiesare currently adopted in ESPRESSO and willvery likely constitute the core for the controlof future E-ELT instruments.

3. Industrial trends

The key point when analyzing industrial trendsfor the future E-ELT is the level of “standard-ization”. Only components adhering to widespread accepted standard (possibly, IEEE cer-tified) could overcome or mitigate the obsoles-cence guaranteeing interoperability and inter-changeability considering the long foreseentime span for the E-ELT instrumentation (atleast 10 years from now for first operations,followed by 20-30 years of lifetime). Basedon this, the proposed control of the futureHiRes spectrograph, follows standard indus-trial trends (where applicable) in the followingareas:

– fieldbus distribution;– time synchronization;– high performance camera communication;– unified architecture;– device control.

The proposed solutions are not only “paper-work”, but are currently being applied (for thefirst time at the VLT for a permanent instru-ment) to the ESPRESSO control case; addi-tional information could be found in (Baldiniet al. 2014) and references therein.

3.1. Fieldbus distribution

The current trend in control paradigm departsslightly from the VLT adopted one by super-seding the centralized vision and by puttingmore emphasis on distributed architecturesinterconnecting intelligent nodes (like PLCs,see Fig. 1). The interconnections could be

Fig. 1. Example of Beckhoff Automation PLCCPU and modules

500 Di Marcantonio: Industrial solutions trends for E-ELT

achieved by dedicated networks on which pro-prietary (or open) protocols run, as well as bydedicated networks based on standard Ethernetinfrastructure. For our applications, the lat-ter seems the most appropriate choice sincethis will allow to fully reuse future infras-tructure that will be certainly available at theE-ELT not imposing additional requirements.Example of real-time protocols that couldrun on such Ethernet-based fieldbus systemsare EtherCAT (standardized in IEC 61158)invented (and used by COTS provided) byBeckhoff Automation or PROFINET used bySiemens.

3.2. Time synchronization

Even though at the time of writing this pa-per no definitive requirements are imposedby E-ELT to external consortia involved inthe feasibility studies for the instrumentation,the time synchronization protocol is an excep-tion in this context. Several tests have beenmade in the past years and the final choiceis to rely, for time synchronization, on thePrecision Time Protocol (PTP, standardized asIEEE 1588), which allows to reach accuracyin the sub-microsecond range. Due to verypositive results, also VLT (and consequentlyESPRESSO) is going to adopt the same pro-tocol (superseding the current custom solu-tion available at Paranal). Both Siemens andBeckhoff Automation offer COTS componentsbased on this protocol (as well as other ven-dors) which guarantees interoperability andlong lifetime also looking ahead towards E-ELT instruments.

3.3. High performance cameracommunication

Triggered by the ESPRESSO requirements amarket survey was conducted to analyze cur-rent trends in CCD cameras communication.Again gigabit-Ethernet based communicationhas been an imposed initial strict requirement(as opposed to USB or Camera Link connec-tions) due to limits in cabling scheme andpossible re-use of existing network infrastruc-

ture. Among the very wide palette of pro-ducts found, the cameras supporting the GigEVision R© protocol introduced in 2006 by (andsupported by) the European Machine VisionAssociation presented a promising answer tocope both with high performance communica-tion needs, as well as interoperability and inter-changeability issues and thus could be seen asa valuable candidates for future E-ELT applica-tions; further details could be found in (Baldiniet al. 2014).

3.4. Unified architecture

OPC Unified architecture (OPC UA) is anindustrial (standard) communication protocoldeveloped by the OPC foundation aimed toconnect various kind of systems and devicesover TCP networks. The interoperability isachieved through a server-client approach: theserver is (usually) supplied by the hardwaremanufacturer and represents the interface be-tween internal proprietary protocols and OPCUA standard. Through the “server”, controlnodes are exposed to outside world in a stan-dardized way which could be than accessedby dedicated “clients”. The client implementsthe OPC UA communication by means ofAPIs and allows the user application to accessthe exposed nodes. By properly separating theclient logic from the communication protocol,this approach guarantees full interoperability:devices from different vendors could be seam-lessly exchanged without affecting high levelsoftware and, vice-versa, high level softwarecould be improved and/or changed without theneed to maintain the initial chosen hardware.This represent a big achievement if comparedto the existing Paranal VME world and it is atthe base of the chosen ESPRESSO control ar-chitecture (Baldini et al. 2014).

3.5. Device control

Among all the industrial trends analyzed in thispaper the control of the devices will eventuallydepends on the technological choices imposedby the E-ELT. Therefore it is still premature toforesee or propose a definitive choice for the

Di Marcantonio: Industrial solutions trends for E-ELT 501

Fig. 2. ESPRESSO distributed control architecture based on fieldbus interconnecting Beckhoff AutomationCOTS components

control of the vital parts of an instruments (likemotorized functions, lamps, sensors) consider-ing also that a design of HiRes does not existyet. Nowadays however, PLCs (ProgrammableLogic Controllers), are an example of widespread devices used in this field and representa valid choice also looking ahead. They are es-sentially a hard real-time systems where out-puts are produced based in response of inputconditions. The way outputs are done is fullyprogrammable and this offer a huge flexibilityallowing with the same PLC to control almostany functionality the system will require. PLCprogramming languages are standardized inIEC61131-3 open international standard; theirusage coupled with the PLCOpen motion con-trol standard specifications allows to write ef-ficient and flexible PLC code that is vendorand product-independent as has been proved inESPRESSO case (see Fig. 2).

4. From ESPRESSO to HiRes

All the technologies presented in this paper arecurrently in use in ESPRESSO which could

be reasonable seen as a precursor and test bedof HiRes@E-ELT. The main difference is thatin HiRes several spectrograph modules (in-stead of one) are foreseen, but for the rest theoverall conceptual design is quite similar: nomovable parts inside the modules to guaran-tee ultra-stability and complex front-end forproperly feeding the light. The ultimate les-son learned from ESPRESSO is that relying onCOTS solutions adhering to widespread indus-trial standards lowers procurement and mainte-nance costs and seems the correct way to mini-mize obsolescence by guaranteeing product in-teroperability and inter-changeability.

References

Baldini, V., et al. 2014, Proc. SPIE, 9152,915228

Duhoux, P., et al. 2014, Proc. SPIE, 9152,91520I

Ghedina, A., et al. 2014, Proc. SPIE, 9145,914553

Megevand, D., et al. 2014, Proc. SPIE, 9147,91471H