CERN Accelerating science

Optical Data Transmission for LHC Experiment Phase-2 Upgrades

Optical Readout and Control Systems have become ubiquitous with the LHC-era of Particle Physics Experiments, with each generation of detector upgrades bringing a need to transfer ever-greater amounts of data.  Just like in our everyday lives where we now take for granted the ability to watch movies on our mobile phones, so the particle physicist takes for granted the availability of technology to allow the collection of more and more data that can be analysed to detect the faintest of signals. Each generation of experiment challenges us developers to conceive faster, smaller, and more radiation tolerant opto-electronic systems.

The Versatile Link Plus (VL+) project has developed an optical data transmission system capable of transferring up to 10 Gb/s of data to and from the innermost detectors that must sustain the highest radiation levels during operation of the HL-LHC.  These optical links will be deployed in all of the so-called Phase-2 Upgrades that the LHC experiments are preparing in order to exploit the physics potential of the increased luminosity that the HL-LHC will provide. The VLproject is a collaboration between CERN, Fermilab, Southern Methodist University, and the University of Oxford. 

Our team within the EP-ESE group is responsible for the development of the optical transceiver that converts the electrical data produced by the detectors to optical signals that can be transmitted over 50-100 m of optical fibre to the shielded underground control rooms for further processing. The optical transceiver module (VTRx+), shown in the photograph below, is a miniaturised object measuring only 10 mm x 20 mm and is below 2.5 mm in height. It is light enough to be placed throughout the future pixel and tracker detectors of ATLAS and CMS. Radiation tolerance is also a key parameter for any component for future use inside the HL-LHC detectors and this has been designed into the components and assembly of the VTRx+.

VTRx+ optical transceiver module (Credits: CERN)

The VTRx+ module contains four optical transmitters: Vertical Cavity Surface Emitting Lasers (VCSELs) that convert electrical inputs, driven by a custom-designed laser driver ASIC, to modulated light. The VTRxmodule also contains one optical receiver: a photodiode and custom transimpedance amplifier (TIA) to turn the optical input into an electrical signal for the control of the detector. The completed module with its optical fibre pigtail is able to operate in the intense magnetic- and radiation fields that will be encountered in the HL-LHC experiments. The strongest magnetic field present in the HL-LHC detectors is the 4T field of the CMS Solenoid, which sets the tolerance limit for the VTRx+ module. The radiation levels used to qualify the VTRx+ module are 1 MGy of ionising dose, 1015 neutrons/cm2 and 1015 hadrons/cm2. Over twenty years of studying the effect of radiation from different particle species and energies on optoelectronic components allow us to confidently simulate the radiation field in the HL-LHC detectors using a single radiation source. Thus one radiation test to a total fluence of 3x1015 20 MeV neutrons/cm2 available at the Cyclotron facility of the Université Catholique de Louvain-la-Neuve in Belgium is enough to qualify the modules and components, which greatly simplifies the problem.

An extensive multi-year development programme has allowed us to both select suitable components and refine the module design to meet the requirements. We have evaluated VCSELs and photodiodes from multiple vendors in the framework of a CERN Market Survey that has culminated in the recent signature of contracts worth a little over 1 MCHF. Careful evaluation of device performance was necessary including environmental testing over the full specification range from -35 to +60 °C, as well as multiple radiation tests. The VCSEL and photodiode production samples for final validation of their radiation tolerance are scheduled for delivery in the first Quarter of 2020 together with the first large batches of these components. We must validate every VCSEL and photodiode production wafer because these are commercial parts that offer no guarantee of tolerance to our harsh environment and we must eliminate the very small probability that minor process improvements at the manufacturer lead to changes in the excellent radiation tolerance that we have observed over many years of testing.

The ASIC chipset used in the VTRx+ consists of a quad channel VCSEL driver and a single channel transimpedance amplifier (TIA) for the photodiode signals. Both designs have been made in standard CMOS processes using special design techniques to increase the overall radiation tolerance of the circuits.  The driver has just passed its final design review and has been submitted for fabrication in a 65 nm process, while the TIA has already been produced in a 130 nm process and is currently undergoing wafer-level testing.

The VTRx+  module design has passed through ten iterations to evaluate different configurations of electrical- and optical connection, in particular the method for attaching the optical fibre. This careful evaluation has allowed us to settle on an extremely thin (2.5 mm) module that fits the space constraints of our most challenging detector applications. The final thickness of the VTRx+  is to be compared to the typical thickness of 10 mm of our previous generation of optical transceiver module that is currently being deployed in the ALICE, ATLAS, CMS, and LHCb Phase-1 upgrades. A very important step in the VTRx+  project was the decision taken late last year to pursue the production of a CERN module design over the purchasing of a commercial design. This decision will lead to significant cost savings to the experiments while maintaining the high performance of the module.

We have just completed the assembly of the first large batch of a little over 350 prototype VTRx+ modules following an extensive survey of companies to find those capable of handing the very tight assembly tolerances (+/- 5 µm) required to achieve good performance. The CERN-designed modules were produced by a German industrial partner, one of the very few European companies identified during our survey. While this first large batch of prototypes was produced in Germany, their fibre pigtails were attached at CERN where we also carried out the testing. This allowed us to trial and refine our test methods and setups in preparation for transferring them to the industrial partner for the series production. Although we had produced tens of prototypes at a time while iterating on the design, this was also the first time we had significant statistics from testing to validate our 90-95% yield target for the final assembly of around 60k modules that will begin in 2020. Although we were just shy of our target for this batch, discussions with the industrial partner have identified some easy to implement process improvements that we are confident will allow us to reach our yield target in production. The modules have now been distributed to end users in ATLAS and CMS for inclusion in system level tests.

The VTRx+  module production will start in the first half of 2020. A first pre-series will be produced in order to fine-tune the assembly and testing process at the manufacturer. We will then carry out the full qualification of the modules before giving the green light to start the series production of approximately 60000 modules at the end of 2020. We are looking forward to what promises to be another busy year!