- Email: [email protected]
Data acquisition remote node powered over the communications optical fiber
Fusion Engineering and Design 96–97 (2015) 64–69
Contents lists available at ScienceDirect
Fusion Engineering and Design journal homepage: www.elsevier.com/locate/fusengdes
Data acquisition remote node powered over the communications optical fiber Antonio J.N. Batista ∗ , Jorge Sousa, Bruno Gonc¸alves Instituto de Plasmas e Fusão Nuclear, Instituto Superior Técnico, Universidade de Lisboa, 1049-001 Lisboa, Portugal
a r t i c l e
i n f o
Article history: Received 18 September 2014 Received in revised form 21 May 2015 Accepted 10 June 2015 Available online 21 June 2015 Keywords: Power-over-fiber Galvanic isolation Remote data acquisition Synchronous data acquisition Nuclear fusion High energy physics
a b s t r a c t Large nuclear fusion reactors, like ITER, will have harsh electromagnetic environments nearby the machine. Foreseeing the necessity for special data acquisition remote nodes, on difficult access locations and as close as possible to the experimental devices, motivated the system design. The architecture is based on the power-over-fiber technology recent advancements and respective implementation aim is to attain a proof of concept for the fusion technology field and others, e.g., high energy physics, industry, etc. The design intends the replacement of traditional copper cables and power supplies, vulnerable to electromagnetic interference, by the communications optical fiber of the data acquisition remote node. Optical fibers provide galvanic isolation, immunity to noisy electromagnetic environments and simultaneously can supply power to the remote node electronics. System architecture uses a laser power converter (array of photovoltaic cells) to convert the laser light, from the optical fiber, into electricity. The generated electrical power is enough for powering the remote node electronics and optoelectronics, such as an ADC, a low power FPGA and an optical transmitter. The laser power converter is also used as the communications receiver and from which the acquisition clock is recovered, providing synchronism between remote data acquisition nodes. Descriptions of the system architecture, tested implementations and future improvements are presented. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Commercial power-over-fiber (PoF) applications have started to be used mainly for remote monitoring of high voltage power networks [1]. PoF has evolved continuously through the last decades and the main components of such technology are Laser Power Converters (LPC), optical fibers and Vertical Cavity Surface Emitting Laser (VCSEL) diodes. State of the art LPCs [2] have optical to electrical power conversion efficiencies around 50% and are starting to have an affordable cost. This can lead to the LPCs usage in nuclear fusion or high energy physics. Unlike copper cables, optical fibers provide high voltage galvanic isolation, are immune to Electro-Magnetic Interference (EMI), provide high speed high distance communications and last, but not the least, optical fibers can carry energy. This mix of characteristics allows the implementation of remote data acquisition systems powered by laser light, avoiding the use of galvanic isolated power
∗ Corresponding author. Tel.: +351 218 419 113. E-mail address: [email protected] (A.J.N. Batista). http://dx.doi.org/10.1016/j.fusengdes.2015.06.048 0920-3796/© 2015 Elsevier B.V. All rights reserved.
supplies normally vulnerable to the high magnetic fields of larger tokamaks. Nowadays small laser diodes can generate enough photonic power to illuminate a LPC over long optical fibers. Also several laser diodes can be accommodated on printed circuit boards, with generous dimensions, as the ones of advanced instrumentation standards such as ATCA [3], MTCA.4 [4] or AXIe [5]. One possible application example, in the nuclear fusion arena, is data acquisition for the magnetics diagnostic of machines with long operation, like ITER. Magnetics diagnostics in larger tokamaks are normally connected to the data acquisition system by several copper cables and connectors across a long distance, which increases signal noise due to EMI and generates small signal offset voltages due to the metal junctions of cables and connectors. The metals junction voltage is temperature dependent (thermoelectric effect) and represents a serious problem on ITER, since magnetics diagnostic signals need to be integrated during the long operation time of the tokamak. The integration result will be affected by a huge error usually called drift [6]. Another problem is the necessity of galvanic isolation between signals normally accomplished using digital isolators and isolated DC–DC converters, as the ones used in the ATCA data acquisition boards at JET [7] or in the ATCA/AXIe boards for the ITER Fast Plant
A.J.N. Batista et al. / Fusion Engineering and Design 96–97 (2015) 64–69
65
Fig. 1. Simplified system architecture.
System Controller (FPSC) prototype [8]. Typical isolated DC–DC converters used in data acquisition systems are vulnerable to magnetic fields above ∼10 mT, starting to fail and not providing the needed reliability. The above mentioned problems can be diminished if a PoF data acquisition remote node is used, since copper cables distances and connectors number are significantly reduced between the system and the magnetic coils of the diagnostic. Also isolated DC–DC converters are not needed anymore and the system can be built with non-magnetic components and hard radiation ICs. The system prototype architecture, tests and future improvements are described in Sections 2–4, respectively. 2. System architecture Conceptually the system architecture is very simple and is depicted on Fig. 1, the real implementation is slightly more complex and is explained in Section 3. Essentially a laser sends light through a fiber optic to a LPC where is converted into electricity for powering the PoF remote acquisition node. The laser light is pulse modulated with a system clock, which is recovered from the LPC as the acquisition clock for the Analog to Digital Converter (ADC) of the remote node. The analog signal from the sensor/diagnostic is digitized by the ADC and sent to the host system. All PoF remote acquisition nodes are synchronous with the system clock.
Fig. 2. Prototype for testing different implementations.
3. Tested system implementations Data acquisition systems in harsh environments, where temperature, EMI and radiation are high, need to be as simple as possible to be reliable. The balance between functionality and reliability depends on the application. To verify the system architecture feasibility three different implementations (Fig. 2), with increasing complexity, were tested and are described, respectively, in Sections 3.2–3.4. Section 3.1 explains the LPC power and clock recovery circuit. 3.1. LPC power and clock recovery circuit To synchronize the data acquisition of all PoF remote nodes a system clock is sent to the LPC by laser pulse modulation (estimated clock frequencies up to ∼5 MHz, LPC and modulation technique dependent). The chosen ADC (ADS7945) is a low power device with 2 MSPS, 14 bits resolution and serial interface. The ADC needs a 2 MHz clock for pacing the acquisition and a 40 MHz clock to
Fig. 3. Circuit to recover the 2 MHz acquisition clock and power generation for the buck-boost converters.
generate the serial data. On all three different PoF remote node implementations the 40 MHz clock was generated from the 2 MHz acquisition clock recovered from the LPC. The circuit to recover the 2 MHz acquisition clock is shown on Fig. 3. A transimpedance amplifier based on the low power operational amplifier OPA836 combined with the low power comparator LMV7219, constitute a simple and efficient circuit to retrieve the clock from the LPC. The LPC (PPC-6E) can produce up to 500 mW of power and above 6 V of output voltage. To generate the required power rails for the Integrated Circuits (IC), of the PoF remote acquisition node, buckboost converters TPS63000 and TPS63060 have been utilized. Buck-boost converters have high efficiencies and allow the automatic generation of defined output voltages from higher or lower
66
A.J.N. Batista et al. / Fusion Engineering and Design 96–97 (2015) 64–69
Fig. 4. System implementation with PLLs on host and remote acquisition node.
input supply voltages. This aspect is essential since LPCs have an output voltage variation depending on the load.
3.2. Phase locked loops on host and remote node
Fig. 5. Acquisition clock, data clock and serial data at host side displayed on the oscilloscope.
Fig. 4 shows the specific system implementation diagram and experimental test setup. The ADC timing and interface signals (Fig. 6) need to be generated, by the remote node electronics, and be compliant with the ADC specifications. A programmable phase locked loop (PLL) IC (CY22381) was used to accomplish that task. The 2 MHz acquisition clock recovered from the LPC is used as the PLL reference clock. The PLL is programmed to have as output signals the ADC acquisition clock (2 MHz) and the ADC data clock (40 MHz). Finally the ADC serial data is sent by the laser diode of the remote node, over an optical fiber, to the photodiode of the host. On host side the same PLL IC and clock reference are used to replicate the acquisition clock and data clock. Fig. 5 depicts the measured signals.
Fig. 6. ADC timing and interface signals: 2 MHz acquisition clock (CS), 40 MHz data clock (SCLK) and serial data (SDO).
A.J.N. Batista et al. / Fusion Engineering and Design 96–97 (2015) 64–69
67
Fig. 7. System implementation with a Manchester decoder on host and encoder on remote node.
Remote node and host PLLs are synchronized but the solution has the inconvenient of requiring a phase adjustment between the data clock and serial data at host side (not done in this setup, the phase tuning depends of the optical fiber length and requires additional logic). Fig. 9. Remote node Manchester encoder.
3.3. Manchester decoder on host and encoder on remote node To avoid the phase adjustment, referred in Section 3.2, between the data clock and serial data signals at host side, the Manchester coding [9] can be used as a solution. Manchester encoders/decoders can be easily built with low power discrete logic and only a few components are needed for their implementation. Fig. 7 contains the required changes, relatively to Section 3.2, in order to include the Manchester coding. The host PLL was replaced by a Manchester decoder and a Manchester encoder was added to the remote acquisition node. Figs. 8 and 9 show, respectively, the decoder and encoder implementations. Low power discrete logic, XOR (NC7SZ86), Flip-flop (SN74LVC2G74) and Schmitt-trigger delay buffers (74LVC1G17) have been used.
Fig. 10. Acquisition clock, data clock and serial data at host side displayed on the oscilloscope.
Fig. 8. Host Manchester decoder.
The host measured signals with an oscilloscope are in Fig. 10 (the signals are noisier than the ones of the previous section since the Manchester decoder was implemented with discrete logic ICs).
68
A.J.N. Batista et al. / Fusion Engineering and Design 96–97 (2015) 64–69
Fig. 11. System implementation with a FPGA on remote node.
3.4. FPGA on remote node
4. Future system improvements and tests
When more flexibility is required on the remote node a low power Field Programmable Gate Array (FPGA) is necessary. The used FPGA was an AGLN125V2 with a programmable PLL inside. The PLL and discrete logic of the Manchester encoder on the remote node implementation (Section 3.3) was replaced by the FPGA, Fig. 11. Measured signals at the host, with an oscilloscope, are shown on Fig. 12. One drawback of using a FPGA is the increase of electronics complexity, which can reduce the system reliability.
The existing prototype implementations can be improved (Fig. 13) and new tests can be done in the future. The next sections will address some of those possibilities. 4.1. Super capacitor An application that consumes more power than the LPC can provide, but working in burst mode, can use the stored energy in a super-capacitor (high capacity and small dimensions). The supercapacitor will replace the capacitor C of Fig. 3 and will be charged between application bursts. 4.2. Operational amplifier and chopper circuit The analog front end of the remote node can be enhanced adding a chopper circuit followed by an operational amplifier before the ADC. The LPC power is enough to include more ICs. The chopper or equivalent solution is needed when the acquired signals are integrated over a long time. This way, electronic offsets introduced by the ICs between the chopper stage and the ADC are canceled. Details for chopper based integrators can be consulted in Ref. [6]. 4.3. Sending data to the remote node
Fig. 12. Acquisition clock, data clock and serial data at host side displayed on the oscilloscope.
If not only the clock but also data is needed to be sent to the remote node a Manchester encoder can be implemented to drive the laser modulator at host side. Afterwards a Manchester decoder implemented in the FPGA of the remote node recovers the clock and data sent by the host.
A.J.N. Batista et al. / Fusion Engineering and Design 96–97 (2015) 64–69
69
Fig. 13. Future improvements of the PoF remote data acquisition node.
4.4. Single fiber for full-duplex communications For long distances, between the host and remote node, using a single optical fiber to carry energy and full-duplex communications could decrease costs. A Wavelength Division Multiplexing (WDM) duplexer device to withstand power lasers must be developed.
require to be executed on a properly designed printed circuit board. 5. Conclusions
A local oscillator can be included in the design if the remote node does not need to be synchronized with the host or other remote nodes. The oscillator can also be included for the cases where synchronization is needed but the optical connection can be lost during short periods of time. The PLL inside the FPGA can be set to switch from the LPC recovered clock to the local oscillator clock when the optical connection is lost, and vice versa if the connection is restarted again.
The implemented prototype has demonstrated the feasibility of a PoF remote node, with synchronization between nodes as required by data acquisition systems of nuclear fusion experimental devices. The system could provide an alternative solution for data acquisition on magnetics diagnostic of large tokamaks. Straightforward ground loops elimination, high voltage galvanic isolation, EMI noise reduction and potentially the drift decrease of integrated signals, are benefits that worthwhile further system evolving. A new prototype with the discussed enhancements (Fig. 13) in Section 4 will be developed. Exhaustive testing is also mandatory to evaluate system performance and reliability.
4.6. Remote node and instrumentation standards
Acknowledgments
To process the data from the sensor/diagnostic and send it to the real time control or storage networks the remote node requires to be interfaced with an instrumentation standard such as the ATCA, MTCA.4, or AXIe.
This project has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement number 633053. IST activities also received financial support from “Fundac¸ão para a Ciência e Tecnologia” through project PestOE/SADG/LA0010/2014. The views and opinions expressed herein do not necessarily reflect those of the European Commission.
4.5. Remote node local oscillator
4.7. Tests Several tests need to be performed to evaluate the developed system. Important tests to do are, for example: jitter measurement of the recovered acquisition clock; communications bit error rate; maximum allowed distance for the optical fiber; power available on the remote node. Harsh environment tests are also needed to evaluate the remote node withstanding to temperature, high magnetic fields and neutron/gamma radiation. The environment tests results will dictate system improvements, like the use of radiation hard optical fibers and components, in order to mitigate possible problems such as optical fiber browning effects or FPGA Single Event Upsets (SEU). Based on an in-house handmade with wiring the prototype inevitably produced the noisy signals presented. The new tests
References [1] http://www.jdsu.com/productliterature/Intermountain-casestudy-FINAL.pdf [2] A.W. Bett, et al., III–V solar cells under monochromatic illumination, in: 33rd IEEE Photovoltaic Specialists Conference, 2008. [3] http://www.picmg.org/openstandards/advancedtca/ [4] http://www.picmg.org/openstandards/microtca/ [5] http://www.axiestandard.org/specifications.html [6] A. Werner, W7-X magnetic diagnostics: performance of the digital integrator, Rev. Sci. Instrum. 77 (2006) 10E307. [7] Antonio Batista, et al., ATCA control system hardware for the plasma vertical stabilization in the JET tokamak, IEEE Trans. Nucl. Sci. 57 (2) (2010) 583–588. [8] A.J.N. Batista, et al., Control and data acquisition ATCA/AXIe board designed for high system availability and reliability of nuclear fusion experiments, Fusion Eng. Des. 88 (2013) 1332–1337. [9] R. Forster, Manchester encoding: opposing definitions resolved, Eng. Sci. Educ. J. 9 (6) (2000) 278–280.
Contents lists available at ScienceDirect
Fusion Engineering and Design journal homepage: www.elsevier.com/locate/fusengdes
Data acquisition remote node powered over the communications optical fiber Antonio J.N. Batista ∗ , Jorge Sousa, Bruno Gonc¸alves Instituto de Plasmas e Fusão Nuclear, Instituto Superior Técnico, Universidade de Lisboa, 1049-001 Lisboa, Portugal
a r t i c l e
i n f o
Article history: Received 18 September 2014 Received in revised form 21 May 2015 Accepted 10 June 2015 Available online 21 June 2015 Keywords: Power-over-fiber Galvanic isolation Remote data acquisition Synchronous data acquisition Nuclear fusion High energy physics
a b s t r a c t Large nuclear fusion reactors, like ITER, will have harsh electromagnetic environments nearby the machine. Foreseeing the necessity for special data acquisition remote nodes, on difficult access locations and as close as possible to the experimental devices, motivated the system design. The architecture is based on the power-over-fiber technology recent advancements and respective implementation aim is to attain a proof of concept for the fusion technology field and others, e.g., high energy physics, industry, etc. The design intends the replacement of traditional copper cables and power supplies, vulnerable to electromagnetic interference, by the communications optical fiber of the data acquisition remote node. Optical fibers provide galvanic isolation, immunity to noisy electromagnetic environments and simultaneously can supply power to the remote node electronics. System architecture uses a laser power converter (array of photovoltaic cells) to convert the laser light, from the optical fiber, into electricity. The generated electrical power is enough for powering the remote node electronics and optoelectronics, such as an ADC, a low power FPGA and an optical transmitter. The laser power converter is also used as the communications receiver and from which the acquisition clock is recovered, providing synchronism between remote data acquisition nodes. Descriptions of the system architecture, tested implementations and future improvements are presented. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Commercial power-over-fiber (PoF) applications have started to be used mainly for remote monitoring of high voltage power networks [1]. PoF has evolved continuously through the last decades and the main components of such technology are Laser Power Converters (LPC), optical fibers and Vertical Cavity Surface Emitting Laser (VCSEL) diodes. State of the art LPCs [2] have optical to electrical power conversion efficiencies around 50% and are starting to have an affordable cost. This can lead to the LPCs usage in nuclear fusion or high energy physics. Unlike copper cables, optical fibers provide high voltage galvanic isolation, are immune to Electro-Magnetic Interference (EMI), provide high speed high distance communications and last, but not the least, optical fibers can carry energy. This mix of characteristics allows the implementation of remote data acquisition systems powered by laser light, avoiding the use of galvanic isolated power
∗ Corresponding author. Tel.: +351 218 419 113. E-mail address: [email protected] (A.J.N. Batista). http://dx.doi.org/10.1016/j.fusengdes.2015.06.048 0920-3796/© 2015 Elsevier B.V. All rights reserved.
supplies normally vulnerable to the high magnetic fields of larger tokamaks. Nowadays small laser diodes can generate enough photonic power to illuminate a LPC over long optical fibers. Also several laser diodes can be accommodated on printed circuit boards, with generous dimensions, as the ones of advanced instrumentation standards such as ATCA [3], MTCA.4 [4] or AXIe [5]. One possible application example, in the nuclear fusion arena, is data acquisition for the magnetics diagnostic of machines with long operation, like ITER. Magnetics diagnostics in larger tokamaks are normally connected to the data acquisition system by several copper cables and connectors across a long distance, which increases signal noise due to EMI and generates small signal offset voltages due to the metal junctions of cables and connectors. The metals junction voltage is temperature dependent (thermoelectric effect) and represents a serious problem on ITER, since magnetics diagnostic signals need to be integrated during the long operation time of the tokamak. The integration result will be affected by a huge error usually called drift [6]. Another problem is the necessity of galvanic isolation between signals normally accomplished using digital isolators and isolated DC–DC converters, as the ones used in the ATCA data acquisition boards at JET [7] or in the ATCA/AXIe boards for the ITER Fast Plant
A.J.N. Batista et al. / Fusion Engineering and Design 96–97 (2015) 64–69
65
Fig. 1. Simplified system architecture.
System Controller (FPSC) prototype [8]. Typical isolated DC–DC converters used in data acquisition systems are vulnerable to magnetic fields above ∼10 mT, starting to fail and not providing the needed reliability. The above mentioned problems can be diminished if a PoF data acquisition remote node is used, since copper cables distances and connectors number are significantly reduced between the system and the magnetic coils of the diagnostic. Also isolated DC–DC converters are not needed anymore and the system can be built with non-magnetic components and hard radiation ICs. The system prototype architecture, tests and future improvements are described in Sections 2–4, respectively. 2. System architecture Conceptually the system architecture is very simple and is depicted on Fig. 1, the real implementation is slightly more complex and is explained in Section 3. Essentially a laser sends light through a fiber optic to a LPC where is converted into electricity for powering the PoF remote acquisition node. The laser light is pulse modulated with a system clock, which is recovered from the LPC as the acquisition clock for the Analog to Digital Converter (ADC) of the remote node. The analog signal from the sensor/diagnostic is digitized by the ADC and sent to the host system. All PoF remote acquisition nodes are synchronous with the system clock.
Fig. 2. Prototype for testing different implementations.
3. Tested system implementations Data acquisition systems in harsh environments, where temperature, EMI and radiation are high, need to be as simple as possible to be reliable. The balance between functionality and reliability depends on the application. To verify the system architecture feasibility three different implementations (Fig. 2), with increasing complexity, were tested and are described, respectively, in Sections 3.2–3.4. Section 3.1 explains the LPC power and clock recovery circuit. 3.1. LPC power and clock recovery circuit To synchronize the data acquisition of all PoF remote nodes a system clock is sent to the LPC by laser pulse modulation (estimated clock frequencies up to ∼5 MHz, LPC and modulation technique dependent). The chosen ADC (ADS7945) is a low power device with 2 MSPS, 14 bits resolution and serial interface. The ADC needs a 2 MHz clock for pacing the acquisition and a 40 MHz clock to
Fig. 3. Circuit to recover the 2 MHz acquisition clock and power generation for the buck-boost converters.
generate the serial data. On all three different PoF remote node implementations the 40 MHz clock was generated from the 2 MHz acquisition clock recovered from the LPC. The circuit to recover the 2 MHz acquisition clock is shown on Fig. 3. A transimpedance amplifier based on the low power operational amplifier OPA836 combined with the low power comparator LMV7219, constitute a simple and efficient circuit to retrieve the clock from the LPC. The LPC (PPC-6E) can produce up to 500 mW of power and above 6 V of output voltage. To generate the required power rails for the Integrated Circuits (IC), of the PoF remote acquisition node, buckboost converters TPS63000 and TPS63060 have been utilized. Buck-boost converters have high efficiencies and allow the automatic generation of defined output voltages from higher or lower
66
A.J.N. Batista et al. / Fusion Engineering and Design 96–97 (2015) 64–69
Fig. 4. System implementation with PLLs on host and remote acquisition node.
input supply voltages. This aspect is essential since LPCs have an output voltage variation depending on the load.
3.2. Phase locked loops on host and remote node
Fig. 5. Acquisition clock, data clock and serial data at host side displayed on the oscilloscope.
Fig. 4 shows the specific system implementation diagram and experimental test setup. The ADC timing and interface signals (Fig. 6) need to be generated, by the remote node electronics, and be compliant with the ADC specifications. A programmable phase locked loop (PLL) IC (CY22381) was used to accomplish that task. The 2 MHz acquisition clock recovered from the LPC is used as the PLL reference clock. The PLL is programmed to have as output signals the ADC acquisition clock (2 MHz) and the ADC data clock (40 MHz). Finally the ADC serial data is sent by the laser diode of the remote node, over an optical fiber, to the photodiode of the host. On host side the same PLL IC and clock reference are used to replicate the acquisition clock and data clock. Fig. 5 depicts the measured signals.
Fig. 6. ADC timing and interface signals: 2 MHz acquisition clock (CS), 40 MHz data clock (SCLK) and serial data (SDO).
A.J.N. Batista et al. / Fusion Engineering and Design 96–97 (2015) 64–69
67
Fig. 7. System implementation with a Manchester decoder on host and encoder on remote node.
Remote node and host PLLs are synchronized but the solution has the inconvenient of requiring a phase adjustment between the data clock and serial data at host side (not done in this setup, the phase tuning depends of the optical fiber length and requires additional logic). Fig. 9. Remote node Manchester encoder.
3.3. Manchester decoder on host and encoder on remote node To avoid the phase adjustment, referred in Section 3.2, between the data clock and serial data signals at host side, the Manchester coding [9] can be used as a solution. Manchester encoders/decoders can be easily built with low power discrete logic and only a few components are needed for their implementation. Fig. 7 contains the required changes, relatively to Section 3.2, in order to include the Manchester coding. The host PLL was replaced by a Manchester decoder and a Manchester encoder was added to the remote acquisition node. Figs. 8 and 9 show, respectively, the decoder and encoder implementations. Low power discrete logic, XOR (NC7SZ86), Flip-flop (SN74LVC2G74) and Schmitt-trigger delay buffers (74LVC1G17) have been used.
Fig. 10. Acquisition clock, data clock and serial data at host side displayed on the oscilloscope.
Fig. 8. Host Manchester decoder.
The host measured signals with an oscilloscope are in Fig. 10 (the signals are noisier than the ones of the previous section since the Manchester decoder was implemented with discrete logic ICs).
68
A.J.N. Batista et al. / Fusion Engineering and Design 96–97 (2015) 64–69
Fig. 11. System implementation with a FPGA on remote node.
3.4. FPGA on remote node
4. Future system improvements and tests
When more flexibility is required on the remote node a low power Field Programmable Gate Array (FPGA) is necessary. The used FPGA was an AGLN125V2 with a programmable PLL inside. The PLL and discrete logic of the Manchester encoder on the remote node implementation (Section 3.3) was replaced by the FPGA, Fig. 11. Measured signals at the host, with an oscilloscope, are shown on Fig. 12. One drawback of using a FPGA is the increase of electronics complexity, which can reduce the system reliability.
The existing prototype implementations can be improved (Fig. 13) and new tests can be done in the future. The next sections will address some of those possibilities. 4.1. Super capacitor An application that consumes more power than the LPC can provide, but working in burst mode, can use the stored energy in a super-capacitor (high capacity and small dimensions). The supercapacitor will replace the capacitor C of Fig. 3 and will be charged between application bursts. 4.2. Operational amplifier and chopper circuit The analog front end of the remote node can be enhanced adding a chopper circuit followed by an operational amplifier before the ADC. The LPC power is enough to include more ICs. The chopper or equivalent solution is needed when the acquired signals are integrated over a long time. This way, electronic offsets introduced by the ICs between the chopper stage and the ADC are canceled. Details for chopper based integrators can be consulted in Ref. [6]. 4.3. Sending data to the remote node
Fig. 12. Acquisition clock, data clock and serial data at host side displayed on the oscilloscope.
If not only the clock but also data is needed to be sent to the remote node a Manchester encoder can be implemented to drive the laser modulator at host side. Afterwards a Manchester decoder implemented in the FPGA of the remote node recovers the clock and data sent by the host.
A.J.N. Batista et al. / Fusion Engineering and Design 96–97 (2015) 64–69
69
Fig. 13. Future improvements of the PoF remote data acquisition node.
4.4. Single fiber for full-duplex communications For long distances, between the host and remote node, using a single optical fiber to carry energy and full-duplex communications could decrease costs. A Wavelength Division Multiplexing (WDM) duplexer device to withstand power lasers must be developed.
require to be executed on a properly designed printed circuit board. 5. Conclusions
A local oscillator can be included in the design if the remote node does not need to be synchronized with the host or other remote nodes. The oscillator can also be included for the cases where synchronization is needed but the optical connection can be lost during short periods of time. The PLL inside the FPGA can be set to switch from the LPC recovered clock to the local oscillator clock when the optical connection is lost, and vice versa if the connection is restarted again.
The implemented prototype has demonstrated the feasibility of a PoF remote node, with synchronization between nodes as required by data acquisition systems of nuclear fusion experimental devices. The system could provide an alternative solution for data acquisition on magnetics diagnostic of large tokamaks. Straightforward ground loops elimination, high voltage galvanic isolation, EMI noise reduction and potentially the drift decrease of integrated signals, are benefits that worthwhile further system evolving. A new prototype with the discussed enhancements (Fig. 13) in Section 4 will be developed. Exhaustive testing is also mandatory to evaluate system performance and reliability.
4.6. Remote node and instrumentation standards
Acknowledgments
To process the data from the sensor/diagnostic and send it to the real time control or storage networks the remote node requires to be interfaced with an instrumentation standard such as the ATCA, MTCA.4, or AXIe.
This project has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement number 633053. IST activities also received financial support from “Fundac¸ão para a Ciência e Tecnologia” through project PestOE/SADG/LA0010/2014. The views and opinions expressed herein do not necessarily reflect those of the European Commission.
4.5. Remote node local oscillator
4.7. Tests Several tests need to be performed to evaluate the developed system. Important tests to do are, for example: jitter measurement of the recovered acquisition clock; communications bit error rate; maximum allowed distance for the optical fiber; power available on the remote node. Harsh environment tests are also needed to evaluate the remote node withstanding to temperature, high magnetic fields and neutron/gamma radiation. The environment tests results will dictate system improvements, like the use of radiation hard optical fibers and components, in order to mitigate possible problems such as optical fiber browning effects or FPGA Single Event Upsets (SEU). Based on an in-house handmade with wiring the prototype inevitably produced the noisy signals presented. The new tests
References [1] http://www.jdsu.com/productliterature/Intermountain-casestudy-FINAL.pdf [2] A.W. Bett, et al., III–V solar cells under monochromatic illumination, in: 33rd IEEE Photovoltaic Specialists Conference, 2008. [3] http://www.picmg.org/openstandards/advancedtca/ [4] http://www.picmg.org/openstandards/microtca/ [5] http://www.axiestandard.org/specifications.html [6] A. Werner, W7-X magnetic diagnostics: performance of the digital integrator, Rev. Sci. Instrum. 77 (2006) 10E307. [7] Antonio Batista, et al., ATCA control system hardware for the plasma vertical stabilization in the JET tokamak, IEEE Trans. Nucl. Sci. 57 (2) (2010) 583–588. [8] A.J.N. Batista, et al., Control and data acquisition ATCA/AXIe board designed for high system availability and reliability of nuclear fusion experiments, Fusion Eng. Des. 88 (2013) 1332–1337. [9] R. Forster, Manchester encoding: opposing definitions resolved, Eng. Sci. Educ. J. 9 (6) (2000) 278–280.