Radiation Test Report Paul Scherer Institute Proton Irradiation Facility

Transcription

1 the Large Hadron Collider project CERN CH-2 Geneva 23 Switzerland CERN Div./Group RadWG EDMS Document No. xxxxx Radiation Test Report Paul Scherer Institute Proton Irradiation Facility Responsibility Tested by: Giovanni SPIEZIA / Paul PERONNARD Group: EN-STI-ECE Date start: 8/03/20 Date end: 20/03/20 Equipment type concerned: Power Converter DUT identification and operating conditions DUT name DUT type Voltage Lot code INA4 OpAmp +/-5V Unknown DUT name DUT type Voltage Lot code OPA2227_ OpAmp ( st board) +/-5V OPA2227_2 OpAmp (2 nd board) +/-5V and +/-8V Unknown Unknown DUT name DUT type Voltage Lot Code TL072C OpAmp +/-5V Unknown DUT name DUT type Input Voltage Output tolerance Lot code TL43 Voltage Reference +5V 2% Unknown DUT name DUT type Input Voltage Output tolerance Lot code TL432 Voltage Reference +5V 0.5% Unknown DUT name DUT type Voltage Output tolerance Lot code LM404 Voltage Reference +5V 0.% Unknown

2 Page 2 of 35 Description of the test setup All the devices share the same setup for the monitoring of Single Event Latchup and Single Event Transient. A Tektronix DPO7254 oscilloscope (2.5GHz, 40GS/s) was connected to the output of the Device Under Test via a 50 ohms coaxial cable. The oscilloscope was configured in glitch trigger mode. Either a rising or falling edge can be detected by the trigger. The oscilloscope is remotely controlled by its Ethernet interface. The input impedance of the scope was set to M ohms. Most of the tested components cannot drive a 50 Ohm load. To avoid problems of contamination, the scope was located in the maze of the bunker. Figure : Trigger configuration panel A BNC cable of 0 meters was used to connect the input of the instrument to the output of the DUT. Figure 2 shows the response of the test apparatus to a pulse with duration of 5ns and amplitude of 00mV. The pulse was generated by a Stanford DG535. The horizontal scale is 20ns and the vertical one is 20mV. The observed response shows that the test setup is able to detect a SET similar to the test pulse. Table gives the configuration of the trigger for the different irradiation sessions. The width and the level of the glitches were changed to increase the possibility to detect Single Event Transients. Figure 2: Observed pulse with 0 meters of cable

3 Page 3 of 35 Table : Trigger setup Run # DUT id Width (ns) Level (mv) TL TL TL TL LM LM INA INA OPA2227_ OPA2227_ 5 20 TL OPA2227_ OPA2227_ The DUT is powered with a dual output DC power supply (Agilent E3648A). The currents are monitored by the power supply itself. In case of a SEL, an event is generated and stored with the Graphical User Interface via a GPIB link. The threshold level to detect a SEL was set to 50 ma for the positive and negative outputs. The table given below shows the nominal current consumption for each device for both positive and negative supplies. Table 2: Current consumption of the tested boards DUT id Current (ma) INA OPA TL LM404 + NA

4 Page 4 of 35 TL43 + NA TL432 + NA The output of each DUT was monitored with a Data Acquisition/Switch Unit Agilent 34970A. Each channel is sampled sequentially by the software. A delay of 00ms was inserted between channel acquisition to allow a good settling time for the signals and to not stress the multiplexer of the DAQ (relays). Each printed circuit board supports 6 chips for the voltage references and 3 chips for the amplifiers. It is noted that two amplifiers are available in a single chip for the TL072/OPA2227. All the devices, apart the INA4, are biased as suggested by the Power Converter group, according to the usage of the device. The INA4, a differential amplifier, is configured with a gain of 0. The output equals 3.5V as the differential input voltage is set to 0.35V by the resistor dividers. For the other amplifiers, TL072 and OPA2227, there are three configurations: (i) first amplifier gain 2 and second amplifier mounted as a follower; (ii) first amplifier with gain -2 and second amplifier mounted as a follower; (iii) first amplifier with gain - and second amplifier mounted as a follower. Figure 3: Bias condition for TL43/TL432

5 Page 5 of 35 Figure 4: Bias condition for LM404 Figure 5: Bias conditions for INA4

6 Page 6 of 35 Figure 6: Test circuit for OPA2227/TL072

7 Page 7 of 35 All the instruments were controlled by the GUI. The GUI (written in C#) stores the current consumptions for all power supplies, set the input voltage for the op-amps and get the information (duration, amplitude) on the detected SETs. Figure 7: Overview of the GUI

8 Page 8 of 35 Beam conditions The tests have been carried out at the Proton Irradiation Facility (PIF) of PSI. The initial proton beam for PIF is delivered from the PROSCAN accelerator with the help of the primary energy degrader, which allows setting a few discrete initial beam energies in the range from 250 MeV down to 30 MeV. The beam is subsequently guided to the Experimental Area where the PIF facility is located. A further set of collimators is placed before the DUT to set lower energy. For the above test, the primary energy was set to 230 MeV. The beam flux is measured by means of two ionization chambers IC and IC2 which are placed upstream and downstream the collimators, respectively. The counting of the IC and IC2 are calibrated to measure the beam flux at the position of the DUT, which is cm far away the IC2 (downstream the collimator). That operation is done before the test by comparing the counting of the IC and IC2 with respect to the output of a scintillator detector which is mounted where the DUT will be placed ( cm from the IC2). The TID is provided by the facility and is given in Gy (Si). The table 3 reports the beam conditions, specifying the final total fluence and TID for each run. Run # DUT id Input Voltage TL43 NA 230 Table 3: Beam conditions Energy [MeV] Flux Fluence (p/cm 2 ) TID (Gy).65E E+ 2 TL43 NA E+08.70E TL43 NA E+08 4 TL432 NA E Dose rate (Gy/h) E E LM404 NA E E LM404 NA 230 NA NA NA INA4 300mV E E INA4 300mV OPA2227_ 4V OPA2227_ 4V E E E+08.00E E E TL072 4V E E OPA2227_2 V E+08.90E OPA2227_2 0.5V E E

9 Page 9 of 35 The dose rate is higher than the one expected for the LHC tunnel (tenths of Gy per year) which would be unpractical to test. Therefore, ELDR (Enhanced Low Dose Rate) effects might not be correctly evaluated. A TID, 4 times higher than the one fixed as a target (50 Gy) was reached in order to compensate for possible ELDR effect. Observed failure modes No latchup, with current higher than 50 ma, and no SET, larger than 00 mv and longer than 5 ns, were observed during the whole experiment. Cumulative effect results Table 4 gives the cumulative TID for each tested device. The effects of the TID on the outputs and current consumption are given in the subsections below for each tested device. Table 4: Cumulative TID for each DUT DUT id Cumulative TID (Gy) INA4 220 OPA2227_ 200 OPA2227_2 29 TL LM TL TL INA4 Figures depicts the outputs of INA4 as a function of the accumulated dose. Runs 7 and 8 are joined together. A warm up period always precedes the test under beam.

10 Page 0 of 35 Figure 8: Drift observed on the st output of INA4 Figure 9: Drift observed on the 2 nd output of INA4 Figure 0: Drift observed on the 3rd output of INA4 For a gain G=0, considering the amplifier offset (±00 µv), and the gain accuracy (±0.05%), the amplifier output has to remain stable within a band gap of about ±2 mv. The obtained results show that no significant drift is observable up to about 00Gy. From that point, the output voltage of the amplifier starts to increase and it is out of the specifications at about 30 Gy. The discontinuities located at around 70Gy are artefacts due to a power cycling and should be ignored. Table 5 gives some noticeable measurement points regarding the accumulated dose while table 6 gives the drift values. The power consumption remained stable for the whole experiment.

11 Page of 35 Table 5: Noticeable measurement points for INA4 TID (Gy) Output (V) Output 2 (V) Output 3 (V) Table 6: Measured drift on the outputs of INA4 TID (Gy) Drift (mv) Drift 2 (mv) Drift 3 (mv) OPA2227_ Figures 0-5 shows the different outputs for the first OPA2227 board. Run 9 and 0 are joined for each output. Run 0 starts at 53Gy. A power cycling is performed between the 2 runs. About 5 minutes separate the two runs. Figure : Output of OPA2227_ for runs 9-0. Gain2

12 Page 2 of 35 Figure 2: Output 2 of OPA2227_ for runs 9-0. Gain2, buffer output. Figure 3: Output 3 of OPA2227_ for runs 9-0. Gain -2

13 Page 3 of 35 Figure 4: Output 4 of OPA2227_ for run 9. Gain -2, output buffer Figure 5: Output 4 of OPA2227_ for run 0. Gain -2, output buffer.

14 Page 4 of 35 Figure 6: Output 5 of OPA2227_ for runs 9-0. Gain -. Figure 7: Output 6 of OPA2227_ for runs 9-0. Gain -, output buffer. Giving the offset voltage of the amplifier and its uncertainty, the output has to remain stable within ± mv. The drift on the outputs is less than mv excepted for outputs 3 and 4. The drift measured during the run 9 on the output 3 is 3.5mV while the drift on output 4 is -2.7mV. For run 0, the drift on the outputs are quite the same and below mv except for the output 4 where it is 2.4mV. The jump between two runs is likely due to the measurement setup and should not be considered. Therefore, one can conclude that this amplifier remains within its specifications up to 200 Gy. Regarding the power supplies, a decrease of about 5 ma of the power consumption was observed on both positive and negative outputs. Figure 6 depicts the phenomenon. It should be noted that the plot shows the current of the whole board with 6 amplifiers and a single one consumes between 3.7 and 4.0 ma.

15 Page 5 of 35 Therefore, the current drift is out of the specifications of about 3 ma. This effect did not affect at all the functioning of the device during the test. Figure 8: Drift observed on the power consumption of the OPA2227_ board. OPA2227_2 V input Figures 9 to 24 depict the behaviour of the outputs while the input is set to V. There is no evidence of drift unlike the previous board with the input set to 4V. Figure9: OPA2227_2 Output. Gain 2.

16 Page 6 of 35 Figure20: OPA2227_2 Output 2. Gain 2, output buffer. Figure2: OPA2227_2 Output 3. Gain -2.

17 Page 7 of 35 Figure22: OPA2227_2 Output 4. Gain -2, output buffer. Figure23: OPA2227_2 Output 5. Gain -.

18 Page 8 of 35 Figure24: OPA2227_2 Output 6. Gain -, output buffer. OPA2227_2 0.5V input Figures 25 to 30 show the behaviour of the outputs while the input is set to 0.5V. Like previous run, no drift could be put in evidence. The amplitude of the bump located between 23-57Gy is less than mv. Figure25: OPA2227_2 Output. Gain 2.

19 Page 9 of 35 Figure 26: OPA2227_2 Output 2. Gain 2, output buffer. Figure27: OPA2227_2 Output 3. Gain -2

20 Page 20 of 35 Figure28: OPA2227_2 Output 4. Gain -2, output buffer. Figure29: OPA2227_2 Output 5. Gain -

21 Page 2 of 35 Figure30: OPA2227_2 Output 6. Gain - output buffer Like the first board, the components remain in the specification given by the datasheet up to 200 Gy. 00 Gy were cumulated with input V and 00 Gy with input 0.5 V. Regarding the power supplies, a decrease of about ma of the power consumption was observed on both the negative and the positive outputs. It seems that the current consumption depends on the output voltage; in fact the decrease is less evident on the board OPA2227_2 which was fed with V and 0.5 V input voltage. TL072 Figures 3-36 show the different outputs for the TL072 board. One should note the step occurring at around 34Gy on all the outputs. The drift on outputs and 2 (+8V) is pretty low (500uV for output 2). The outputs 3 and 4 (-8.8V) show a drift of 3.2mV before 34Gy and remain stables after this TID value. The drift on outputs 5 and 6 is about mv before 34Gy and 2mV between 34Gy and 200Gy. Unlike the previous run, there was no power cycling during the run. The drift of the output can be considered within the specifications given by the datasheet (max input offset voltage: 0mV, 3mV typical).

22 Page 22 of 35 Figure 9: TL072 - Output. Gain 2 Figure 20: TL072 Output 2. Gain 2, output buffer.

23 Page 23 of 35 Figure 2: TL072 Output 3. Gain -2 Figure 22: TL072 Output 4. Gain -2, output buffer

24 Page 24 of 35 Figure 23: TL072 Output 5. Gain -. Figure 24: TL072 Output 6. Gain -, output buffer.

25 Page 25 of 35 Figure 25: TL072 Current consumption The TL072 board has shown a drift on the power supplies. Regarding the power supplies, a decrease of about 2 and 3 ma of the power consumption was observed on the negative and the positive outputs, respectively. Figure 37 depicts the phenomenon. It should be noted that the plot shows the current of the whole board with 6 amplifiers and a single one consumes between and 2 ma with no load and output equal to 0. Therefore, the current drift could be considered within the specifications. This effect did not affect at all the functioning of the device during the test. LM404 Figures show the outputs of the voltage references obtained during the run 5.

26 Page 26 of 35 Figure 26: LM404 Output Figure 27: LM404 Output 2

27 Page 27 of 35 Figure 28: LM404 Output 3 Figure 29: LM404 Output 4

28 Page 28 of 35 Figure 30: LM404 Output 5 Figure 3: LM404 Output 6 There is no evidence of degradation of the outputs except for the first chip which shows a drift of about 2.5mV. The other outputs fluctuate between a peak to peak value of mv. No drift was observed on the power supply. The output voltage tolerance given by the manufacturer is +-0.% which means 6.82mV in this case. Therefore, the component remains within this tolerance up 200Gy. TL43 Figures depict the behaviour of the outputs of the TL43 voltage references as a function of the total ionizing dose of 205Gy (runs to 3). The outputs remain quite stable until a TID of 60Gy. Then, the average drift for the outputs is 4.6mV. The maximum drift of 6.4mV was recorded on output 5. The power

29 Page 29 of 35 consumption remained stable during the whole irradiation time. The outputs remained within the tolerance of 2% (35mV) for the whole irradiation time. Figure 32: TL43 Output Figure 33: TL43 Output 2

30 Page 30 of 35 Figure 34: TL43 Output 3. Figure 35: TL43 Output 4

31 Page 3 of 35 Figure 36: TL43 Output 5 Figure 37: TL43 Output 6 TL432 Figures 50 to 55 show the behaviour of the TL432 references as a function of the TID (run 4). The drift remains below 4mV until 92Gy. From this TID value, a step occurred on the outputs. The step can be either positive or negative. No voltage step was observed on the output 5. Table 7 gives the step values for each output. That could also be an artefact of the measurement system. Nevertheless, the values remain within the datasheet specification which indicates an output tolerance of 0.5% (~34mV).

32 Page 32 of 35 Table 7: Output steps observed on TL432s Output # Step value -4.8mV 2-9.mV 3 4.5mV 4 3.5mV 6-6.7mV Figure 38: TL432 Output

33 Page 33 of 35 Figure 39: TL432 Output 2 Figure 40: TL432 Output 3

34 Page 34 of 35 Figure 4: TL432 Output 4 Figure 42: TL432 Output 5

35 Page 35 of 35 Figure 43: TL432 Output 6 Conclusions During this test campaign, 6 analog components have been exposed to a proton particle beam, 3 operational amplifiers and 3 voltage references. Six chips of each voltage reference and three chips of each amplifier (TL072 and OPA2227 contain 2 amps in a single chip) were tested. The test setup was designed to detect events such as SETs and SELs. Regarding the SET detection, the test apparatus is able to detect a pulse of 5ns with amplitude of 00mV. With these parameters, no events were triggered for any of the tested components. It was shown by the literature that it is difficult to detect events on most of the analog components while they are exposed to proton irradiations. All the components have shown a weak sensitivity to the TID resulting on a drift on the outputs. Regarding the voltage references, all of them have their output that remains within the tolerance given by the manufacturers. As far as the amplifiers are concerned, their output drift remain within the specifications, apart for the INA 4 which goes out of the specifications at TID of 30 Gy. The dose rate in the LHC tunnel is much lower than the one at which the test was done. Therefore, ELDR (Enhanced Low Dose Rate) effects might not be correctly evaluated. A TID, a factor 4 higher than the one fixed as a target (50 Gy) was reached in order to compensate for possible ELDR effect. Finally the performance degradation due to the TID can depend on the component batch.