HV-Supply for the ALICE-TPC Readout Chambers: Test of the ISEG EHQ F020p High Voltage Module R. Renfordt1), H.R. Schmidt2) 1) I IKF, University of Frankfurt, 2) GSI Darmstadt Introduction This paper reports on test results of an ISEG EHQ F020p high voltage module. This unit has been chosen for evaluation as the outcome of a tendering procedure for ALICE-TPC readout chamber HV supplies carried out at GSI in spring 2000. The specification required in the tendering and the nominal specs of the EHQ F020p module are listed in Table 1. requested EHQ F020p Vmax [V] 2000 2500 HV set resolution [V] <1 0.04 HV output accuracy [V] <1 ±0.44 HV ripple [mV] <50 < 20 Temperature coefficient <12 1 <5 Imax [] 200 200 current resolution [nA] <20 4 programmable current trip √ √ Ethernet interface √ (√)2 CAN bus interface √ √ Table 1 Comparison of requested and given specifications 1 For a detailed discussion of the measured temperature coefficient refer to Sec. III.4 2 Third party product The purpose of the tests described below was to verify the HV accuracy in terms of ripple, relative set resolution, reproducibility and the dependence on temperature as far as these values are relevant for the proper operation for the TPC readout chambers. It was not intended to provide a comprehensive test of the ISEG specifications. II Test Setup II.1 Hardware The test setup is sketched in Figure 1. It consists of one 16-channel EHQ F020p HV module, a custom made 16-channel voltage divider box, a CAMAC LeCroy 8252 scanning ADC in a CAMAC crate with a HYTEC™ CAMAC controller, a PEAK™ CAN bus controller with parallel port interface and a PC running Windows NT 4.0 equipped with a HYTEC™ PCI interface card. The divider box contains a resistor chain of 150 and 1 M yielding a factor 150 downscaling of the HV output. To match the impedance of the box to the scanning ADC the resistor chain is followed by a buffer amplifier with gain =1. This setup allows to monitor simultaneously the voltage and current from the internal EHQ F020p ADC’s and from the external measurement via divider box and CAMAC ADC. The load is similar to the one expected under ALICE running conditions (≈ 10 A). In addition, the temperature of the module and the ambient temperature was monitored with several PT100-elements and also recorded via the LeCroy 8252 scanning ADC. Figure 1 PC - Windows NT 4.0 CAN-bus CAMAC-controller 16 CAN-controller ADC LeCroy 8252 ISEG EHQ F020p divider 150:1 Schematic sketch of the measuring setup 2 II.2 Software A custom control program, written in BORLAND C++, allows to record the measured parameters to disk as well as to control all other features of the HV module. The CANbus driver routines were provided by ISEG, the CAMAC driver routines by H. Stelzer. 1401 1400 U meas-ISEGN [V] 1400.5 1399.5 1399 0 10 20 30 40 50 60 70 80 time [min] Figure 2 II.3 Example of the measurement cycle for one channel Accuracy of the Measurement The precision of the LeCroy 8252 ADC is 12 bit for a range of 10 V, folded with the division factor of 150 from the resistor chain, an accuracy of 366 mV/bit is reached. The precision of the ISEG internal measurement is 50 mV/bit and 4 nA/bit for the voltage and current measurements, respectively. It should be noted that the precision of the external measurement is only relative, since no special selection or calibration of the resistors in the divider box was done. 3 III Test Results Figure 2 shows one HV channel as example of the measurement cycles: the HV is ramped up at a fixed speed (typically 0.1 Vset/sec), then the voltage is kept at its nominal value for a certain period of time and thereafter ramped down again. The question under investigation is whether after repetitive switching on/off the nominal voltage is always restored with the specified accuracy. For the actual measurement all 16 channels of the ISEG module were switched simultaneously. 548 on/off switches, internal (ISEG) HV measurement 1250.2 U HV Channel # 1 nominal =1250 V 1250.15 1250.05 U meas/ISEG [V] 1250.1 1250 1249.95 1249.9 0 200 400 600 800 1000 relative time [min] Figure 3 Voltage values after switching on/off cycles for one channel measured internally by the ISEG module III.1 Switching On/Off Figure 3 gives, as representative sample, the voltage of channel #1 as function of time. Each point corresponds to a measurement following a switching off/on. Altogether each channel was switched on/off 548 times with about 2 minutes time between successive cycles. As can be seen from the figure the stability is ±50 mV, corresponding to the fluctuation of one LSB. Figure 4 summarizes the measurement of all 16 channels: plotted is the distribution of 16 548 4 measurements. The plot shows that the average over both channels and the time is within =50 mV. These results are independently checked by the external measurement. Figure 5 shows the measured voltages from the divider box (ADCchannel 150 10000/4096 mV). The measurement of the different channels varies up to 30 volts due to the non-calibrated resistor chains or ADC’s. However, the fluctuation in time of the individual channels is again of ±1 LSB, indicating that the external measurement confirms the reproducibility of the HV setting. 548 on/off switches, 16 HV channels, internal (ISEG) HV measurement 4000 entries 3000 2000 1000 0 1249.75 1250 1250.25 U [V] 1250.5 ISEG Figure 4 16 channels Distribution of voltage value for 548 switching cycles for all 5 548 on/off switches, external (CAMAC) HV measurement 1335 1330 CAMAC [V] 1325 U 1320 1315 1310 1305 0 200 400 600 800 1000 relative time [min] Figure 5 Voltage values after switching on/off cycles for all 16 channels measured by the external CAMAC ADC 1310.5 HV channel #1 HV channel #2-16 1309.5 U meas/CAMAC [V] 1310 1309 1308.5 0 200 400 600 800 1000 time [sec] Figure 6 Voltage values of channel #1 (red, thin curve) while switching channel #2-16 (blue, thick curve) simultaneously from 0 to V max (measured via the external CAMAC ADC) 6 III.2 Mutual Influence The mutual influence of channels was tested by setting one channel to the nominal voltage (1250V) while switching all other channels periodically from 0 to Vmax. The behavior of HV can be seen from Figure 6 indicating no significant change in voltage of the selected channel (#1) while switching all other channels. III.3 Settling Time Figure 7 and Figure 8 give views of the leading edge of one cycle as in Figure 2 to demonstrate the time needed until the voltage is stable. As can be seen the HV ramps up until it reaches a voltage about 5 Volts below Vset and then settles slowly within about 20 second to the nominal voltage without any overshoot. 1500 U meas/ISEG [V] 1000 500 0 0 20 40 60 80 100 time [s] Figure 7 Leading edge of one cycle 7 1410 1400 U meas/ISEG [V] 1390 1380 1370 1360 1350 1340 0 20 40 60 80 100 time [s] Figure 8 III.4 Enlarged view of the leading edge of one cycle Temperature Dependence The temperature dependence, again shown as representative sample, is depicted for channel #16 in Figure 9, Figure 10 and Figure 11. A marked day/night variation is visible in a 24 hrs measurement both for the voltage and the current. Both values are anti-correlated with the also measured ambient temperature, shown as the red curve. The variation in voltage is and =153 mV for the internal and external measurements, respectively. The corresponding range in temperature is Tpeak-peak ≈ 7 °C. Figure 12 shows the same parameters as those in Figure 11, however, for an extended period of 6 days. A correlation analysis of the {Ti, Vi} samples yields temperature coefficients, cT:=1/V dV/dT, of 12 externally and internally (as, e.g., in Figure 10) measured voltages, respectively. Both numbers have to be interpreted carefully, the first one because it includes the temperature dependence of the load resistors as well as the temperature dependence of the external measurement circuit, the latter one because it probably is in feedback with the generated voltage. In any case, however, the numbers are comparable with the value quoted by ISEG for the temperature coefficient (<5 8 8.8 30 6/12/2001 11:09 6/13/2001 9:16 28 26 T [°C] Imeas/ISEG[A] 8.795 8.79 24 8.785 22 T [ °C] Imeas/ISEG [A] 8.78 0 5 10 15 20 25 20 relative time [hrs] Figure 9 Time dependence of the internally measured current (black symbols) and the measured ambient temperature (red curve). The black curve is a fit through the measured currents. 1250.2 30 6/12/2001 11:09 6/13/2001 9:16 1250.15 28 26 1250.05 T [°C] U meas/ISEG[V] 1250.1 1250 24 1249.95 22 T [°C] 1249.9 Umeas/ISEG [V] 1249.85 0 5 10 15 20 25 20 relative time [hrs] Figure 10 Time dependence of the internally measured voltages (black symbols) and the measured ambient temperature (red curve). The black curve is a fit through the measured voltages. 9 1314.5 30 6/12/2001 11:09 6/13/2001 9:16 28 [V] 1314 T [°C] meas/CAMAC 26 U 24 1313.5 22 T [°C] Umeas/CAMAC[V] 1313 0 5 10 15 20 25 20 relative time [hrs] Figure 11 Time dependence of the externally measured voltage (black symbols) and the measured ambient temperature (red curve). The black curve is a fit through the measured voltages. 1314.4 28 1314.2 26 1314 22 1313.6 20 T [°C] U CMAC [V] 24 1313.8 1313.4 18 1313.2 16 1313 14 1312.8 0 1000 2000 3000 4000 5000 6000 7000 8000 time [min] Figure 12 Time dependence of the externally measured voltage (blue symbols) and the measured ambient temperature (red curve) for an extended period of 6 days. 10 III.5 Low Voltage Supply Failure The effect of dripping LV supply voltages due to potential failures in the LV power supplies has been investigated: the 24V or the 5V supply voltages was turned off slowly while recording the output HV. In both cases the HV switched off rapidly when the low voltage went below a certain level. Above this level the HV output did not change with varying low voltage. The modules stayed switched off when bringing one or the other voltage back up, i.e., the module delivered HV again only after setting the voltage explicitly. III.6 Ripple Ripple (and noise) is specified to be < 20 mV at full load for voltages above 400 V and voltage differences between channels below 1400 V. In the present case the ripple was measured employing an AC coupled probe. All channels were set at 1250 V and loaded with 8.8 scope used (2 mV) no ripple was detected. IV Summary In summary, this specific set of measurements confirms that the module meets the specifications required for operating the ALICE TPC readout chambers in terms temperature variations, ripple, as well as the reproducibility of Vset. 11