SVT Gains and Offsets from UCSB TestBench

This document summarizes the procedure used to extract electronics gains, offsets and noise on a Type-B layer 2 module with a hybrid fully loaded (7 chips/side) with first version radiation-soft chips.

1. All measurements were performed with 100 nsec shaping time.



2. The clock frequency was 40 MHZ.



3. We used the prototype wire-wrap DAQ-link and HDI-link cards, with transistor-based level shifting. We used the prototype wire-wrap matching card for the earlier measurements on the unbonded hybrid. For the measurements on the bonded detector, we switched to the PC version of the matching card.



4. The hybrid and the matching card were connected through a a set of two short single layer flat kapton circuits and a home-made cable harness. In this harness, digital signals were transmitted on 50 ohm coaxial cables with their shield tied to digital ground.
   
   
5. Power was supplied by a set of benchtop power supplies. Voltage levels were set at the beginning of each run using the sense lines brought out from the HDI. Currents on the 5V and 2V supplies to the hybrid were monitored with digital multimeters in series with the power lines. Silicon currents were monitored using nanoammeters.



6. Gains, offsets, and noise were measured with the detector in the ring frame. A noise measurement was repeated after attachment of the ribs, with the detector in the test-beam box using tails provided by the INFN-Milano group. This test yielded consistent results. Noise measurements were also made using the new version of the data transmission system with the module in the test-beam box. Initially, these measurement showed one order of magnitude higher noise levels. After some reconfiguring of cabling etc., the noise as measured with the new data transmission was reduced and was found to be comparable to the noise measured with the old data transmission system (although still worse by 10% or so).
   
   
7. Gains Offsets and Noise



1. Measurements of gains and offsets were based on a series of threshold scans at fixed charge injection (with different charge injection values for different runs).



2. The charge injection circuitry was fired for only one channel per chip at a time.



3. Typically, for each threshold and charge injection setting, each channel was fired 50 times.



4. The 50% firing point and the noise for each channel were extracted from a fit of the threshold turn-on curves to an error function using binomial errors and constraining the top of the threshold curve to 100% efficiency. Two typical fits for bonded channels are shown in Figure 1. (Postscript version of Figure 1).
   
   
   
   Figure 1: Results of threshold scans at fixed charge injection for a bonded channel on the n- (phi-) side (Chip 3, Channel 40) and a bonded channel on the p- (z-) side (Chip 10 Channel 40). What is plotted is the number of firings as a function of threshold measured in Threshold DAC counts. Note that here threshold decreases left-to-right.
   
   The fit parameters P1 and P2 in Figure 1 give the 50% turn on point and the gaussian noise component in Threshold DAC counts respectively. As we will describe below, one Threshold DAC count corresponds to a charge of approximately 400 electrons.



5. The gains and offsets were then extracted by fitting the 50% firing points for each channel as a function of different charge injection settings. The fits to extract gains and offsets for the same two channels of Figure 1, are shown in Figure 2. (Postscript version of Figure 2).
   
   
   
   Figure 2: Linearity fits for a bonded channel on the n- (phi-) side (Chip 3, Channel 40) and a bonded channel on the p- (z-) side (Chip 10 Channel 40). On the horizontal axis we plot the value of the injected charge, in units of Calibration DAC counts; on the vertical axis we plot the 50% turn on point, in unit of Threshold DAC counts.
   
   The fit parameters P1 and P2 in Figure 2 give, respectively, offset and gain in units of DAC counts. Note that with this convention, a channel with zero offset (in mV) would have an offset of 63 Threshold DAC counts.
   
   
6. To convert from Threshold DAC and Calibration DAC counts to SI units, we used the following factors:



• Threshold DAC: 4.29 mV/count



• Calibration DAC: 8.217 mV/count



• Charge injecting capacitor: 65 fF



• Using the above values, Calibration DAC: 65*0.008217=0.51 fC/count



7. The gain and offset variations for two chips bonded to the detector are shown in Figure 3 (Postscript version of Figure 3).
   
   
   
   Figure 3: Gains and offsets for chip 3 (n- or phi-side) and chip 10 (p- or z-side)
   
   
8. The noise for all channels in the module for a typical run is shown in Figure 4 (Postscript version of Figure 4).
   
   
   
   Figure 4: Noise in Threshold DAC counts for all channels on the n- (phi-) and p- (z-) sides of the detector as measured in a typical threshoold scan.
   
   To convert the noise measurement from Threshold DAC counts to number of electrons, we use the conversion factors given above, and a typical gain of 65 mV/fC (see Figure 3):
   
   
   • 1 fC = 65/4.29 = 15.15 Threshold DAC counts
   • 1 fC = 6250 electrons
   • 1 Threshold DAC count = 6250/15.15 = 410 electrons
   
   Thus, the typical noise seen is approximately
   
   
   • ~ 2.6 Threshold DAC counts = 1070 electrons on the n- (phi-) side
   • ~ 2.1 Threshold DAC counts = 860 electrons on the p- (z-) side
   
   which can be compared with the latest predictions of Issy Kipnis' simulation for a Layer 2 detector:
   
   
   • ~ 810 electrons on the n- (phi-) side
   • ~ 660 electrons on the p- (z-) side
   
   All channels on the module are included in the plots of Figure 4, i.e. known bad channels and unbonded channels have not been removed. The pileup of channels at zero noise is due to dead channels. One of the seven chips on the n-side of the detector is effectively dead, since both channel and calibration masks appear to be stuck in an OFF state. The other clump of channels with noise of order 0.8 Threshold DAC counts is due to unbonded channels. Note also that some of the spread in the noise values is due to the statistical uncertainty on the noise measurement, which is typically of order 0.2 Threshold DAC counts (see the fits in Figure 1):



8. Channel-by-channel gains, offsets, and noise are tabulated in the EXCEL spread sheet which contains also the bonding map and the list of bad channels. This spread sheet is available through a link from the UCSB test-beam page. More plots/information is also available on request.

Return to UCSB test-beam page