Bench test of the ADS16 chip at PNPI


  • Introduction.

  • -The ADS16 chip was tested, channel by channel, at PNPI with the use of the test stand prototype. Here we present analysis of the data which were provided by A.Golyash as the tables. The data and results are very preliminary. The goal of the analysis is to establish methods and parameters to characterize the chip performance.
  • Time vs input charge.

  • - The mean of time distribution as a function of input charge. Fig.1 .
  • Threshold vs JTAG code.

  • - The thresholds controlled by JTAG code were calibrated in fC on the bench too. The results are in Fig.2
  • Input charge vs DAC code calibration.

  • - Input signal vs DAC code dependence was calibrated at Ccal=0.84 pF. The results are here . The input charge (fC) vs DAC code data were fitted by the polinomial of degree 3.
  • Thresholds and noise.

  • - The thresholds above (see Fig.2 ) were obtained at PNPI with the use of the linear interpolation of the threshold curve between the points with 0.1 and 0.9 efficiencies. This can bias the threshold estimate if the threshold curve has "abnormal" shape around this points ( a few examples could be seen in Fig. 3 below). More accurate approach using fit by the Fermi distribution gives thresholds which are lower by about 10%. The results for each channel (chip 1) at three different JTAG settings (JTAG=150,170 and 190) and Cin=0.84 pF are on the pp 1-4 of the postscript files below (Fig.3a,b,c). Page 5 gives fitted thresholds and the slopes for each channel as the corresponding parameters P1 and P2 of the Fermi function eff=P3/(1. + exp(-(Q - P1)/P2)). In Fig.5-8 the fit was done with 2 parameters only,
    eff=400/(1. + exp(-(Q - P1)/P2)).
    - Having Fermi distribution we can get resolution finctions by differentiating the Fermi function (see pp 1-4 as well). They are histogrammed and fitted by Gaussian distribution on the p.6. Note that resolution function is fitted by Gaussian rather close except the 3-5% tails, p.7 (also, the derivative of the Fermi function is not a Gaussian). From this we can estimate the noise of each channel as RMS, provided by HBOOK or RMS as the fitted parameter (p.8). In average the noise is about 1.2-1.4 fC and fitted RMS are lower (due to non-gaussian tails).
    - Results at Cin=0 pF:
    Threshold JTAG = 150 Fig.3a
    Threshold JTAG = 170 Fig.3b
    Threshold JTAG = 190 Fig.3c
    - The summary for the fitted thresholds and noise at JTAG=150,170 and 190 is in Fig.4
  • Thresholds and noise at different Cin.

  • - The calibration curve Input Signal vs DAC code is here .
    - Results at Cin=0 pF:
    Threshold JTAG = 150 Fig.5a
    Threshold JTAG = 170 Fig.5b
    - Results at Cin=55 pF:
    Threshold JTAG = 150 Fig.6a
    Threshold JTAG = 170 Fig.6b
    - Results at Cin=125 pF:
    Threshold JTAG = 150 Fig.7a
    Threshold JTAG = 170 Fig.7b
    - The summary for the fitted thresholds and noise at JTAG=150,170 and Cin above is in Fig.8

    bondar@fnal.gov teren@fnal.gov
    Last modified: Mon May 31 10:30:00 CST 1999