Performance of CMP16F chips 1 - 90
(Data from the ASIC Chip Test Stand at Fermilab, May - August, 2000)

blue_ball.gif Introduction.

Ninety 16-channel preamplifier-discriminator chips, CMP16F (Jan. 2000 MOSIS submission), were installed on 16-channel anode front-end boards, AD16, and tested on the ASIC test stand at Fermilab during May - August, 2000. Descriptions of the chip and the anode front-end board, as well as the preliminary results from tests on the first 28 boards, can be found in the documentation for the EMU Electronics System Review at CERN, 09/18/00 (see CMU Review page). See also talks given by T.Ferguson ( talk) and by N.Bondar ( talk) on EMU meetings at CERN (June 4, 2000) and Fermilab (Sep.8, 2000).
Here, we present the results of measurements of the threshold, noise, gain, discriminator offset, time resolution and slewing time for 87 boards. Three boards (58, 59 and 77) were excluded at the very beginning of the tests as being malfunctioning.
blue_ball.gif Conditions.

The shape of the calibration input test pulse sent to each board is shown here. The maximim amplitude of the test pulse is controlled by a DAC code, and was measured with a digital scope as a function of the DAC code. The resulting data were fitted by the function A(mV) = A0 + A1*DAC + A2*DAC*DAC (with A0 = -1.81, A1 = 0.185 and A2 = -1.06E-06). The corresponding input charge, Qin, was calculated using Qin(fC) = A(mV) * Cinj / R, where Cinj = 1.6 pF and the factor R corresponds to the attenuation used of either 16 dB (R = 6.31) or 6 dB (R = 2.0) (see (Calibration) for the 16 dB attenuator). Irregularities in the differences between the fit values and the data in the picture are due to switching the scope scale.
The threshold of the discriminator in the chip was controlled by changing the discriminator input voltage, Ud(mV). For the CMP16F chip and for all previous submissions, the actual discriminator threshold level, Uthr, is related to Ud by Uthr(mV) = 1600 mV - Ud. Therefore, the threshold goes down as the Ud value increases (e.g. Qthr = 0 at Ud = 1600 mV and Qthr is approximately 17 fC at Ud = 1400 mV).
blue_ball.gif Threshold, noise, gain and discriminator offset measurements.

The preamp chip threshold (Qthr) and noise for each channel have been measured at three different Ud settings (Ud = 1000, 1200 and 1400 mV), using an attenuation of 16 dB, and at four values of input capacitance Cdet (Cdet = 0, 56, 100 and 180 pF). See here for details of the method used. At Cdet = 0 pF and 100 pF, the data are available for boards 1-28, 39-57, 60-76 and 78-83, for a total of 70 in all. At Cdet =56 pF, for boards 1-38, and at Cdet = 180 pF, for boards 1-90, excluding boards 58, 59 and 77.
Results for Cdet = 180 pF are presented in Fig.1_180pF. Page 1 gives the Qthr and noise distributions; pg. 2 - the Chi2 fit to the threshold curve for all three Ud settings; pg. 14 - the means of the Qthr and noise values for each board versus board number; pg. 15 - the residuals for the Qthr and noise distributions, which are the differences between each channel's value and the mean value for that chip. Note that the average value of the noise obtained in this analysis is around 2.1 fC instead of 1.5 fC as found in the previous analysis of boards 1-28. This is because we now fit the threshold curve to a more convenient Gaussian error function, instead of a Fermi function. To address this issue properly, we need in future tests to decrease the step size in the DAC code for the input signal from 20 to 5, and eliminate the binning effect. Another feature is the use of the Chi2 in this analysis. One can see that the mean Chi2 is close to one only for Ud = 1400 mV. At higher thresholds, the fit is not as good as for Ud = 1400 mV.
The gain and the discriminator offset (the difference between the nominal 1600 mV value and the observed value) are obtained from a fit to the Qthr values at all three Uthr (page 18 in Fig.1_180pF). Their distributions can be seen on pg. 29, the mean values on pg. 32 and the residuals on p. 31. The means and widths of the distributions for all boards together, as well as the widths of the channel residuals, are given in Table 1 below. A comparison with the same table for the CMP16E chip (our previous submission, CMP16E) shows significant improvement in Qthr and the Qthr residual distributions. Their widths are 1.5 times less than for the CMP16E chip. This is likely due to the AMI transition from a 1.2 micron to a 1.5 micron technology between the two submissions.
Results at Cdet = 0, 56 and 100 pF are shown in Fig.1_0pF, Fig.1_56pF and Fig.1_100pF. All parameters except the noise are independent of Cdet (see Fig.1_Cdet where the mean values of the parameters are presented).
There is an interesting feature in the data taken at Cdet = 0 pF. Channel #16 in all boards has systematically different values from the rest of the channels for Qthr, noise and gain (see pages 16 and 33 in Fig.1_0pF).
blue_ball.gif Measurements of the time resolution and its variation (Fig. 2).

The time resolution of the chip was measured at Ud = 1400 mV (Qthr = 17 fC) with an input signal amplitude in the range 50 fC < Qin < 180 fC (an attenuation of 16 dB). Data are available for boards 1-32, 35-57, 60-76 and 78-90, a total of 85 in all. The time resolution was calculated for each channel of the measured board as the RMS of the time distribution for that channel at a given Qin. Page 1 in Fig. 2 presents the RMS at Qin = 50, 100 and 150 fC, and the variation of the RMS, which was defined as Max(RMS) - Min(RMS) in the interval of 50 fC < Qin < 180 fC. The remaining pages have distributions plotted for Qin = 100 fC. See the summary given in Table 2 below.
blue_ball.gif Measurements of the mean time and the slewing time (Fig. 3).

Data were taken at Ud = 1400 mV and attenuations of 6 dB and 16 dB. Two attenuators were needed in order to take into account the contribution of the pulser slewing time. Data are available for boards 1-42, 44-45, 47-49, 51-57, 60-76 and 78-86, a total of 80 boards in all. The measured parameters are the mean of the time distribution for each channel at Qin = 100 fC, and the slewing time, which was defined as Max(mean)-Min(mean) in the interval of 46 fC < Qin < 578 fC for each channel (page 1 in Fig. 3). Table 3 presents the characteristics of the time measurements.
blue_ball.gif Tables.

Table 1. Threshold, noise, gain and discriminator offset at Cdet = 180 pF (from Fig.1_180pF).
The means and sigmas are from a Gaussian fit.

ParameterMean SigmaPage #
Qthr (fC) at Ud = 1400 mV17.3 2.61
Qthr residual (fC) at Ud = 1400 mV - 1.3 15
Noise (fC) at Ud = 1400 mV2.1 0.1 1
Noise residual (fC) at Ud = 1400 mV - 0.08 15
Gain (mV/fC) 9.0 0.3 29
Gain residual (mV/fC) - 0.2 31
Discriminator offset (mV)51.2 21.5 29
Discriminator offset residual (mV) - 11.5 31

Table 2. Time resolution at Ud = 1400 mV (Qthr = 17 fC) and Cdet = 180 pF (from Fig.2).
The means and sigmas are from a Gaussian fit.

ParameterMean SigmaPage #
Time resolution RMS (ns) at Qin = 50 fC 1.7 0.2 page 1
Time resolution RMS (ns) at Qin = 100 fC 0.7 0.08 page 1
Time resolution RMS (ns) at Qin = 150 fC 0.6 0.07 page 1
Variation of RMS (ns) at 50 < Qin < 180 fC 1.2 0.2 page 1
Time resolution RMS residual (ns) at Qin = 100 fC - 0.07 page 4

Table 3. Mean time and slewing time at Ud = 1400 mV (Qthr = 17 fC) and Cdet = 180 pF (from Fig.3).
The means and sigmas are from a Gaussian fit.

ParameterMean SigmaPage #
Mean time (ns) 123.8 1.2 page 1
Mean time residual (ns) - 0.4 page 4
Slewing time (ns) 3.1 0.6 page 1
Slewing time residual (ns) - 0.35 page 4

CMU CMS home page
Last modified: Tue Nov 27 18:15:00 CST 2000