ACC Readout Electronics

Schematics

• Readout Board (19Kbyte eps).

Design & Production

• Talk on preliminary Readout Specifications (99 kByte Postscript) and design considerations.

• An updated (Mar. 28, 1998) design note (117Kbyte PS in progress) for the 9U VME board is now available. Some figures in that note are available as separate files:
  • mqt301.eps (19Kbyte eps).
  • vga_filter.eps (10Kbyte eps).
  • qtot_timing.eps (10Kbyte eps).

• Status Update (830 Kbyte PS) Presented at November 1997 Collaboration Meeting.

• Production Schedule (8 Kbyte PS) updated, Dec. 29, 1997.

• Schedule for various Readout Components. (8 Kbyte PS) Updated, June 23, 1998.

Test Results

Production Testing of Boards

• Test procedures. (175 Kbyte PS) May 2, 1998.

• Board summary

Preproduction 9U VME Prototype

A number of minor problems with the 1st 9U prototype were identified and repaired and a FAST OR output for triggering was added. All 26 channels on this board were assembled. Bench test data are summarized below.

• Width vs channel and pedestal vs. channel data are shown in this plot. As can be seen, the channel-to-channel variations are small. The RMS variation in width mode is typically 0.25 ns (consistent with TDC resolution). The RMS variation in pedestal mode is typically 0.50 ns.

• Pedestal subtracted output width vs channel for various input pulseheights are shown in this plot. The RMS variation in pedestal mode is typically between 0.5 ns at small pulseheights to 1.2 ns at larger pulseheights. For the larger pulseheights, the RMS is dominated by pulse-to-pulse variations in the test pulser output.

• Q/T gain vs channel as determined from the previous data is shown in this plot. The control voltage on the VGAs was set to 0V. There is a 100 ohm reverse termination resistor between the amplifier that drives the MQT chip and the lumped delay line.

• The pedestal subtracted output width for various control voltage settings of the variable gain amplifier is shown in this plot. These data were taken after changing the amplifier reverse termination resistor from 100 ohms to 27 ohms, which increases the typical gain from 1.5 ns/pC to 2.4 ns/pC for a VGA control voltage of 0V. These data are summarized in this plot, which shows the gain as a function of the VGA control voltage. Also shown is the RMS noise (pedestal width) for each control voltage setting. Recall that in all cases the full scale output is about 4000 ns.

• The ACC MQT board has provision for an external gate. Proper operation of this gate is shown in this plot. The upper histogram shows the timing spectrum obtained without the external gate. The large spike in the middle (note the semilog scale) corresponds to the in-time events. The events at positive time most likely correspond to flouresence of the plastic scintillator or after pulsing in the FMPMT. The events at negative times are mostly pure accidentals. Both types of pulses are eliminated in the lower plot, where an external gate derived from the tigger PMT is applied. The small number of events not rejected by the gate are due to pileup in the gating logic.

• Discriminator curves for four channels selected at random are shown in this plot. The discriminator voltage was set to 100 mV, with a VGA control voltage of 0V. The plot shows that the "turn on range" for a typical channel is about 1.25 mV, or about 20% of the setting. The minimum threshold that can be set is about 2~mV. This is shown here, where the noise count rate is plotted versus threshold setting.

First 9U VME Prototype

This board was a full size 9U x 400 mm VME board. Only two of the 26 channels on the board were assembled. The board was tested on the bench and in the beam at KEK.

• Draft BELLE Note on the ``analysis'' (1.3 MByte PS) of the Dec. 1997 beam test has been prepared.

CAMAC Prototype

• Timing and pulseheight spectra (120 Kbyte PS) obtained using a FMPMT and a prototype 9U VME board. The timing spectrum shows the decay time of the plastic scintillator. The pulseheight spectrum shows a number of interesting features. In particular, one can two broad bumps. The possibility that the two bumps are due to single- and double-photonelectron events is small, since the observed probability of seeing NO hits is about 96%. There is also an strong peak near zero. This peak is present for out of time (noise) hits as well.

  The glitch in the pulseheight spectrum is not yet fully understood, although its origin has been traced to the MQT300A chip.

  The lego plot of charge versus time (440 Kbyte PS) shows that the noise is qualitatively different from the single photoelectron signal. The intime structure at t=0 is the single photoelectron signal. At other times there are significant numbers of pulses that are much smaller than the single photoelectron pulseheight. Indeed, the lower limit on pulseheight is imposed by the Q/T threshold, which is 3 mV. The FMPMT was operated at 1600 V in zero field. The preamplier gain was x10.

  These pulses are NOT electronic noise, as was verified by carrying out runs with the FMPMT HV turned off (in such cases there are essentially no triggers). It is interesting to note that the SPE signal also has a strong component at small pulseheights.

  The tremendous dynamic range of the LeCroy MTD133 chips is illustrated in the leading-edge timing spectra (135 Kbyte PS). The upper plot shows all events that satisfy the 3 mV online threshold cut. The lower plot shows the spectrum after a Q>25 (Q in "ns") cut is applied. This cut rejects about 90% of the accidental background hits, which as noted populate the small pulseheight region.

  Additional bench-test results can be found in Status Update (830 Kbyte PS) which was presented at November 1997 Collaboration Meeting.