Time Equalizer Board

Prototype Test

W.S. Aug/21/01


Abstract: The Time Equalizer board is under development. This single width CAMAC unit will be used to align timing signals from the Cerenkov counters (either the Maryland Cerenkov array or Rochester T0's) for the PHOBOS experiment at RHIC.

Design goals


Date: Fri, 13 Apr 2001 16:05:46 -0400 (EDT)
From: Wojtek Skulski <skulski@phobtof.pas.rochester.edu>
To: Frank Wolfs <wolfs@pas.rochester.edu>,
     PHOBOS TOF Group <tof@liststar.pas.rochester.edu>

Board form factor               CAMAC single width
Number of channels              16
Signal in and out               ECL
Connectors                      ribbon cable, two connectors (in and out)
Shortest possible delay tpd     6.5 ns
Shortest possible delay step    10 ps
Number of steps                 256
Longest possible delay          1 milisecond
Delay tempco                    7.5 ps/degree C
Delay jitter                    10 ps
Single step size                settable by swapping resistors
Max trigger rate per channel    60 MHz
Min output pulse width          3 ns
Max output pulse width          several ns (set once per chip with a R/C)

Remarks:

1. The total delay is given by

          td = tpd + N*(range/256)   N = 0,1,2,...,255

  i.e., the delay range is divided into 256 equal steps. There is a
  minimum propagation delay tpd of 6.5 ns.

  The step size of 10 ps is only applicable at the shortest range of 2.5 ns.
  At longer ranges the step size will be longer. For example, if the range
  is set to 25 ns, then the step size will be 100 ns, and the total
  delay will be 6.5 ns to 31.5 ns.
 

The schematic and the board layout

Both the schematic and the board layout were done during off-shift hours in the TOF apartments, both on-site and at the NAT Center. I appreciate the patience of my roommates while I was working on the board. Jan Toke of the Rochester Nuclear Chemistry group reviewed the schematic and pointed out a couple of mistakes

In order to achieve the design goals I used layout techniques appropriate for high-speed analog design, such as four-layer printed circuit board with separate power and ground planes. A few components, such as pull-up resistors and a DIP-switch, are meant only for initial testing without the benefit of a full-blown CAMAC interface, which will be assembled at a later stage.

The prototype

The prototype PCB was produced by Sierra Proto Express, whose web site can be reached under a funny name www.2justforyou.com. Despite the funny name, their service was quick and very professional. I was more than pleased with both the price and the quality of the PCBs.
The prototype
Figure 1. (Click here to see a more detailed photo). The actual prototype with two delay chips soldered in (black rectangles near the upper-right corner and in the center of the board). The input/output ribbon cable sockets are on the right, the CAMAC connector is on the left. The wire loops soldered in other channels were used to measure the input-output propagation time for the unpopulated channels. The blue DIP switch was used to dial in the value of the delay during the tests. The silver square on the left is the area where the CAMAC interface chip will be soldered.

I soldered the surface-mount chips, resistors, and capacitors myself. The soldering of the 28-pin delay chips in PLCC packages was a bit tricky. The pins of those chips are tucked under the chip body like a letter "J", what is not really meant for hand soldering.
 

The first test

The first tests were performed on Aug/16, using a NIM pulser, ECL-NIM-ECL converters, a TAC, and an ADC.  Only one channel out of sixteen was populated with components. The time range of the delay chip was set to 2.5 ns using a socketed resistor.

Date: Fri, 17 Aug 2001 12:33:00 -0400 (EDT)
From: Wojtek Skulski <skulski@phobtof.pas.rochester.edu>
To: PHOBOS TOF Group <tof@liststar.pas.rochester.edu>
Subject: Equalizer board tests

The sigma contribution of the board itself (disentangled
from the data) is 5.2 ps. This number is in fact 3 times
smaller than the TAC channel width in this measurement.

First tests
Figure 2. (The picture size was shrunk, what may have introduced a distortion. Click here for the full-size picture). The sequence of peaks corresponding to the delay word = 0, 64, 128, 192, 255. The left-most peak was acquired with the board bypassed (inputs shorted to outputs). The minimum propagation delay equal to 6.6 ns is indicated with a horizontal arrow.

The very small jitter introduced by the board is evidenced by a slight increase of the width of the peaks, as compared with the left-most one. The reference peak without the board is only 2 ADC channels wide. The board adds one more channel to the width. This is consistent with the specs of the delay chip, which specifies resolution of one delay step = 10ps (in the present case 2.5ns / 255 steps). From the direct comparison of the spectra we see that the increase in the width is indeed consistent with one delay step.

Fitting such narrow peaks with a gaussian is only approximate. Therefore, the exact value of the board's time jitter equal to 5.2 ps (one sigma) should be treated with a grain of salt. More importantly, the width of the time peaks is entirely consistent with the intended use of the board.

Reference peak
Figure 3. The reference peak (with the board bypassed) is less than 2 channels wide.
 
 

The peak after the board
Figure 4. The width of the peak after delaying by the board has increased by about one TAC channel, i.e., 15 ps. The increase in "sigma" of the peak is about 5 ps.

The CAMAC  interface

After completion of the tests described above I soldered the CAMAC interface chip (a PQFP package with 208 pins). I used a 15-watt soldering iron from Radio Shack, which performed surprisingly well for its price of $9. Send me an e-mail if you are interested in details, because there are a few tricks to pay attention to when soldering such chips.

The chip was then programmed over the JTAG port with the help of the Cypress ISR cable. The CAMAC interface worked right out of the box. Two small mistakes in the interface logic were identified and corrected, and the chip was reprogrammed with a new version of the CAMAC interface logic.

The board responds to two CAMAC commands: F(16)Ax to set the delay word for the channel "x", x=0..15, and F(0)Ax to read the programmed delay word from channel "x". The CAMAC commands are the same as in the case of CAMAC discriminators. This will facilitate writing the Labview GUI.

The conclusion and the plans

The tests indicate that the equalizer board will perform its statutory duty. Several things need to be done:
  1. Assemble and debug the CAMAC interface using the current prototype (done, see above).
  2. Add a hardware OR to the board layout.
  3. Procure and test the 2nd (and hopefully final) version of the PCB.
  4. Assemble and test the actual production boards.
  5. Modify the Labview interface to interact with the board(s) to be installed in the tunnel.
  6. Install the appropriate number of the boards in the tunnel, connect to the Cerenkov and/or T0 detectors.
  7. Comission the whole system.
Questions? Comments? Send an e-mail to skulski@pas.rochester.edu .

Back to top of this page | Back to PHOBOS home page.

(C) Wojtek Skulski 2001.