Fix broken DCIMs

The problem:
There are a variety of small and large problems with different DCIM boards. For example, stuck bits, bad packet format, Glink does not lock on some channels, etc. Some of these problems (e.g. the stuck bits) are easy to fix, some may be hard to fix. We need to fix more of the boards for year-3 running.

What skills are required:
Probably anyone with some electronics skills can fix many of the problems. I am worried that the fact that the top channel in most DCIMs has problems with Glink lock is something wrong in layout. This problem may be harder to solve.

Time estimate:
~1 month to fix the easier stuff, unpredictable for the Glink lock problem in the top channel, but I'll guess another ~1 month.

Some details:
Records of the Q/A testing of DCIM boards are kept here and the test procedure is described here. Actually, the test procedure is not exactly what we usually do, but it will give you the basic idea of what needs testing. In fact, we have found that the only reliable way to test DCIM modules is with a more or less complete setup including:

  • a full crate of DCIM modules + a TCIM module
  • a few real MCMs to supply data packets
  • some DCMs
  • access to the standalone version of the PHENIX DAQ
  • a LV power supply similar to the one used in the IR
    We have most of the parts for such a setup in the side room of bldg 1008. We will have to remove a crate full of DCIMs + a TCIM from the IR. We may have to make a few test cables. We need a more or less complete setup for tests because some of the clock related problems depend on the number of channels connected and/or the current being drawn from the power supplies.

    I have found that stuck bits are normally caused by broken traces or solder bridges. These are usually around the Xilinx FPGAs. Therefore, the job requires good eyes, steady hands, and patience in addition to some electronics skills.

    John Sullivan
    last update: 16-Jan-2002