FIC Status Page

 

Updates

11-01-01 We have been working extensively with Rick Kwarciany from FCC for several weeks and have made many modifications. It is still a work in progress.

Recently, at the test stand we found that the FIC, or Fiber Input Converter module was not functioning as we thought it should. On various tests we performed the FIC showed several different problems. This page is a summary of the tests we have performed along with those that we can still perform for more statistics, and the results of these tests.

Test #1 (FIC TESTING PROGRAM TEST)

The first test that was performed was back in May. This test was from a program that runs on Windows written by James Heinmiller from UIC. This test used a module called the TXGLINK sender built by Saclay. The TXGLINK sender was used to create data which was sent over G-link to The VTM and subsequently to the FIC. The FIC was then read out over VME and the data read was compered to the data sent. We commonly saw errors while performing this test. These errors were mostly duplicated words or skipped words read from the FIC.The pattern of data was a single word incremented by hex 0x1. The table below is a summary of the FIC's that were tested and which showed errors.

FIC# Ch 0 Ch 1 Ch 2 Ch 3 Explanation of errors or notes
1 No errors Errors Errors No errors This test was 10^8 bits or more, Ch 1 errors were duplicated words, Ch 2 errors were skipped words
2 No errors No errors No errors No errors This test was 10^8 bits or more.
3 No errors Errors No errors Not tested  
4-11 Not tested Not tested Not tested Not tested These FIC's have not been tested with this test
12 No errors Errors No errors No errors The errors in channel 1 were all duplicated words

Test #2 (FIC FIFO TEST)

This test was designed to test the FIC fifo's over VME and was written to run on the Alpha by Reinhard Schwienhorst. The program writes and reads from the FIC fifo's over VME two channels at a time. The test pattern in this case is a large fixed pattern. This was not tested extensively but the results are below. The FIC testing program written by James also has this functionality but was not used extensively.

FIC# Ch 0 Ch 1 Ch 2 Ch 3 Explanation of errors or notes
1 Errors Errors Errors Errors These errors were either single bit errors or repeat/skipped words
2 Errors No errors No errors No errors The errors were the same as mentioned above
5 Errors No errors No errors No errors These were only single bit errors
8 Errors Errors No errors No errors These were single bit errors

Test #3 (Questionable Method of Testing)

This test uses one Alpha that sends events over VME to the FIC which sends them to the MBT and compares them back at the Alpha. The table is below. This test is probably invalid because a couple days later we found that the FIC does not work well with an MBT in the same crate as was the case here.

FIC# Ch 0 Ch 1 Ch 2 Ch 3 Explanation of errors or notes
1 Errors Errors Errors Errors Repeated words
2 No errors No errors Errors Errors Missing words in event
3 Errors Errors Errors Errors Missing words
4         Not tested
5 No errors No errors No errors No errors  
6         Not tested
7 Errors Errors Errors Errors Repeated words
8 Not tested Not tested Errors Errors
9 No errors No errors No errors No errors  
10 Errors Errors No errors No errors Repeated words
11         Not tested
12 Errors Errors Errors Errors missing/repeated words

Test #4 (Alpha -> VME -> FIC -> hotlinks -> MBT -> MBus -> Alpha (compare))

This test was done with two programs called fic2send and mbtbc, both written for the Alpha by Reinhard Schwienhorst. The Alpha writes to the FIC fifo's over VME and the events of a known test pattern are transmitted to an MBT in another crate. The MBT is read out over MBus with another alpha. The table for this test is below.

FIC# Ch 0 Ch 1 Ch 2 Ch 3 Explanation of errors or notes
1 No errors No errors No errors No errors Test was run to 10^9 bits
2 No errors No errors No errors No errors "
3 No errors No errors No errors No errors "
4 No errors No events No errors No errors Ch 1 doesn't send events.
5 No errors No errors No errors No errors Test was run to 10^9 bits
6 No errors Errors Errors No errors Ch 1: bit 1 has errors. Ch 2: missing word errors
7 No errors No errors No errors No errors 10 ^9 bits
8 No errors No errors No errors Errors "
9         Not tested
10 No errors No errors No errors No errors 10^9 bits
11         Not tested
12         Not tested

Test #5 (Bit3 -> VME -> GSTM -> G-Link -> FIC -> hotlinks -> MBT -> MBus -> Alpha(compare))

This test was designed to write test events using the bit3 card to the GSTM which sends them over G-link to the FIC which then transmits to an MBT in another crate which is read out over MBus by an Alpha. The program that runs the GSTM is called GSTM CONTROL written by Jason Wennerberg. A shortcut to it is available on the desktop of D0NTMSU1. The program that reads out the MBT from the Alpha is called mbtbc. This table is complete and is below. Where not specified, the tests were conducted to 2.56*10^8 bits or 100000 of our test events.

FIC# Ch 0 Ch 1 Ch 2 Ch 3 Explanation of errors or notes
1 No errors No errors No errors No errors Tested to over 10^9 bits
2 No errors No errors No errors No errors Ch 0 tested to over 10^10 bits. Other channels tested to 10^8 bits
3 No errors No errors No errors No errors Tested to 10^8 bits
4 No errors Dead Channel No errors No errors Channel 1 doesn't pass any events, the rest were tested to 10^8 bits
5 Errors No errors No errors No errors Ch 0: error is 0x3ff apears as the first word of the event after a couple thousand events
6 No errors Not tested No errors No errors Ch 1: has a problem, with reframming the MBT and therefore the test won't run
7 Errors No errors Errors No errors Ch 0: error is 0x7ff as the first word, Ch 1: error is 0x3ff as the first word.
8 No errors No errors No errors Errors Ch 3: lots of single bit errors, stuck bit maybe?
9 No errors No errors No errors No errors Tested to 10^8 bits
10 No errors No errors No errors No errors Ch 0 was tested to 10^10 bits.
11 No errors No errors Errors No errors Ch 2: error is 0x3ff as the first word.
12 No errors No errors No errors No errors Tested to 10^8 bits

Some Error Analysis

On 8-17-01 we have found that FIC#5 does not fail on this test if emission of the FIC is turned off while the FIFO is being written and then on afterwards. This points to the error being a timing issue. I've also noticed that when the error we see (0x3ff as the first word) occurs, the Gofl LED on the front panel of the FIC is on. The FIC manual states that this LED goes on when a protocol error occurs in the FIC. (bit 16 not high in first word or bit 16 high inside the event).

Test #6 (Bit3 -> VME -> GSTM -> G-Link -> FIC -> Bit3(compare))

This test was started on 8-13-01. The test is called GSTM FIC VME Tester and is a modified version of the GSTM CONTROL program. The modifications make it so that the FIC is read out directly over VME, after being written to over G-link. This is made to continue Test #1, which was not finished because the french TXGLINK module had stopped functioning. The table will be updated as the test is continued.

FIC# Ch 0 Ch 1 Ch 2 Ch 3 Explanation of errors or notes
1 No Errors No Errors No Errors No Errors Tested to 2.56*10^8 bits
2          
3 No Errors No Errors No Errors No Errors Tested to 2.15*10^9
4          
5 No Errors No Errors No Errors No Errors Tested to 2.56*10^9 bits
6          
7          
8          
9          
10 No errors No errors No errors No errors Tested to 2.56* 10^8 bits
11          
12 No Errors No Errors No Errors No Errors Tested to 2.56*10^8 bits

Test #7 (Bit3 -> VME -> FIC -> hotlinks -> MBT -> MBus -> Alpha(compare))

This test was performed on FIC #7 which has been tested to 2.3*10^10 bits and has not failed. This test writes with the bit3 card over VME directly into the FIC FIFO's where the data is transmited to an MBT and subsequently to the Alpha where the patern is compared. This test shows us that the timing problem we see is generated somewhere in the interface between the VTM and the FIC fifo's.

 

Page created by Jason Wennerberg on 8-13-01

Last update at 2:45PM CDT 8-20-01

Questions comments: jasonw@fnal.gov