From: Lennart Osterman [Lennart.Osterman@hep.lu.se]
Sent: Friday, November 26, 2004 12:42 PM
To: Luciano Musa
Cc: Hans-Ake Gustafsson
Subject: Altro test statistics

Dear Luciano,

I have collected some figures from the ALTRO test.

From the total of 45167 tested chips, 37515 has been fully accepted.
7652 are considered bad in one way or the other.

 Of all the bad ones...
2194  has problem with the power, no other tests has been done

1171  has power OK but some error in the configuration registers,
      no other tests has been done

 570  has error in the pedestal memory, the tests above is OK, but the 
      tests belove are not considered

2072  has error in the event-memory, the tests above is OK, but the tests
      belove is not considered.

 707  has error in the data processing part, the tests above is OK, but 
      the tests belove is not considered.

 723  has error in the A/D converter, all tests above is OK, the test 
      belove is not considered

 168  has some other error, all above is OK.

 ~47  has some mechanical damage like bended pins or the like, and was
      never tested.

3133  has power and A/D converter OK. No other tests considered


Some comments;
 The test has been done in the following order: 
  1 Power
  2 Registers
  3 Pedestal memory
  4 Event memory
  5 Data processing
  6 A/D converter
  7 Miscellaneous

If the power test failed, the test was stopped. Nothing else was tested.

If the power was OK but some of the configuration registers was not working, no more tests was done. This means that there might be some chips with perfectly working A/D-converters among the 1171 chips that were rejected due to configuration register errors.

The test of the A/D converters is done in test-mode.

The miscellaneous test include things as a test of the 8-event memory, and a test with a sinus input and readout through the event memory, test of various error conditions, and some other things that don't fit into the other categories.

On the chips that have memory failure, I don't have information about if it is a single bit or a more fatal error. This information can be extracted from the log-files, but it requires some programming.

I still have about 700 chips that was rejected once. But some of them was rejected due to problems with the test equipment. This chips will be retested and I expect that about 30% of them will pass the test. This will of course change the figures above slightly.

I hope you can use the information above. If something is unclear don't hesitate to ask me.

Lennart