Subject: 2nd/End chip problem

 The "End" chip, whichever the number of chips on hybrid,  
must send the "end" signal, an unique trailer pattern. This
should be issued when the chip is configured as "End",  
however, due to a bug in the 1st version of the ABCD logic,
this "end trailer" can be generated when the "datain0/  
datain0B" is tied to logical "0/1", or "DGND/VDD". We have
  bonded the "datain0/datain0B" of the 2nd chip (which is 
the "End" chip) to "DGND/VDD" and then the DAQ chain works.
Thanks, Daniela.

					Y.Nobu UNNO

===========================================  From macina@
   cern.ch Fri Jun 26 20:14:01 1998 Date: Fri, 26 Jun 1998 
12: 45:42 +0200 (METDST)  
From: Daniela Macina <macina@cern.ch>
To: UNNO Yoshinobu <unno@post.kek.jp> 
Subject: Re: Reading two chips on the hybrid

Dear nobu,
I think that your problem is due to a bug in the abcd logic
. A slave chip doesn't send the trailer and so the daq  
hungs up. What you should do is to put a third chip. In  
this case you will be able to read the second chip. Here it
is the mail I sent around to explain in detail what the  
problem is. Anyway putting one more chip should do the  
trick. If you don't have a third chip or if the hybrid is  
fully populated you can apply the trick of connecting the  
datain to gnd and datainB to VDD through a resistor as  
explained in the following mail.
               Good luck
                  daniela

---------- Forwarded message ----------
Date: Mon, 4 May 1998 18:01:56 +0200 (METDST) From: Daniela
Macina <macina@cern.ch> To: Peter Phillips <P.W.Phillips@   
rl.ac.uk>,
    Gareth Moorhead <moorhead@liszt.ph.unimelb.edu.au>,
    "tim.dubbs" <dubbs@scipp.ucsc.edu>
Cc: Wladyslaw DABROWSKI <dabrowsk@sunvlsi.cern.ch> Subject:
abcd trailer problem

Hello everybody,
 I have talked to Daniel La Marra and Francis Anghinolfi  
about the "last chip" trailer problem. Daniel has made some
simulations and he found the same problem we found. The  
explanation is the following:

 unfortunately the internal logic foresees that the data  
entry has to be zero during the trailer transfer. If the  
entries of the circuit that receives the "datain" lines are
not connected, the internal status of these lines will be  
undefined and they can be 0 or 1.

 Since the datain lines and the datainbp (datainbypass )  
lines have different layouts, it may be that the datainbp  
lines are set to zero and the datain lines are set to 1.  
This explains why Gareth, using the datainbypass lines in  
the last chip, got the chain working. Anyway the success of
this method will depend on the chip itself (in my case  
doesn't work).

 According to Daniel and Francis if we force the datain to
  be zero, as suggested by Tim, we should succeed in 100% 
of  the cases. We should tie datain to gnd and datainB to 
VDD.  However with this method we get about 40 mA more 
current on the hybrid, due to the presence of a resistor of
about 120  ohm in the datain and datainB lines. Anyway this 
shouldn't  be a big problem. Francis suggested me to try 
first with  just one bond between datain and gnd. If this 
is not enough , then I should bond datainB to VDD.

 Let me know if this is somehow unclear. By tomorrow  
morning I should have 6 chips on the hybrid and I'll try  
the 3 methods (bypass lines, datain bond to gnd and datain
  bond to gnd + datainB bond to VDD).
                           Ciao
                                 Daniela