Log entry time 02:30:28 on October23,2009

Entry number 296184

01:10 Sasha is doing Moller studies, so I take the chance to
test the ADC18 code.

The safe, roll-back versions are happex3_ts_ok5.crl and
hapAdc18Lib_ok2.c.  They will be restored afterwards, regardless
of the outcome, per agreement with Kent.

With the new codes, the ADC18 parameters are set in prestart.
The reason is that a hard-reset is done in prestart, so one
must program them after that.

Run 13534 ... Ovs=0.  

Run 13535 ... Ovs=1.   No beam at the moment.
Rampdelay=0, Integ=6600
Rate looks ok.  No deadtime.

Run 13536 ... Ovs=1,  Rampdelay=0, Integ=6400
Similar to previous run, smaller integ time.

Run 13537 ... Ovs=3.  Rampdelay=0, Integ=3200
Rate = 128 Hz and no deadtime.  Looks good.

With ovs=3 (oversampling by 4):
The pedestal on det2 goes down a factor of 4 as expected
but the pedestal became a bimodal distribution (fig 1).
Meanwhile a nearby empty channels looks normal (fig 2)
so I doubt it's a ADC18 problem per se.  
Another note is that only the first event is "bad", and
this is expected because the "go" occurs on first event.
No other "bad" events, no deadtime, etc.

Run 13538 is with ovs=9 and Integ=1300.  Rate is 300 Hz.  Good.
Run 13539 is the same.

For ovs=9 :
The det2 ped becomes 700, which is right.  Some other channels
look strange --- e.g. roc23 slot0 chan0 (bcm2) has base=peak=0.
Will investigate a bit more ...