cb_pcidas64: Residual data samples at start of acquisition

Hi all,

Experimenting with comedilib/demo/mmap.c with Computer
Boards pcidas64/16.

Repeatedly executing "mmap" with "-nx", which sets the
channel count to x, with a dead short on channel 0
Analog input, I see the shorted channel's data change
position within the data scan.  In other words, the
short may appear as channel 0, I Control-C & rerun
mmap & I may see the short in column 10, or any other
column.  Clearly there is some residual data somewhere
offsetting the digital data by varying amounts.

For anyone that has not tried mmap, it produces a
tabular output to the CRT, one column per channel.  I
modified the output slightly, subtracting 32768 from
the A/D count values which, combined with range 0
(+-10V) range, makes the shorted channel stand out. 
The short gives about 10 A/D counts while the
remaining channels drift at several thousand counts.

I have found a work around however.  By executing
"mmap -n1", the next execution of mmap works properly.

Can anyone tell me how to clear all buffers & reset
all pointers prior to starting a continuous analog
acquisition command?

Setup:
Slackware 9.1, 2.4.25 kernel
comedilib & comedi from CVS except cb_pcidas64 taken
from comedi-0.7.67 since CVS version hung my system.

I'd appreciate the help folks,
Mike Reedy


	
		
__________________________________
Do you Yahoo!?
Yahoo! Photos: High-quality 4x6 digital prints for 25¢
http://photos.yahoo.com/ph/print_splash

Received on 2004-04-20Z01:03:36