LV5.1, NT4, serial ports, and CPU load

[L. Ravi Narasimhan]

Hello,

I have looked through old threads on LV-serial interfacing under NT but
am not able to resolve the following point.

I have an instrumentation driver supplied by EG&G to communicate with
one of their lock-in amplifiers via the serial port (4800 baud.)

ftp://ftp.egginc.com/divisions/iras/SigRec/egg510x.llb

I run LV5.1 under NT4, SP4 on a 350MHz Dell with 128MB RAM and 6GB hard
disk. There is an NI PCIMIO 16E4 DAQ card in the system, as well.

Problem is that the serial port VI takes 100% of the CPU according to
task manager and this does not seem right to me. Buffered I/O from the
DAQ card is not possible concurrently with the serial port VI. One or
the other will run, but not both. Asynchronous acquisition
_will_ work
with the serial VI.

Is this a fundamental problem with NT or do I need to reconfigure
something? Based on past threads, I have looked at my system
configuration and drivers. All seem to be correct and the most current
version.

Thanks,

--- Ravi Narasimhan


--
Ravi Narasimhan
Dept. of Physics and Astronomy, UCLA
http://www.physics.ucla.edu/~oski

[Mike Ayers]

> Problem is that the serial port VI takes 100% of the CPU according to
> task manager and this does not seem right to me. Buffered I/O from the
> DAQ card is not possible concurrently with the serial port VI. One or
> the other will run, but not both. Asynchronous acquisition _will_ work
> with the serial VI.

I faced the same problem - even worse (trying to asynchronously work
with two serial ports read/write). None of the stock VIs handle this
correctly. My solution was a double loop - the inner loop uses "Bytes
at Serial Port.vi" and "Wait (ms)" to poll the serial port in
nonblocking fashion and avert the dreaded lockup, and the outer loop
reads a character at a time and concatenates it to the shift registered
result string, while checking for EOL. The result w
orks beautifully,
and should be able to support as many serial ports as you care to
enable.


/|/|ike