Experimental Physics and
Industrial Control System

You can diagnose problems like this by turning on asynTrace on the underlying serial port.  In order to have the asynTrace output not impact the performance significantly you can send the output to a file:

 

# Send the asynTrace output to a file.  Substitute the name of your asyn port for the serial device below.

asynSetTraceFile “serial1”,0,”serial_output.txt”

# ASYN_TRACEIO_HEX

asynSetTraceIOMask “serial1”,0,4

# ASYN_TRACE_ERROR and ASYN_TRACEIO_DRIVER

asynSetTraceMask “serial1”,0,9

 

To turn off the debugging output

# ASYN_TRACE_ERROR

asynSetTraceMask “serial1”,0,1

 

To set the file back to stderr

asynSetTraceFile “serial1”,0,””

 

The output file will give you a series of time-stamped messages when messages were sent to the device and when replies were received.  That should help to determine if the serial I/O is the bottleneck.

 

Are you running at the fastest possible baud rate?

 

Mark

 

 

 

Hi,

We are using RS232.  Using the XFlash unit as configured through spec does not show the same delay as the IOC.  Disabling the Temperature monitoring PV (:ReadTemperature) which is on a 1s scan-rate, seems to alleviate the problem a bit.

Thanks,

-ZB

On 2/6/13 11:00 AM, Phillip Sorensen wrote:
> Hello,
>
> We are tring to setup the Rontec XFlash MCA with epics using the R7.2
> version of mca.
>
> The problem we are having is there seems to be a delay of some number of
> seconds after sending a command until we see an update in the MEDM
> screen.  For example clicking the start and stop buttons for the
> acquisition show this.  I we look at the XFlash unit it seems to update
> the front panel immediately, but the MEDM screen shows a delay.  The
> spectrum update (which is set to 1 second) also seems to take 5 to 10
> seconds per update.
>
> We are not getting any timeout errors.  Any suggestions?

Hi, Phil.

Over what physical link are you communicating with the XFlash?  If it's
RS-232, it could just take that long to transmit the data over the slow
link.

Lewis