RE: Asyn Delay

["Mark Rivers" <rivers@cars.uchicago.edu>]

Hi Rob,
 
> Unless I can set up more than one port for this device I will have to
> live with the >1 sec read rate?

I think that's correct.  Since this is a TCP/IP device you should be able to use a second asyn port, and get around that problem.  Then I guess its a question of what the VXI-11 device does at the other end when there are multiple requests arriving.
 
If this were a local serial port you would have no choice but to keep the read rate > 1 second.
 
Mark
 

________________________________

From: Robert Emery [mailto:duma@u.washington.edu]
Sent: Wed 5/16/2007 7:37 PM
To: Mark Rivers
Cc: tech-talk@aps.anl.gov
Subject: Re: Asyn Delay



Mark,

My "read" records were each processing at 1Hz.  If I slowed down the
"read" rate, I would still see a delay problem with my "set" messages
and that problem would appear at the "read" scan rate.

After I changed the "read" records to be linked with FLNKs I could see
(using wireshark) that "set" messages were able to go out between the
"read" messages and my delay problem was gone as long as I kept the scan
rate of the first "read" record (which started the chain of the FLNKs)
slower that 1 sec.

At 1 sec., the FLNKd  chain of 12 "read" records (each record has to
wait about 50 msec for a response) can not finish before the first one
starts the chain again.  From the messages that come out it looks like
the message queue is backing up which causes delays for my "set"
messages and causes my "read" messages to also be delayed and out of
sequence.

Unless I can set up more than one port for this device I will have to
live with the >1 sec read rate?

Thanks for your help.

Rob

Mark Rivers wrote:
> How are your "read" records being processed.  Is each processing at 1Hz, or is it a set of records linked with FNKL links, with only the first record scanning at 1Hz.
> 
>   If it they were linked then it seems to me that the "set" messages would have to wait for at most one "read" request to process before they went to the head of the queue (since there should be at most 1 read request in the queue.  But without testing it I can't be sure.
> 
> Mark
> 
>
> ________________________________
>
> From: Robert Emery [mailto:duma@u.washington.edu]
> Sent: Mon 5/14/2007 5:26 PM
> To: tech-talk@aps.anl.gov
> Subject: Asyn Delay
>
>
>
> I am using ASYN to communicate with 12 power supplies via an E5810
> Ethernet/GPIB gateway from a Linux IOC.  I am periodically (1 sec)
> reading 24 parameters from the supplies (voltage and current for each
> supply) and am setting the current (3 supplies at a time) when the
> operator turns a dial at the console.  The problem that I am having is
> that when the current/voltage requests are processed I see about a 1 s
> delay in the response to the current set messages.
>
> It takes the power supplies about 50 ms to respond to a request for
> current or voltage information.  If these requests are backing up in the
> ASYN port queue, I would make sense that I would see this type of delay.
>
> Can I speed up the processing of my current set messages by setting up a
> second ASYN port on the same IOC connected to the same gateway?  With a
> higher priority?  Or is there a better way to speed up my set messages?
>
> Thanks
> Rob Emery
> UW Medical Center
>
>
>
>