RE: Gentec S-link power meter problem with ending I/O interrupt measurement

[bohmm <bohmm@fzu.cz>]

Hi Tomáš,
 I didnt specify some important things about the setup, you are right:

I am using Ethernet device attached to a PC where IOC with Stream is 
running. The device is set up to trasmit data and the data exchange 
speed depends on the frequency of external trigger

I have an EPICS record configured SCAN="I/O Intr", DTYP=stream and that 
has only the  "in" command defined in the .proto file, as you said.

The amount of data:
 Just doubles (values of power measurement) received at the frequency of 
the external trigger (for us no more than 10 Hz).

When in Interrupt mode, the device receives the data automatically at 
the frequency of the ext. trigger.
Interrupt mode for this power meter means, that device receives the data 
whenever external trigger occurs, on the other hand in polling mode, you 
first send a request to the device ("out" .proto) and obtain a response.
(Here I am not sure what yout meant so - Data also can be received 
non-periodically in interrupt mode, depends on the nature of ext. triger 
signal)

I think this is not the case of fast rate data transfers.
I should also mention, that when doing a telnet to the Gentec directly, 
and typing in the command to stop the I/O interrupt mode, everything 
works fine.
Peter Owens, the original author of the stream protocol (which otherwise 
works perfect) also says he has seen cases where a busy device can't 
handle a serial port - but works fine over an Ethernet port, which he 
thought is this case.

I hope this helps to clarify

Thanks
Marek


Dne 2015-04-16 13:37, Mazanec Tomáš napsal:
> Hi Marek
>
> I'm not sure if I understand your setup correctly, so...
>
> you're using RS232 device attached to a computer where an IOC with
> streamDevice is running in order to acquire data from it.
> If the device is set to transmit data, it does that at 10Hz rate.
> An EPICS record is configured with SCAN="I/O Intr", DTYP=stream....
> and reads the data using streamDevice protocol command which has only
> "in" command defined.
>
> If so, it should work smoothly.
> Of course, it depends on data. What is the amount of data and what
> structure it has ?
> And then it depends on serial line speed.
>
> Which of the device modes (Polling, Interrupt) sends the data
> automatically and periodically at 10Hz ?
> Does the interrupt mode means the data are sent in the same way as
> poll but non-periodically ?
>
> I've checked S-LINK's product webpage. It mentions some "Fast rate
> data transfers to PC" at 10kHz rate. Is this the case ?
>
> I'm interested, we have got one S-LINK in lab too. However its
> integration to EPICS may not happen at all.
>
> Thanks,
> Tomas Mazanec
>
>
> ________________________________________
> Od: tech-talk-bounces@aps.anl.gov [tech-talk-bounces@aps.anl.gov] za
> uživatele bohmm [bohmm@fzu.cz]
> Odesláno: 14. dubna 2015 12:07
> Komu: Tech talk
> Předmět: Gentec S-link power meter problem with ending I/O interrupt
>  measurement
>
> Hello,
>   I am an EPICS newbie and trying to implement statistics functions for
> our energy meter. Our laser system is running at 10 Hz and the energy
> meter has to catch every pulse energy/power value for the statistics to
> be computed properly. So the energy meter is running with an external
> trigger. I am communicating with the device via StreamDevice using
> serial commands.
>
>   The device supports two acquisition modes: Polling ("Free running") 
> and
> I/O interrupt (device is generating interrupts).
>
> The problem is, once I get into I/O Interrupt mode @ 10 Hz, the device
> is so busy receiving the data, that any other byte streams send to it
> via StreamDevice do not get through. The consequences are that I am not
> able to send the command to end the I/O Interrupt mode once I started 
> it
> and get back to Polling ("free running"), and also that the devices 
> does
> not receive its other statuses.
>
> At really low repetition rates (@ 0.5 Hz) this problem does not occur,
> the device has "time to answer".
>
> Did anybody deal with a similar problem? Is it possible to solve 
> without
> writing a driver?
>
>
> Best regards
>
> Marek Böhm

--
___________________________________
Ing. Marek Böhm, RP2 Control Engineer

Institute of Physics AS CR, v. v. i.

HiLASE Centre

Za Radnicí 828

252 41 Dolní Břežany

M:    +420 606 259 414

T:    +420 314 007 726

E:      bohmm@fzu.cz

W:    http://www.hilase.cz <http://www.hilase.cz/>

______________________________________________________________________