RE: StreamDevice communication problems

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

There is something I don't understand about your database.  The read_volt record is scanned a "1 second".  But it also has a forward link (FLNK) to itself?  That seems strange.  Why are you doing that?

Mark

________________________________
From: tech-talk-bounces@aps.anl.gov [tech-talk-bounces@aps.anl.gov] on behalf of Tito Körner [tito.koerner@gmail.com]
Sent: Monday, August 26, 2013 10:33 AM
To: tech-talk@aps.anl.gov
Subject: StreamDevice communication problems

Hello,

I am using Streamdevice (http://epics.web.psi.ch/software/streamdevice/doc/index.html) to control a high voltage box. The box has 5 levels with 8 channels each, so it controls 40 channels. The boxes receive commands and send output via telnet. Currently I only have two records - one for reading one for writing values. To access all channels I use a template. The problem is whenever I give one channel the command to change the value - the entire traffic gets mixed up:

My .db looks like this:

record (stringout, "$(P):BOX:$(BOXNO):$(LEVELNO):$(CHANNELNO):set_volt")
{
    field (DTYP, "stream")
    field (OUT,  "@$(PROTO) set_voltage($(LEVELNO),$(CHANNELNO)) termBox$(BOXNO)")
}

record (ai, "$(P):BOX:$(BOXNO):$(LEVELNO):$(CHANNELNO):read_volt")
{
    field (DTYP, "stream")
    field (INP, "@$(PROTO) read_voltage($(LEVELNO),$(CHANNELNO)) termBox$(BOXNO)")
    field (SCAN, "1 second")
    field (FLNK, "$(P):BOX:$(BOXNO):$(LEVELNO):$(CHANNELNO):read_volt")
    field (PACT, "")
    field (STAT, "")
    field (SEVR, "")
    field (VAL, "")
}

My protocol file looks like this:

terminator = CR LF;
LockTimeout = 10000;

read_voltage {
    out "read_adc floats_csv \$1 \$2";
    in "%*d.%*d.%*d %*d:%*d:%*d,%f";
    @mismatch{out "read_adc floats_csv \$1 \$2";  in "%*d.%*d.%*d %*d:%*d:%*d,%f";}
}

set_voltage {
    out "SetVpmF \$1 \$2 %s";
    in "DAC: \$1   CH: \$2   U: %*f  Uk: %*f";
    in "Nachregelung aktiv!";
    in "Messwert: %*f   DAC: \$1   CH: \$2   Value: %*f   korrigiert: %*f";
    in "Messwert: %*f   DAC: \$1   CH: \$2   Value: %*f   korrigiert: %*f";
    in "Messwert: %*f   DAC: \$1   CH: \$2   Value: %*f   korrigiert: %*f";
    in "Messwert: %*f   DAC: \$1   CH: \$2   Value: %*f   korrigiert: %*f";
    in "Messwert: %*f   DAC: \$1   CH: \$2   Value: %*f   korrigiert: %*f";
    in "Messwert: %*f   DAC: \$1   CH: \$2   Value: %*f   korrigiert: %*f";
    in "Messwert: %*f   DAC: \$1   CH: \$2   Value: %*f   korrigiert: %*f";
    in "Messwert: %*f   DAC: \$1   CH: \$2   Value: %*f   korrigiert: %*f";
    in "Messwert: %*f   DAC: \$1   CH: \$2   Value: %*f   korrigiert: %*f";
    in "Messwert: %*f   DAC: \$1   CH: \$2   Value: %*f   korrigiert: %*f";
}

I created a python script that prints all values to the terminal. As long as only the read record is activ everything works out. But when I use another terminal to set a voltage like: 'caput CB:HV:BOX:19:0:0:1480:set_volt' the voltage values start to  'jump around'. I monitored the network traffic with wireshark and saw this:

> read_adc floats_csv 2 5
01.01.1900 04:23:55,1345.003
> read_adc floats_csv 2 6
01.01.1900 04:23:55,1345.123
> read_adc floats_csv 2 7
01.01.1900 04:23:55,1345.156
> read_adc floats_csv 3 0
01.01.1900 04:23:56,1346.075
> read_adc floats_csv 3 1
01.01.1900 04:23:56,1343.827
> read_adc floats_csv 3 2
01.01.1900 04:23:56,1376.977
> read_adc floats_csv 3 3
01.01.1900 04:23:56,1346.537
> read_adc floats_csv 3 4
01.01.1900 04:23:56,1341.462
> read_adc floats_csv 3 5
01.01.1900 04:23:56,1368.079
> SetVpmF 0 0 1480
read_adc floats_csv 3 6
DAC: 0   CH: 0   U: 1480.000  Uk: 1584.800
Nachregelung aktiv!
Messwert: 1479.545   DAC: 0   CH: 0   Value: 1480.350   korrigiert: 1585.165
Messwert: 1479.941   DAC: 0   CH: 0   Value: 1480.395   korrigiert: 1585.212
Messwert: 1479.935   DAC: 0   CH: 0   Value: 1480.445   korrigiert: 1585.264
Messwert: 1480.045   DAC: 0   CH: 0   Value: 1480.410   korrigiert: 1585.228
Messwert: 1480.046   DAC: 0   CH: 0   Value: 1480.375   korrigiert: 1585.191
Messwert: 1479.941   DAC: 0   CH: 0   Value: 1480.420   kread_adc floats_csv 3 6
orrigiert: 1585.238
Messwert: 1480.052   DAC: 0   CH: 0   Value: 1480.379   korrigiert: 1585.196
Messwert: 1479.963   DAC: 0   CH: 0   Value: 1480.408   korrigiert: 1585.226
Messwert: 1480.063   DAC: 0   CH: 0   Value: 1480.359   korrigiert: 1585.175
Messwert: 1479.963   DAC: 0   CH: 0   Value: 1480.388   korrigiert: 1585.205
> read_adc floats_csv 3 7
01.01.1900 04:23:59,1344.063
> read_adc floats_csv 4 0
01.01.1900 04:23:59,1344.070
> read_adc floats_csv 4 1
01.01.1900 04:23:59,1346.169
> read_adc floats_csv 4 2
01.01.1900 04:23:59,1357.607
> read_adc floats_csv 0 0
01.01.1900 04:24:00,1352.217
> read_adc floats_csv 0 1
01.01.1900 04:24:00,1340.962

As you can see (marked bold) in between the command 'SetVpmF 0 0 1480' and the response from the box there are two out commands from the read_volt record. The problem now is - the first response from the box after the response to the set_voltage command is '01.01.1900 04:23:59,1344.06. This is actually the voltage value belonging to the command 'read_adc floats_csv 3 6'. But because the read_volt record sent this command in between the command and response of  the set_volt record, Epics 'thinks' this is the value for 3 7. At the StreamDevice website it says that a protocols first out command locks the device for exclusive access until the protocol is finished. But here this does not seem to work because obviously the read_volt record sends the out command of its protocol while the protocol called in the set_volt record is active. I hope someone can help me to figure out what's going wrong and how to fix it.

Regards,
Tito Koerner