Re: StreamDevice reading waveforms

[Anthony Pietryla <afp@anl.gov>]

Dirk,

I tried a form of that where the first (horizontal) record used I/O intr and forward linked to the vertical record. I would get one value in each waveform then I would receive the error about “surplus input”.

Tony

On Apr 13, 2015, at 9:05 AM, Dirk Zimoch <dirk.zimoch@psi.ch> wrote:

Hello Anthony,

In principle you can also use "I/O Intr" for the two longin records. They should trigger for each line. But I cannot guarantee this.

record (longin, "$(P)$(R)TBT_Horizontal_tmp")
{
 field (DTYP, "stream")
 field (INP,  "@spark.proto getTBT_X $(PORT)")
 field (SCAN, "I/O Intr")
 field (FNLK, "$(P)$(R)TBT_Horizontal")
}

Similar for vertical.

Terminator = "\n";

# only trigger reading
getTBT_XY {
 out "TBT_XY 3";
}

# processes the I/O Intr records once for each input line
getTBT_X_line {
   in "%i %*i";
}

getTBT_Y_line {
   in "%*i %i";
}




Unfortunately EPICS isn't very good at multi-dimensional arrays.

Dirk

On 13.04.2015 14:59, Anthony Pietryla wrote:

Hi Dirk,

Thank you for your reply. Unfortunately, there are many more than three
rows to read, up to 256k samples/channel. I used a value of three for my
debugging.

One suggestion I received was to read everything into one huge array and
use an asub record to parse the values.

Thanks,
Tony

On Apr 13, 2015, at 3:54 AM, Dirk Zimoch <dirk.zimoch@psi.ch
<mailto:dirk.zimoch@psi.ch>> wrote:

Hello Anthony,

For waveform records StreamDevice repeats the format for each element
and optionally reads a separator in between. While this allows to read
a row, it cannot read a column. And it probably does not work with
redirection (I have never tried).

This makes it a bit complicated to read your device. But hopefully not
impossible. You need a few more records.

I understand that the reply always consists of 3 rows. Then I would
try the following:

Have two longin records, one for X and one for X.
Redirect to these records as you do now, but do it 3 times:

Terminator = "\n";
getTBT_XY {
  out "TBT_XY 3";
  in   " %(\$1TBT_Horizontal_tmp)d %(\$1TBT_Vertical_tmp)d";
  in   " %(\$1TBT_Horizontal_tmp)d %(\$1TBT_Vertical_tmp)d";
  in   " %(\$1TBT_Horizontal_tmp)d %(\$1TBT_Vertical_tmp)d";
}

This processes the two longin records 3 time each. For the waveforms
use two compress records and do a FLNK from the longin records to the
compress records:

record (longin, "$(P)$(R)TBT_Horizontal_tmp")
{
 field (FNLK, "$(P)$(R)TBT_Horizontal")
}

record (compress, "$(P)$(R)TBT_Horizontal")
{
 field (INP,"$(P)$(R)TBT_Horizontal_tmp")
 field (ALG, "Circular Buffer")
 field (NSAM,"3")
}

The same for Vertical.

I am not quire sure about the compress record settings. Maybe you need
to try a bit.
Maybe you may want to write to the .RES of the compress before
starting. In that case you can use additional seq records.

Hope this helps.

Dirk

On 09.04.2015 23:01, Anthony Pietryla wrote:

I have a device which uses SCPI protocol and returns sets of waveforms
in columns where I can request how many rows to return. For instance I
can query the device for 3 values of position data and it will return 2
columns (X & Y) and 3 rows. I want to read the columns into separate
waveform records. I am having problems with the protocol file for which
I need help.

The database records and the protocol file are listed below. When I
process the record only the first values (first row of data) are written
to each record. With asyn debugging turned on the console shows all
three sets of data:

epics> 2015/04/09 13:52:36.877 L0 wrote
TBT_XY 3

2015/04/09 13:52:36.878 L0 read
 2000   6000
 7143   4286
 4286   1429

caget shows only the first value stored:

caget -# 5 B:Spark:TBT_Horizontal
B:Spark:TBT_Horizontal 5 3333 0 0 0 0

caget -# 5 B:Spark:TBT_Vertical
B:Spark:TBT_Vertical 5 -1111 -3333 0 0 0

*******************
The protocol file has:

getTBT_XY {
  Separator = "\n";
  out "TBT_XY 3";
  in   " %(\$1TBT_Horizontal)d %(\$1TBT_Vertical)d";
  ExtraInput = Ignore;
}

**************
The database records are:

record(waveform, "$(P)$(R)TBT_Horizontal")
{
  field(DESC, "TBT Horizontal")
  field(DTYP, "stream")
  field(INP,  "@spark.proto getTBT_XY($(P)$(R)) $(PORT)")
  field(NELM, "262144")
  field(EGU,  "um")
  field(FTVL, "LONG")
}

record(waveform, "$(P)$(R)TBT_Vertical")
{
  field(DESC, "TBT Vertical")
  field(DTYP, "Soft Channel")
  field(NELM, "262144")
  field(EGU,  "um")
  field(FTVL, "LONG")
}

********************
When I comment out the “ExtraInput = Ignore;” line in the protocol file
I get error messages:

2015/04/09 15:53:58.170 L0 wrote
TBT_XY 3

2015/04/09 15:53:58.170 L0 asynOctetBase interrupt
2015/04/09 15:53:58.171 L0 read
 3333  -1111
-3333  -5556
-2000      0

2015/04/09 15:53:58.271 L0 B:Spark:TBT_Horizontal: 29 bytes surplus
input "<0a> -3333  -5556<0a> -200..."
2015/04/09 15:53:58.271 L0 B:Spark:TBT_Horizontal: after 13 bytes:
"  3333  -1111"

caget -# 5 B:Spark:TBT_Horizontal
B:Spark:TBT_Horizontal 5 3333 0 0 0 0

caget -# 5 B:Spark:TBT_Vertical
B:Spark:TBT_Vertical 5 -1111 -3333 0 0 0

--
Anthony Pietryla
Principal Controls Engineer
Advanced Photon Source
Argonne National Laboratory

Phone: 630-252-7430
Fax: 630-252-6123