Experimental Physics and
Industrial Control System

Pavel,

I don't think I saw an answer to this message from Rod, and I agree with him, your calculation of the size seems incorrect. How did you get 70 KB for a waveform?  Tektronix scopes can send the waveforms in binary with 1 byte per channel or 2 bytes per channel.

In fact I just did a test with a Windows machine to a Tektronix TDS380 scope.  This is not the same model you have, but it is similar.  Its waveform is 1000 points, rather than 2500.

I connected COM1 on the PC to the scope and used an EPICS asyn record to communicate running on a Windows EPICS IOC.  I set the baud rate to 19200.  I used an IDL script to send the commands and read the data back all through the asyn record.  The IDL script is basically the one contained in the asynRecord documentation here:

http://www.aps.anl.gov/epics/modules/soft/asyn/R4-18/asynRecord.html#Example2

I have attached the actual script I used.

Here is the IDL command I used to run the script.  It gets the system time (in seconds as a double), reads the waveform from the scope, gets the time again, and prints the elapsed time:

IDL> startTime=systime(1) & read_tds200, stop=1000, '13TEST:asyn1', data & print, 'Time=',systime(1)-startTime

Here is the output:



Time=      0.66956306



So it only took .67 seconds!

I then query and plot the data
DL> help, data
DATA            LONG      = Array[999]
IDL> plot, data

The waveform looks as expected.  The total number of bytes transfered during the waveform read was 1006 (each channel is encoded as 1 byte; one can ask for 2-bytes to get more resolution, but I don't think you need it for your purposes.

Note that 1006 bytes at 19200 baud should take about 0.5 seconds, very similar to what I measured, 0.669 seconds

If we scale this up to your scope with 2500 channels it would take only about 1.7 seconds per channel.  Multiplying by 4 channels=6.7 seconds.  You said it takes 10-15 seconds for the screen dump. So it would actually be 2-3 times faster to read the actual waveforms. Plus you can plot the data directly in medm or CSS widgets.

Mark

________________________________________
From: [email protected] [[email protected]] on behalf of Rod Nussbaumer [[email protected]]
Sent: Friday, May 04, 2012 9:23 AM
To: Pavel Masloff

Cc: EPICS Tech Talk
Subject: Re: [Scopes] BMP image record??

Pavel:

We must be talking about different things. If your waveform is 2500
points, and takes 73.5 KB, that works out to 29.4 bytes per point.
Sounds like very high resolution! Even if that waveform was transferred
in some ASCII format, that is still very verbose.

I've never tried to measure the GPIB throughput, but I am fairly certain
that the overall throughput in any case that I've encountered is limited
by the scope itself, and the operation of the scope in terms of its
triggering setup and the signal it is measuring. On some scopes (eg TDS
380), I can swamp it's CPU with GPIB communications, and effectively
cripple its functionality. That is an old scope, and I doubt that modern
scopes would be similarly afflicted. Perhaps the hardcopy function would
be a good case for a throughput test.

   ---   rod.


Pavel Masloff wrote:
> Hi Rod,
>
> We have 2500 points per waveform. Our PCs are equipped with COM ports
> only, no GPIB cards. One waveform is 73,5 KB to be exact. The RS232
> connection speed is maximum 19200 Kb/s at maximum. For the time being I
> just want to check whether the experiment went fine, no false
> triggering, etc. In the future, I would like to analyze the waveforms in
> matlab via CA. So this will be handy.
> What is the speed of the GPIB interface on Tek scopes, anyway?
>
> On Fri, May 4, 2012 at 1:42 AM, Rod Nussbaumer <[email protected]
> <mailto:[email protected]>> wrote:
>
>     Pavel:
>
>     Every Tektronix scope I've set up communications with had a 500
>     point waveform (as well as some substantially larger ones).
>     Presently, I'm acquiring those at 10 waveforms per second (over
>     GPIB). Are you sure that would not suffice for you? Is GPIB an
>     option? Much much faster.
>
>
>        ---   rod.
>
>
>     Pavel Masloff wrote:
>
>         Mark,
>
>         A waveform weighs 70kB. 4 waveforms = 280 kB. 280 vs 40. I would
>         have to
>         wait a couple of minutes till the data is processed. 10-15
>         seconds for
>         the screen dmp.
>
>
>
>         On Thu, May 3, 2012 at 11:13 PM, Mark Rivers
>         <[email protected] <mailto:[email protected]>
>         <mailto:[email protected].__edu
>         <mailto:[email protected]>>> wrote:
>
>             It may be time to back up a step.  Why not transfer the
>         waveforms
>             themselves, rather than a screen dump?  There are others in the
>             EPICS community who have already done that for the scopes
>         you are using.
>
>             __ __
>
>
>
>
> --
> Best regards,
>
>
> Pavel Maslov, MS
> Controls Engineer at Pulsed power Lab
> Efremov Institute for Electro-Physical Apparatus
> St. Petersburg, Russia
>
> Mobile: +7 (951) 672 22 19
> Landline: +7 (812) 461 01 01