Experimental Physics and
Industrial Control System

Dirk Zimoch <[email protected]> · Thu, 31 May 2012 17:45:54 +0200

Pavel,

I think I know what's happening.

If the device sends \n\r and you define your terminator as \n, then the

\r is still in the queue.

Normally this does not harm because StreamDevice flushes all input

before sending a new output (command).

But sometimes, the computer is too fast. The \n may be still on the wire

when StreamDevice tries to flush. Thus it arrives after the flush and

appears as the first byte if the next input.

Thus setting your terminator properly (either in asyn or in

StreamDevice) should help.

Note that InputEos of asyn is limited to 2 bytes (1 byte for most GPIB

devices). StreamDevice supports arbitrarily long inTerminators.

When taking a new device into operation, I usually test it without any

terminator setting (letting the read timeout terminate the input). Then

I can see what type of terminator the device really sends (which may be

different to the documentation!) and use that as the inputTerminator.

Dirk

Mark Rivers wrote:

I don’t know about the output EOS, that depends what the scope expects.

But it looks like it is definitely sending you \n\r, so you should set

the InputEos to “\n\r”.

Mark

*From:* Pavel Masloff [mailto:[email protected]]
*Sent:* Wednesday, May 30, 2012 9:09 AM
*To:* Mark Rivers
*Cc:* Dirk Zimoch; EPICS Tech Talk
*Subject:* Re: [StreamDevice] parsing rapid inputs

Hi Mark,

asynOctetSetInputEos("${PORT1}", -1, "\n")
asynOctetSetOutputEos("${PORT1}", -1, "\n")

Should I set them both to "\n\r"?

On Wed, May 30, 2012 at 6:00 PM, Mark Rivers <[email protected]

<mailto:[email protected]>> wrote:

How have you set the input EOS on your asyn port? Have you overridden

this by explicitly defining a terminator in your stream protocol file?

If you define the terminator as “\n\r” for the asyn port and don’t

define it in your protocol file then I don’t think you should have this

problem, because the read won’t complete until it finds and consume the

\r (CR) character.

Mark

*From:* [email protected]

<mailto:[email protected]>

[mailto:[email protected]

<mailto:[email protected]>] *On Behalf Of *Pavel Masloff

*Sent:* Wednesday, May 30, 2012 3:15 AM
*To:* Dirk Zimoch; EPICS Tech Talk
*Subject:* [StreamDevice] parsing rapid inputs

Hello,

I have run into a problem lately. I have sort of a slow device - a

TDS2024 scope. So when I run my IOC, I expect StreamDevice to initialize

my PVs. So there are ~ 20 PVs I want to initialize. And every once in a

while I get the following error message in epics shell:

/2012/05/30 13:02:25.211 L2 TDS2024:probe_ch2: got "<0d>:CH2:PROBE 1"

where ":CH2:PROBE " was expected

/This <0d> character (or CR) is driving me mad. This is because it was

supposed to be the terminator of the previous input (here is the whole

string together with the one above):

/2012/05/30 13:02:25.124 /dev/ttyUSB0 read 13
:CH1:PROBE 1\n

2012/05/30 13:02:25.211 L2 TDS2024:probe_ch2: got "<0d>:CH2:PROBE 1"

where ":CH2:PROBE " was expected/

As you can see, there is an \r missing. So my <0d> comes as a part of

the input / ":CH2:PROBE 1" /- so I get /"<0d>:CH2:PROBE 1" /instead

which causes the error.

How can I fix this?

I have tried using the @mismatch handler, but it doesn't work within the

@init handler.

And you will never know which PV will not initialize next.

As an comparison, my TPS2024 and TPS2024B scopes are fast enough to

stumble upon this. No problem there.

--
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
Land line: +7 (812) 461 01 01

--
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

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System