Re: Andor SDK3 issue - AT_WaitBuffer fails

[Brett Graham <Brett_Graham@hms.harvard.edu>]

Hi Matt,

No worries on the timing of your reply, the information is extremely helpful.

I wasn't even aware of (nor was I informed of from Andor) the /tmp/atdebug.log file. I will look through this for messages that match up with times we see AT_ERR_NODATA errors. It looks like it gets wiped every r

The errors are very sporadic for us. We can go several million grabs before we see an error but can also sometimes get errors within less than a few thousand images.

I've tried retriggering (we're using software triggering) with no luck. However, stopping and starting acquisition does appear to allow us to work around the issue.

I will let you know if I find similar atdebug messages when we encounter the error again.

Thanks again for your help.

--

Brett

On Wed, Aug 2, 2017 at 10:55 AM, Pearson, Matthew R. <pearsonmr@ornl.gov> wrote:

Hi Brett,

Sorry for the late reply.

We still have the same issue with our camera. This is how the EPICS driver starts an acquisition and waits for data (not all the code is pasted here for brevity):

AT_Flush(handle_);
AT_QueueBuffer(handle_, drvBuffers_[x],
                                           (int)imageSize_);
AT_Command(handle_, L"AcquisitionStart”);
AT_WaitBuffer(handle_, &image, &size, AT_INFINITE);

which should block until there is data ready. But AT_WaitBuffer returns immediately with the error code 11 (which means AT_ERR_NODATA).

We’re using SDK 3.9, although I saw it previously with SDK 3.7 and I’ve seen it with SDK 3.12. I didn’t continue using SDK 3.12 because I had other issues with this on RHEL6.

When I exit the EPICS IOC I often saw this exception from the SDK:

TSDK3Exception'
[Thu Sep 22 19:46:07 2016]   what():  TDualCLLogicalControl: Error - Sensor halves are not running in unison

which seemed to be a result of us calling:

AT_GetBool(handle_, L"CameraAcquiring", &acquiring);

on an exit().

In the log file generated by the BitFlow kernel module I see:

../../regcam/rcacquisitionstart_regcamcl.cpp. 184 (PID:5044 TID:2853934848): 21/10/16 18:48:58: startCameralink:        Error with mp_clControl->Start()
../../regcam/cldevice/dualcltentaplogicaldatasource.cpp.  99 (PID:5044 TID:2853934848): 21/10/16 18:48:58:      Wait:   Error waiting on Even, not waiting on Odd, timeout [ -1 ]

Andor seemed to think it was a hardware issue.

Do you see similar errors in the /tmp/atdebug log file?

We were having this issue on the beamline every few hours, and the only way to recover was to restart the application (didn’t need to power cycle anything). I moved the camera from the beamline to my office and ran it there for a while, and it ran for weeks without a single issue. Moving back to the beamline, and using a different PC, and the problem happened again quite regularly. So I came to the conclusion it was environmental noise.

Cheers,
Matt









Data Acquisition and Control Engineer
Spallation Neutron Source
Oak Ridge National Lab

> On Jul 23, 2017, at 7:30 AM, Brett Graham <Brett_Graham@hms.harvard.edu> wrote:
>
> Hi Matthew and Mark,
>
> I recently started getting AT_ERR_NODATA errors when calling AT_WaitBuffer on a Zyla 4.2 (we have 4 and have so far encountered the error on 3 of them, some multiple times).
>
> I found your thread "Re: Andor SDK3 issue - AT_WaitBuffer fails" from October 2016 and was hoping you could provide some advice or update on the error you were getting.
>
> I'm not using Epics but am using the SDK (3.11.30014) on linux (ubuntu 14.04). I've been in contact with Andor and they suggested changing the timeout argument (I was using the exposure time, 25 ms). I doubled the timeout but I just received the error again last night.
>
> Thanks in advance for any information you can provide and sorry if this isn't appropriate for the mailing list.
>
> --
> Thanks,
> Brett