Experimental Physics and
Industrial Control System

Hinxx <[email protected]> · Wed, 13 Jul 2016 13:15:18 +0200

Hi Mark,


On 07/12/2016 11:36 PM, Mark Rivers wrote:

Hi Hinko,


OK. I guess we can expect this change to appear in 2.5, 2.6 release?

Yes, it will be in the next release.


Would you accept a patch that would allow changing read mode?

Yes, I'd be happy to accept a patch to support changing the read mode.  I would like the patch to include:

- Updated RELEASE.md file
- Updated andorDoc.html
- Updated medm screens with widgets for the new PVs.  Other OPI screens are autoconverted from the medm screens.  If you cannot run medm then please send a screen shot of your OPI screen so I can reproduce it in medm.



I have forked the ADAndor on Github and submitted a pull request for

you: https://github.com/areaDetector/ADAndor/pull/7


Pull contains the following changes.


a) Deals with issues I had when trying to run latest code with Andor

Luca R I have here. Basically Luca does not support shutter so I had to

add a check to make sure that initialization in constructor actually

happens all the way.



b) Along the way I stumbled on a situation where initialization would

fail with:



2016/07/13 10:34:26.841 andorCCD:AndorCCD: ERROR: Unknown error

code=20992 returned from Andor SDK.


From atmcdlXd.h:

#define DRV_NOT_AVAILABLE 20992


This then resulted in segfault because enum strings were not allocated,

but setupAcquisition() was called due to PINI processed records, AFAIK.

Power cycling the camera resolved the issue. Nevertheless I moved

allocation of the enums before Andor SDK tries initialization. With this

change IOC will run but report bunch of SDK errors without crashing.



c) Also to avoid division by zero in setupAcquisition(), due to same

reasons mentioned above, I added a flag the allows bailing out early and

not crashing..




For the actual readout mode support I would at least need shutter check

from above pull request ( a) ), if I want to implement and test this new

feature.



I have a patch on top of the above changes that currently allows

selecting Image or Full Vertical Binning readout modes. I guess I need

to wait until first pull request is merged and the issue a new one. I

have modified CSS OPI, RELEASE.md and andorDoc.html as requested, too.



For the Single-, Multi- and Random- track readout modes I would need a

bit more time. It might be better to add support for those in separate

pull requests anyway.


Thanks,
Hinko



Thanks,
Mark


-----Original Message-----
From: Hinxx [mailto:[email protected]]
Sent: Tuesday, July 12, 2016 8:40 AM
To: Mark Rivers; [email protected]
Subject: Re: AndorCCD binning

Hi Mark,

OK. I guess we can expect this change to appear in 2.5, 2.6 release?
We will probably use patched 2.4 until then.

On a related note - Andor SDK has something called SetReadMode() to
setup for different readout modes: Image, Full Vertical Binning, single
track and multi track. Currently andorCCD only works with Image mode.
Would you accept a patch that would allow changing read mode?

Thanks,
Hinko

On 07/12/2016 02:56 PM, Mark Rivers wrote:

Hi Hinko,

Your change looks correct to me, I don't know where the factor of 2 was coming from.  I added that code back in January 2012, but I probably only tested with small values of binning (e.g. 1, 2, 3, 4).  I will make the change and test on our Andor iDus 401 camera.

Mark

________________________________________
From: [email protected] [[email protected]] on behalf of Hinxx [[email protected]]
Sent: Tuesday, July 12, 2016 2:52 AM
To: [email protected]
Subject: AndorCCD binning

Hi Mark,

Binning handling in AD andorCCD support does not allow to produce full
vertical binning (width x 1 image). Looking at the code I noticed that
some magic with trying to figure out the minY might be the problem. Also
when trying to come up with the binY values Andor SDK rejects some valid
binning values because other parameters to SetImage() are invalid.

unsigned int WINAPI SetImage(int hbin, int vbin, int hstart, int hend,
int vstart, int vend)

For example Andor Luca with 1004x1002 detector and binY set to 1002
yields this call to the SetImage():

SetImage(1,1002,1,1004,-1001,0)

Of course it results in error from Andor SDK (-1001 is not valid minY).

Probably same applies to horizontal binning..


After changing some of the lines of the andorCCD.cpp I can successfully
set binY to 1002 and get 1004 x 1 image:

diff --git a/andorApp/src/andorCCD.cpp b/andorApp/src/andorCCD.cpp
index ff09114..2ff92e0 100755
--- a/andorApp/src/andorCCD.cpp
+++ b/andorApp/src/andorCCD.cpp
@@ -1060,12 +1060,12 @@
      getIntegerParam(ADSizeY, &sizeY);
      getIntegerParam(ADMaxSizeX, &maxSizeX);
      getIntegerParam(ADMaxSizeY, &maxSizeY);
-  if (minX > (maxSizeX - 2*binX)) {
-    minX = maxSizeX - 2*binX;
+  if (minX > (maxSizeX - binX)) {
+    minX = maxSizeX - binX;
        setIntegerParam(ADMinX, minX);
      }
-  if (minY > (maxSizeY - 2*binY)) {
-    minY = maxSizeY - 2*binY;
+  if (minY > (maxSizeY - binY)) {
+    minY = maxSizeY - binY;
        setIntegerParam(ADMinY, minY);
      }
      if ((minX + sizeX) > maxSizeX) {

The call to SetImage() is now:

SetImage(1,1002,1,1004,1,1002)


This change might break other Andor detectors, here at ESS and with CEA
partner we can test Andor Newton and Andor Luca.


Thanks,
Hinko





Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System