Gatan K3 installation and setup¶

Gatan K3 installation and setup is identical to K2 except the instrument.cfg and the additional dmsem.cfg settings below.

instruments.cfg¶

• A template for instruments.cfg is in the installed pyscope directory as "instruments.cfg.template". Copy it to
  

  C:\Program Files\myami\instruments.cfg

• Remove SimCam modules in the configuration.

Gatan K3 Instruments.cfg¶

The camera modes are configured as separate cameras as follows.

• K3 is natively Super Resolution. Use binning of 2x2 to get counted camera effect.

[Gatan K3]
class: dmsem.GatanK3
zplane: 50
width: 8184
height: 11520

dmsem.cfg¶

All available options for K2 are also available to K3. You can configure the section [k2] as described in Gatan K2 installation and setup.
Make sure you keep the section header [k2]. Don't change that.

A new section called [k3] includes K3-only settings.

DM_PROCESSING

For now, the default is "gain normalized". This means the returned frame movies are dark subtracted and gain normalized. This means the resulting data are float number. Therefore could not be compressed efficiently.

As of today, service mode access is required from Gatan to use K3 in "dark subtracted" mode which is equivalent of the raw-frame movies in K2 counted and super-res mode. This mode can pair with LZW tiff compression to save a lot of space. To do so set the following:

[k2]
....
SAVE_LZW_TIFF_FRAMES = True
[k3]
DM_PROCESSING = dark subtracted

Testing with pyscope¶

Binning=2x2 (Counted mode)

1. Start DigitalMicrograph
2. From python command line or IDLE:

   import pyscope.dmsem
   k = pyscope.dmsem.GatanK3()
   k.setBinning({'x':2,'y':2})
   k.setExposureTime(200)
   a=k.getImage()
   a.shape
   

• The last command should return the shape of the image arrau

You should expect these to run without error. The getImage() command should give a 2D numpy array like

array([[1000, 3400, 2300, ..., 1000,1200,3000],
         [1000, 3400, 2300, ..., 1000,1200,3000],
         [1000, 3400, 2300, ..., 1000,1200,3000],
         ...,
         [1000, 3400, 2300, ..., 1000,1200,3000],
         [1000, 3400, 2300, ..., 1000,1200,3000],
         [1000, 3400, 2300, ..., 1000,1200,3000],dtype=int16)

The number and dtype depends on the camera.

a.shape command should give a tuple of the camera dimension matching your camera.
For example, (4096,4096)

If you use python shell to do this test, some of the error will cause the shell window to close immediately. Use Python IDLE instead in that case

Testing frame saving¶

You can continue the test above by saving frames, too.

k.setSaveRawFrames(True)
k.setExposureTime(200)
a=k.getImage()
a.shape

* The last command should give you the shape of the summed image, and the frame movie should show up on your frames directory