Project

General

Profile

High/Low gain mode issue using 16bpp

Added by Kyle McKinney almost 5 years ago

I am having issues using 16bit mode with any gain mode other than combine.
Data is transmitted but no matter the light level I get darkness. As I
change the light level the overall average pixel value does change so
something is happening, just nothing meaningful.


Replies (5)

RE: High/Low gain mode issue using 16bpp - Added by Michael Williamson almost 5 years ago

Hi Kyle,

Are you using XCAP (the epix frame grabber) or MityViewer to look at the data?

When you switch to single channel, all of the data scaling is removed and the data is packed into the lower 11 bits of the 16 bit word.

Quite often the data scaling presented by XCAP (and Mityviewer, depending on how you set it up) is trying to linearly map 16 bits of data to the 8 bits on the screen, and so all but 3 bits are chopped in the display and it looks black. The images are there (if you save them and open them in ImageJ or if you take the isolated image and scale it in MityViewer) but, the data is jammed into the LSBs of the pixel.

For a single image buffer in XCAP, you can shift the 11 bits of data in each pixel to the MSB:

- Ensure the capture is in "Unlive"
- Snap an image
- From the PIXCI View window --> Modify menu
- Arithmetic
- Bit Shift Left; Shift Bits set to 4 (or 5)
- Hit apply

Let me know if this helps at all.

-Mike

RE: High/Low gain mode issue using 16bpp - Added by Kyle McKinney almost 5 years ago

I use MATLAB to look at and process the data. Knowing that the data is packed into the lower 11 helps answers my question.
Using only 11bits should produce a maximum pixel value of 2047 but I am getting pixel values that peak at 2067. Is this
supposed to happen?

RE: High/Low gain mode issue using 16bpp - Added by Michael Williamson almost 5 years ago

There is a 20 count bias in the data. It's a bug in the mode you are using.

During calibrated high/low gain blending (normal operation), there is a 20 count bias added (digitally) to each of the channels prior to the blending.

The uncalibrated single channel modes you are using do not disable the 20 count bias. It should.

Subtract 20 counts to get the raw data. We will fix this issue in a future firmware release.

I am sorry for the inconvenience.

-Mike

RE: High/Low gain mode issue using 16bpp - Added by Kyle McKinney almost 5 years ago

Ah, ok well that explains it! No problem, just good to know what is happening.
Thanks.

Kyle

RE: High/Low gain mode using 8bpp-12bpp - Added by Kyle McKinney almost 5 years ago

Mike, is there a similar bias in the 8bit and 12bit modes?
If so what is it?
Does 12bit pack into 11bits in those modes?

    (1-5/5)
    Add picture from clipboard (Maximum size: 500 MB)