Running a Basler raL8192-80km at full speed

[Brett_Percy]

Hi,

I'm trying to set up a Basler raL8192-80km linescan camera to capture images at 80,000 lines per second, which is fairly close to its maximum speed.

I've found that using MAX 14.5 that I can only get half of the maximum scan rate - 40,000 lines per second.

The interface to the camera is a PCI-1433, using Full Camera Link

Looking in the camera file provided for this camera, it is only configured to use 4 taps, which would explain why the scan rate is limited.  I've tried hand-editing the camera file to select the full 8 taps that the camera can use, but this doesn't seem to have any effect.  I'm not sure why the supplied camera file doesn't support the maximum speed of the camera out of the box?

Is there a more correct camera file available for this camera?

Thanks

Brett

[samazing]

Hello,

I apologize for the late response, I have been working to reproduce this issue on my end as it seems that there are a couple different steps that need to be taken to achieve the correct lines/second. I have a few follow up questions and next steps for you to try:

Firstly, you should be configuring your camera from Basler's configuration SW, Pylon: http://www.baslerweb.com/en/products/software/pylon-windows

If you are already doing this, the main configuration settings we care about here are tap configuration and exposure time

You can use Basler's Frame Rate Calculator to determine the appropriate exposure time and tap configuration. It looks like 8 taps and an exposure time of 2 us should give you 80645 lines per second.

Something that I learned in my research is that the camera files provided by NI have built-in serial configuration commands that are sent to the camera upon initialization of the camera session. What this means is that even if you configure your camera in Pylon, the settings may be overwritten by the automatic serial commands. To prevent this, we must modify the camera file. Here are the steps you can follow to do so:

1) First, save a copy of the .icd file so that you may have a back up copy
2) Download and install the NI Camera File Generator from the following link: ftp://ftp.ni.com/support/imaq/pc/ni-imaq/cfgenerator/
3) Open the Camera File Generator, select the NI 1433 as your target board (simulated device is OK), and select Open Existing Camera File to open the file you want to modify. (image1)
3a) You may get an error when opening the Camera File - if you do, go ahead and edit your camera file to be 4 taps temporarily. Remember to change the camera file back to 8 taps by changing lines 55 and 57:

55: NumCameraTaps (4) -> 8
56: NumCameraXZones (1)
57: NumCameraTapsPerXZone (4) -> 8
4) Once you have opened the camera file, go to Settings » Serial Initialization (image 2)
5) From here, delete all of these commands. Again, these configuration commands are sent to the camera upon opening the camera session and are likely the cause of the tap's reverting.

Additionally, there are default settings sent to the camera that you can remove by going to the Camera Control tab and deleting them as well (image 3).

Ideally, after doing this the configuration in Pylon should match the configuration in the camera file which should match the actual camera's configuration.

However, if this does not work, can you please post the following:

1) The camera file in question

2) A screenshot of Pylon, specifically the acquisition configuration (tap configuration, etc)

3) Can you also set the camera to a test image and send a picture of the image acquired from a snap in NI MAX?

Best,

Sam K

[Brett_Percy]

Hi Sam,

Thanks for looking into this.  I've managed to get the camera working with 8 taps using your advice.

I've attached the modifed camera file I ended up with - I've stripped all of the serial commands from it using the Camera File Generator, which might have been excessive but I couldn't get it to work otherwise.

I had some trouble getting the Pylon software to communicate with the camera reliably but setting the maximum baud rate down to 115200 using the Pylon CL Configurator seems to have fixed that.

The final process was 

- set up the desired capture windows size and exposure time using MAX and the original 4 tap camera file

- edit the number of taps in Pylon

- load the modified 8 tap camera file into MAX and save the configuration

Cheers

Brett

[Brett_Percy]

Ok so it turns out this problem is not quite solved after all - I've started a separate topic but it probably belongs back here

http://forums.ni.com/t5/Machine-Vision/Basler-raL8192-80km-has-black-zero-pixels/m-p/3568293/highlig...

The modifications to the camera file have caused a pattern of missing/black pixels in the image:

The above is a test image from the camera, so none of the optics are involved.  Given that the default camera file gives an image that looks fine, this probably has something to do with the 8 tap setting

[samazing]

Hi Brett,

If you take a picture with the camera, does the image appear in a similar manner to the test image? By that I mean, does it show up clear but just in the 'wrong order', or is it essentially noisy static with no discernible qualities? What it looks like is happening here with your test image is that all of the data is coming in correctly, it is simply being pieced together incorrectly. For example, if the camera is sending 8 taps to the framegrabber (as configured by pylon), but the framegrabber was expecting 4 taps (default configuration of .icd file) I would expect to see complete empty spaces throughout the image as we are losing 4 taps worth of data.

Perhaps the number of taps are configured correctly (as we are seeing complete data) but perhaps the geometry is not configured appropriately. Are you able to confirm the tap geometry's match in both the camera configuration and the camera file?

edit: after reading through your posts again I'm realizing now that those black lines are not expected - which would then point us towards a missmatch in the number of taps (as I described above). Is it safe to assume this only occurs with the 8 tap configuration? I will have to look into this further and get back to you.

Best,

Sam K

[samazing]

Can you confirm what bit depth the camera is configured to from Pylon? The camera file you posted is configured for 8 bits and this behavior may coming from having your camera setup with a bit depth of 10 bits.

If there is a missmatch in bit depth there would be a shift in the pixel values that may cause these black lines. (They may not be missing pixels but rather just very low greyscale values).

To test this, can you try changing your camera file to expect 10 bits and let us know the results?

If you don't find success here, can you save the full image acquired from MAX so that we may inspect the exact pixel values. To clarify: save the picture from MAX rather than taking a screenshot.

From the camera's manual you can see the difference between 8-bit and 10-bit configurations:

http://s.baslerweb.com/media/documents/AW00118702000%20racer%20Camera%20Link%20Info%20Frm%20Grab%20D...

Best,

Sam K

[BruceAmmons]

To me it looks like the last two out of eight pixels are not mapped correctly.  They are always zero.  The camera file only shows Class07.dat and Class08.dat, so I assume these mappings are in separate files.  Can you find these files and post them?  They are probably compressed, so it wouldn't do much good, but who knows?  Are you able to specify each line for each bit in the camera file editor?  Have you verified these for the last two pixels?

Bruce

[Brett_Percy]

Hi Sam,

This problem only occurs with the modified 8 tap camera file - I tried using the 4 tap version that NI provides and that works fine

Also the live images I've taken exhibit the same issue, it's just less clear as I don't have the correct lens for the project yet and the images are quite faint

Cheers,
Brett

[Brett_Percy]

Hi Sam,

I don't have the camera at the moment, I've sent it back to NI's office in Sydney to have them see if they can work it out

I did try changing the bit depth to 10 in Pylon but it just caused the camera comms to timeout in MAX, so I couldn't see anything with that changed

Hopefully I shuold get the hardware back later this week so I will take a closer look at it then

[Brett_Percy]

Hi Bruce,

I've attached the two .dat files referenced in the Camera file.  I've had to add an extra .h extension to the filenames as apparently .dat is verboten

The Camera File Editor won't let me specify 8 taps using the NI 1433 - it only seems to support 4 taps

Cheers,
Brett