08-25-2017 04:58 AM
Hello,
I have a PXIe-based image acquisition system:
Initially, we only had 1x NI 8234 (in slot 2). Slot 3 was empty, and everything was working fine. Later on, we added another NI 8234 into slot 3. This was when trouble began.
When we first booted the PXI controller after inserting the new card, Windows automatically tried to install drivers, but failed:
Device Manager had trouble with the new card:
Notes:
I was able to "fix" Device Manager's complaints by right-clicking the device icons and manually updating the drivers:
However, this does not seem enough. When I try to snap/grab images from a camera connected to the new card, I get this:
(image taken from http://digital.ni.com/public.nsf/allkb/4D0ECFAA9AD9FF3E8625792600551A1E instead of my own screenshot)
Jumbo frames is enabled by default, and my firewall is completely disabled.
Here's what I've tried:
I'm out of ideas at this point. What could the issue be?
Thanks in advance!
08-27-2017 11:46 PM
08-28-2017 10:04 AM
Hi JKSH,
I'm glad you were able to narrow down the issue to the chassis itself. Since the HW seems to be defective, I would recommend creating an SR to get the chassis RMA'd.
Thanks,
Peter George
Applications Engineer
National Instruments
08-29-2017 03:30 PM - edited 08-29-2017 03:30 PM
Hi,
I am suspicious there may be some other failure mode other than broken hardware, such as a bug in the BIOS of the controller. Would you be able to try a quick test of setting the camera to a 1x50 AOI and see if it can acquire an image? Then try 1x100, 1x200, and 1x300 (all in Mono8). This would help narrow down if it an issue of packet sizes on the bus being DMA'd.
Eric
08-29-2017 09:10 PM - edited 08-29-2017 09:15 PM
@BlueCheese wrote:
I am suspicious there may be some other failure mode other than broken hardware, such as a bug in the BIOS of the controller. Would you be able to try a quick test of setting the camera to a 1x50 AOI and see if it can acquire an image? Then try 1x100, 1x200, and 1x300 (all in Mono8). This would help narrow down if it an issue of packet sizes on the bus being DMA'd.
Eric
Thanks Eric, I didn't think of that possibility.
I can't test it at the moment because our customer is using the system with 2 (out of 4) cameras, while awaiting the replacement chassis to arrive -- we've already initiated the replacement process with NI Australia. When the new chassis arrives, I'll try your test before swapping the chassis.
If it does turn out to be an issue on the controller rather than the chassis, what are the things I could try to fix it?
FWIW, I did try snapping a 100x100 Bayer RGB (8-bit) image at one stage, but still got the "The system cannot acquire images" error.
EDIT: One last thing I tried before concluding that it was a chassis issue: I swapped the hard drive for a brand new one with a pristine copy of Windows. I installed the IMAQdx drivers with all cards in their respective slots, then rebooted and ran NI MAX. The outcome was the same -- I can acquire images on all slots except the middle one.
08-31-2017 09:15 AM
Hi,
Definitely let me know how your test goes with the new chassis. It is hard to imagine a particular hardware failure condition that would lead to the set of observations you saw, so that is why I am a bit unsure if a new chassis will help. However, a colleague tried reproducing a similar setup and was unable to see the same issue (same chassis, controller, and 8234's in slots 2 and 3 [but no PXIe-6612]).
Eric
09-01-2017 12:55 AM
@BlueCheese wrote:
Definitely let me know how your test goes with the new chassis. It is hard to imagine a particular hardware failure condition that would lead to the set of observations you saw, so that is why I am a bit unsure if a new chassis will help.
The replacement chassis arrived today, and I expect to have access to the customer's system sometime next week. I'll keep you updated.
One thing my colleagues suggested, which I now wish I had checked earlier, is that perhaps a pin inside the chassis got bent. It is hard to imagine that those can bend without impacting my ability to insert/remove a card though. Does this sound plausible to you?
@BlueCheese wrote:
However, a colleague tried reproducing a similar setup and was unable to see the same issue (same chassis, controller, and 8234's in slots 2 and 3 [but no PXIe-6612]).
Thanks for checking!
This is the 4th almost-identical system that we've assembled for this particular customer. The differences are:
In all cases, we used a PXIe-1071 chassis, and had the PXIe-8234 GigE cards in the first 2 slots. Systems 1-3 didn't have any problems with acquiring data from a particular slot, but System 4 did.
09-01-2017 09:17 AM
Hi,
Because those are PXIe cards, the interface for communication with the cards is over a serial link (PCI express-based). If a pin were bent and caused communication issues it would almost certainly prevent the card from even being enumerated by the system at all, since the same exact data path is used.
Thanks for the info about other configurations you tested that didn't have this issue. It might help for narrowing down what's wrong if your new chassis shows the same problem.
Eric
09-06-2017 08:49 PM
@BlueCheese wrote:
Definitely let me know how your test goes with the new chassis.
Your suspicion was correct -- we are getting exactly the same behaviour on the new chassis! The same camera can acquire images from Slots 2 and 4, but not Slot 3.
I also tried acquiring a tiny image (Mono8, 2x2) from that slot, but got "The system cannot acquire images". Reducing packet size from 8000 to 1500 didn't help either.
Any ideas what I should try next? The only thing left I can think of is to try a new controller altogether. Our current controller is:
This is the same model as System 3 which worked fine, but we didn't record its firmware version or hardware revision.
Note: I found out that my earlier test with a PXIe-6612 in Slot 3 was flawed. I had updated the Physical Channel of the DAQmx Linear Position Task, but forgot to update the digital inputs for the A/B/Z channels, so those were pointing at the wrong slot. That's why I was getting routing errors.
09-07-2017 11:25 AM
Hi,
I realized that the test packet feature the IMAQdx driver does will still use maximum-sized packets to validate your connection regardless of the actual size needed for the image. Under the Advanced Ethernet category of attributes on the camera you should be able to disable the Test Packet feature and re-try those smaller images.
Additionally, please confirm that you are using the NI GigE Vision driver on all instances of the 8234 in device manager and there are no instances using the Intel driver.
Lastly, could you try removing the 6612 from the system? This would more closely resemble the setup we replicated here.
Eric