Imaqdx and multiple GigE cameras - Stability issues

[adrrienc]

Hello,

I am using 4 GigE cameras and the 4 ports NI frame grabber. I am recording continuous video over extended time. My application works fine over a long time without error occuring, when using short patch cables (cat5e or cat6a - 2 to 10m). However, when using full length cables (60m), as it needs to be in the real application, I bump into the following problem: one camera (not necessarily the same one) ends up losing connection (sometimes after a quite long time recording- 30min to several hours), probably for a very short time, before being back again on its waiting state (I get the error 0xBFF6900C - Camera has been removed). It must not happen for my application. Cameras are powered through PoE. Ethernet cables are cat6a, shielded foiled twisted pairs.

Any suggestions on what could be the issue here?

Thank you.

[BlueCheese]

If you are sure that the issue tracks cable length, I can think of two possibilities:

• Packet loss - Check the image reception packet loss statistics and make sure you aren't losing packets. If you are losing some there, you may be losing camera discovery packets that will cause the driver to remove it
• Insufficient power - Power over ethernet will have a voltage drop proportional to the length of the cable. This will translate into increased current draws from the camera as cable gets longer and the voltage they receive drops. Some cameras may have built-in diagnostics to read the voltage level, or maybe you can read the timestamp counter to see if rebooted itself.

[adrrienc]

Hello,

Many thanks for your answer. I tried to check the camera diagnostics statistics, time since initialization and time since last bus reset, and the cause seems to be a bus reset. Then I am not sure yet what is the cause of the bus reset itself. I will record the data loss statistics overnight and check if it comes from here. I dont have the possibility of monitoring the power supply with this model unfortunately.

Thank you.

[BlueCheese]

Also be careful with using shielded cables. It is possible in some cases to create a ground loop between the two ends of the cable. Typically you want only one end of the shield to be connected.

[adrrienc]

Ok, that is good to know. I need to check this with the supplier.

Another question: I have a "GigE resend packet" option for each camera. This was disabled until now. Is there any chance that enabling it would help to recover potential lost camera discovery packets, and then mitigate the bus reset issue?

Many thanks for your help.

[BlueCheese]

The resend packets option is purely related to image transfers. The camera discovery mechanism is independent and has its own mechanism for dealing with lost packets. The only way these two items can interact with each other is if there are tons of resent packets that are starving the link for bandwidth and interferring with control traffic from the camera. This is camera-implementation-specific, though, as many cameras reserve/prioritize bandwidth so this can't occur.

[adrrienc]

In order to give an update on this topic:

-After testing it appears these bus resets always occur on the same port of my frame grabber (NI 4 ports, I tried to swap cameras and cables).

-Plugging less cameras does not help.

-I will try to supply power via GPIO instead of PoE.