LabVIEW

cancel
Showing results for 
Search instead for 
Did you mean: 

sbrio 9606 ethernet cable

Hello, I am using the sbRIO 9606 on one of my projects. It works fine if I use a 10ft Ethernet cable from the board to the computer, but if I use a 20m cable, it does not support on time data transfer and I start seeing the data loss. I guess that the FIFO buffer fills up because of the slow transfer rate and finally I start losing the data.
 
Can someone please tell me what could be the reason behind that? Is 20m Ethernet cable not supported by sbRIO 9606?
0 Kudos
Message 1 of 6
(3,913 Views)

You offered next to zero information.  Ethernet is usually good to up around 100m.  It'd be a bit strange to see issues at 20m.  I seriously doubt any company will spec their device to a length of ethernet cable.

 

What's your sampling rate?  How are you connecting the sbRIO? Is it directly to the host or is it going through a router?  How many elements are you reading at a time from the host?  What's the size of your buffer?

0 Kudos
Message 2 of 6
(3,894 Views)

Thanks for the reply,

 

I am sampling at 960000 samples/second. The ethrenet is connected directly to the host. Reading 920000 samples at a time. And the buffer size is 5000000.

 

Can you please clearify that, by saying that " I seriously doubt any company will spec their device to a length of ethernet cable." do you mean that sbRIO 9606 does not work with 100m ethernet cable?

0 Kudos
Message 3 of 6
(3,876 Views)

Your buffer is 5x your read.  Try reading a larger chunk at a time on your host side.

 

Why would saying a company wouldn't spec their thernet even begin to mean that?  Ethernet, by itself, isn't reliable beyond 100m.  If you're going this far, you'd need to add repeaters.  But, none of this matters for your situation.  You're working with 20m, which is well below the 100m problem inherent to ethernet.  I've never seen a company state "we're only going to work with a cable up to x m long."  When I say I doubt I'll see that, I mean it's very unlikely you'll ever have a company give you a length as there is a length inherent to ethernet communications and it'd take needless testing to try to find another value.

0 Kudos
Message 4 of 6
(3,860 Views)

Thanks I got you point. But, here I am working with two different setups. One is working and the other not. 

 

The one that is wokring, I have a 10ft ethernet cable connected through router and I don't see any prolbem even at the sampling rate of 896000 sample/sec and Host transfer of 920000 every cycle.

 

But when I connect with 20m cable, I start having issue. There is another difference between two setup.

 

In the first setup, the card is powered with a dc supply separately and a plane ethernet cable is connected through router.

 

In the second setup, I am connecting a power and communication box that is located roughly 15m away from the card. A plane ethernet cable goes from host to that box, where I tie the DC power and the ethernet together on a 20m long ethernet/power combo cable and that runs all the way to the card, where I separate the power and ethernet. Do you thing there may be connector or cable that would not behave as expected? There are also intemediary mating connectors involved in order to aceive that.

0 Kudos
Message 5 of 6
(3,847 Views)

The 802.3 Ethernet standard dictates compatibility with 100m cables (as listed in the sbRIO user manual), and NI validates all Ethernet devices during design with 100m Ethernet cables.

 

It sounds like there are several variables of difference (power junction box, power/ethernet combo cable, etc.) between your 10 meter setup and the 20 meter setup.  I'd like to ask a few questions to try to isolate the issue.

 

The most likely culprit in my mind is the power/Ethernet junction box.  Can you share more details on this box and the cabling?  I find a picture or diagram contains a 1000 words in these types of scenarios.  Ethernet cabling has a controlled impedance, and if you are splicing this wire with additional conductors for power, this would create a definite impedance discontinuity.  This could lead to signal degredation and performance degredation or data loss.

 

Can you test without the power splicing box by running a direct CAT5 (or faster) Ethernet cable directly from the router to the sbRIO? For the Ethernet/Power cable you are using, is the Ethernet capability rated for CAT5/5e/6 data rates? Can you share a datasheet?

 

I also want to confirm something that I found vague in your previous post.  Is the power+Ethernet cable a Power-Over-Ethernet cable? or is it just a cable that has additional conductors for carrying power?

 

Regards,

Spex
National Instruments

To the pessimist, the glass is half empty; to the optimist, the glass is half full; to the engineer, the glass is twice as big as it needs to be has a 2x safety factor...
0 Kudos
Message 6 of 6
(3,774 Views)