Cat5e: Ok To Coil Excess?

[PeteCress 0]

3 identical cams, identical setups as far as I can see, connected to the same POE switch, and being served up by Blue Iris.

 

One is working a-ok. Other two are showing very slow frame rates - sometimes zero with frozen video streams - albeit higher bit rates.

 

Direct connections to the problem cams seems to support the idea that it's the cams and not something weird with BI.

 

Only diff I can come up with is that the "Good" cam's Cat5e cable is fairly short - maybe eight feet max while the others' are more like 50' and a substantial part of that is excess and is coiled into 2 loops about 8" across and secured with wire ties.

 

Before I drive 200 miles round trip.... could this be the problem?

 

viz: http://tinyurl.com/kf4pbd2

 

Where "East" and "South" are the problem cams and "West" is the "Good" cam.

 

FWIW, the BI presentation of the cams is at http://ExtremeSurfCam.DynDNS.org

[Don Stephens 0]

My only guess is that there isn't enough bandwidth to support it but I'm assuming you took all that into account. I can't think of any reason that it'd be as a result of a small amount of coiled up cable.

 

What kind of equipment is it?

[id5 0]

3 identical cams, identical setups as far as I can see, connected to the same POE switch, and being served up by Blue Iris.

 

One is working a-ok. Other two are showing very slow frame rates - sometimes zero with frozen video streams - albeit higher bit rates.

...

 

Unlike coaxial cable Cat 5 has each strand separately insulated, there is no core strand that could cut through the insulator to the outer strands, you can tie cat5 in knots and it does not make a difference. You can put two devices a couple of feet apart and use a 100m drum of cat5 set between them and you would see no difference in bit rate as each of the four pairs are twisted to make sure that stray signals do not get electromagnetically created.

 

If it is working then the Ethernet chips that drive the ports will only downgrade the connection to 10Mbps if you have badly terminated connectors or have a poor socket connection before they stop working completely and that is enough for 2 1080P cameras at 30fps.

[PeteCress 0]

What kind of equipment is it?

HikVision DS-2CD2032-I cams, TrendNet 8-port POE switch.

 

A pair of Ubiquiti Loco radios link the site to the server about .6 miles away.

 

...you can tie cat5 in knots and it does not make a difference. You can put two devices a couple of feet apart and use a 100m drum of cat5 set between them and you would see no difference in bit rate as each of the four pairs are twisted to make sure that stray signals do not get electromagnetically created.

So much for the cable theory..... -)

 

Thanks for saving me the trip.

 

 

Just ran "LAN Speed Test" from a PC on the site against the server .6 miles away, and maybe I have something: back on 8/11 the same test showed 55/55 MBps Write/Read, but just now it's only showing 7/38 Write/Read.

 

Clearly a significant diff, and now that I'm spending some more time staring at the frame rates, I see that the problem cam moves around. Right now "Parking Lot" is down to 5 FPS and the other three are ticking along at 20.

 

OTOH, Ubiquiti's "Speed Test" built in to the radios shows 57/57 Mbps, and re-running LAN SpeedTest immediately afterward shows the same 7/38 speeds.

 

OTOOH, when I add up BI's idea of each cam's bitrate, I get 1,411 KB/s which I think comes out to 12 Mbps....

 

The difference in the two tests aside, I guess I will spend some more time monitoring the radio link's speed and see if I can find a correlation between 55/55/Good Performance and sub-10/sub-55/Problems.

[ssmith10pn 0]

Make sure the LOS is ok on the radios, Check signal strength and make sure they are still pointed correctly.

Last but not least make sure the radios have the latest firmware.

[PeteCress 0]

Make sure the LOS is ok on the radios, Check signal strength and make sure they are still pointed correctly.

Last but not least make sure the radios have the latest firmware.

Any thoughts on why the Ubiquiti speed test gives pretty good (50-55 MB/s) speeds and LAN SpeedTest gives such a low write speed?

[ssmith10pn 0]

Make sure the LOS is ok on the radios, Check signal strength and make sure they are still pointed correctly.

Last but not least make sure the radios have the latest firmware.

Any thoughts on why the Ubiquiti speed test gives pretty good (50-55 MB/s) speeds and LAN SpeedTest gives such a low write speed?

 

The Ubiquiti speed test shuld be 135 to 160 Mbps.

Don't get Mbps and MBps mixed up.

Also was the camera load on the network when you did the test?

[ssmith10pn 0]

Also can you diagram the physical network layout?

 

It doesn't have to be a drawing but as much detail as possible.

[PeteCress 0]

Also can you diagram the physical network layout?

Maybe a bit busy, but I'm trying to keep it to one page:

http://tinyurl.com/mkfmvcx

 

Also was the camera load on the network when you did the test?

Yes, I currently have no way to turn the cams on/off remotely.... the POE switch is just plugged in and that's it.

 

 

Edit 2014 11-09 11:34:

 

Several contributors to the Ubiquity forum say that the Ubiquity speed test is not reflective of real-world speed - partially because it uses UDP instead of TCP and therefore does not reflect packet loss. They recommended a separate utility and I'll download it and how it's results match up with LAN SpeedTest's.

 

I am starting to suspect a particular run of cable and will try to prove that by swapping connections around so that the PC running the test gets a direct path to the server with no middlemen and no cams online.

Edited November 10, 2014 by Guest

[ssmith10pn 0]

Looks like your using the cablemodem/router for a switch? That could be the bottle neck.

[id5 0]

Assuming your network switch at the shack is 100mbps then your problem is bandwidth on the wireless link. Wireless network equipment is really designed for one direction data streams, the buffers soon fill up and the problem only gets worse as more devices are added, made worse again when another local wireless station uses the same spectrum.

[ssmith10pn 0]

I would also venture to guess that the extreme conditions (Salt) could be working on your RJ45 connections on the Nano stations.

[PeteCress 0]

I would also venture to guess that the extreme conditions (Salt) could be working on your RJ45 connections on the Nano stations.

Now that you have said it, that seems like a very likely cause. Historically, even inside the shack, we have had problems with router ports corroding. Somewhere along the line this year we had an episode of vandalism and somebody else re-wired the cams and the Nano.... and it looks to me like they did not replace the mastic tape I had on it around the Cat5 connection.... and I'm not too sure about the outdoor cam connections either.

 

The presence of 4 cams while I'm running the test is probably confusing matters (6 MB/s uploading, 35 down...) but I have to wonder if there is a chance that corrosion on the right Nano Cat5 contact could affect transmission in one direction and not the other.

 

As far as the Comcast cable modem/switch goes, it's the same setup that was in place back in August when I was getting 55/55 so, good or bad, I don't think it's responsible for the change.

 

As far as two-way traffic goes, even if there were a way to make the radio link one-way, I could not do it because it also supports the shop owner's Internet access.

 

I'm going to drive down there today or tomorrow and start re-plugging/swapping wires, adding mastic to any outdoor connections missing same, and probably spray some of RadioShack's magic goop on the contacts.

[jeromephone 6]

kind of what the other posters said about the connections if part of a connector is bad you might still get 10mps on good pairs. I would look into weatherproof /industrial Jacks as they are sealed up and will extend the life of connections is dirty or weather related conditions you can find them in many places and they are pretty cheap if you don't have to replace them every few months or so. If you do have to rewire look at outdoor rated cat5 as the jell will help with salt problems

[Digiscan 0]

I think technically even coiling can impact signal (at a very minute level), but at the distance you mentioned and the coiling described you'd never be able to measure it.

 

I highly doubt it's as simple as corroded connectors. I think better money is a connector is either so bad it's not working or good enough that it is; we're talking a tiny, tiny, tiny piece of metal compared to the entire length of the cable. I think you'd need to be terribly unlucky that it's corroded just enough to let things "kind of work".

[hardwired 0]

The Nanostations can be sensitive to static damage if you do not use shielded cat5 cable with metal plugs between the power injectors and radios.

 

One thing to do is to run the spectrum analyzer, and pick a channel that's low in usage (the 5.5.10 firmware opens up usage in the UNII-1 band, so that will give you more channel choices). Also, adjust your power levels for as close to a -50 signal level as possible.

 

Also, you need to test the connection with the other user computers disconnected, a virus or streaming/torrent application could be using a lot of bandwidth. (What throughput usage does the Nanostation display in the main page?).

 

A good link (locked steady at MCS15), on a 20MHz channel width, should give you at least 35Mbps of actual throughput. 30 MHz channels will give 65+Mbps. (I try to avoid 40MHz channels, real world usage tends to show lower throughput than 30MHz setting, as 40MHz is two bonded 20MHz channels, with more interference possibilities).

[PeteCress 0]

I'm going to drive down there today or tomorrow and start re-plugging/swapping wires, adding mastic to any outdoor connections missing same, and probably spray some of RadioShack's magic goop on the contacts.

Bit the bullet, drove down today, and found multiple issues:

 

• At the server site, I found a bad RJ45 connection on the back of the server PC. Swapped out the Cat5, and we went up to 34 MB/s on the test. Put the old Cat5 back, wiggling the connector.... and it stayed at 34. Probably should have done some cleaning, but the stuff was back at the cam site.
  .
  
• Back at the cam site I took down the POE switch, slathered all the RJ45 ports with cleaner then with some other goop... let it all dry, re-plugged, and now we have 55/55 Up/Down on the test.
  .
  
• Also back at the cam site, I noticed that the guy who re-mounted the Nano had left the mast twisted so that it aimed about 45 degrees out on the bay instead of along the shoreline at the server site. Straightened it out and noted a couple of dB improvement in "Aim Antenna", but no diff in LAN SpeedTest.
  .
  
• One cam still was not working after all this. It came up on a NetScan, but refused connection. Took it down, took it home with me, and now it's working A-OK. I've been through this with a couple of EdiMax cams.... they give trouble, then work after being brought home. Ultimately, both EdiMax's failed altogether and it looked like a boiled-over capacitor in each case. Maybe this HikVision cam is on the way out too... time will tell.

 

Seems like all my overthinking and theorizing was for naught and that sSMith's observation is the winner.