Ethernet negotiation speed 3000L vs 1000?????

Products ePMP
1

Hi, in two of our towers we tried to replace three 1000 APs with three 3000L APs. From the first moment we started having ethernet negotiation problems, all 3000L could come up only at 10Mbps Full Duplex or hardly at 100Mbps Half Duplex. We tried to replace cables, connect all grounds, power with its own power supply, tried different routers and switches and even replace all new 3000L with other new 3000L but there was no way to make them come up at more than 10Mbps FD stable.

Magicaly, replacing all back with 1000 ethernet links are up at 1Gbps Full Duplex and running since days. 

Yes there is interference on these towers (FM and TV) but nothing more than usual so, what is different with 3000L?

Obviously triyng the 3000L back in the office all links are 1Gbps.

@Cambium: Is 3000L "weaker" in matterials that was used to build it than 1000? Why the difference?

2

Hi,

ePMP3000L Ethernet PHY is not much different from ePMP1000 so it is not an expected result. I am also not aware of other customers reporting similar issues. From what it sounds you tried all the troubleshooting tips so it is really hard to recommend something without deep investigation.

Thanks,

Dmitry

3

Would this have anything to do with e1k AP's that have the "Smart Speed" feature enabled... vs. the e3kL not having this feature?

4

Smart Speed is enabled by default on both ePMP1000 and e3000L. 

 

someone , do you have it disabled on ePMP1000 in configuration by any chance?
5
@someone wrote:

Hi, in two of our towers we tried to replace three 1000 APs with three 3000L APs. From the first moment we started having ethernet negotiation problems, all 3000L could come up only at 10Mbps Full Duplex or hardly at 100Mbps Half Duplex.

Is there any FM radio transmitter on those tower?
6

@Dmitry Moiseev No, smart speed is enabled in 1000

@MW_WISP Yes, there are FM transitters on the towers

7

FWIW, we have not encountered any non-GigaBit negotiation speeds with Force300 or 3000 or 3000L.  That's probably not helpful, but we've replaced 200 with 300, and 1000/2000 with 3000, with no ethernet issues.

8

@MW_WISP Yes, there are FM transitters on the towers

I think this could be the problem. Be sure you're using a *good* STP cable, with *good* RJ connectors crimped to ground.

9
@Eric Ozrelic wrote:

Would this have anything to do with e1k AP's that have the "Smart Speed" feature enabled... vs. the e3kL not having this feature?

I thought the 3000L had it ,  you just don't have the setting to turn it off ?

10
@someone wrote:

Yes there is interference on these towers (FM and TV) but nothing more than usual so, what is different

Long story short, if you are using good double shielded ethernet cable (braid and foil) with grounded connectors (as already mentioned), and still have this issue on towers that also have FM broadcast antennas, you will need ferrite cores installed at the radio ethernet ports.

Two weeks ago, we mounted an FM broadcast on a tower we use for the owner of said tower, and when it rains, we are seeing this with the 300 CSM and PTP550 (both do not have the "Smart Speed" option available, but not sure if Smart Speed would help). The 1000 AP's on this tower are not experiencing any issues. 

This is a 240' AM tower that has been convertered to FM broadcast use (FM translators).

235'; Our 1000's

220' First FM antenna

205'; Second FM antenna

180'; PTP 550

140'; F300 CSM

130'; 900Mhz STL dish

All our equipment is using Cat5e with braid and foil tape (double shielded). Shielded/grounded RJ45's. Radios are bonded to tower. Ethernet surge protectors inside radio shack are bonded to copper ground bus which is ground to tower/electrical ground ring. This equipment, besides the CSM 300 has been up for at quite some time with absolutely no sync issues or even any real amount of ethernet errors. The 300 CSM has been up about 3 months with drops from 1G to 100M once or twice. 

A few weeks ago, we installed the FM antenna at 220' and put in a 1400w FM translator running at 700w (purposely designed so transmitter can run for years with no stress). We installed FM antenna, with proper grounds at top, bottom, and in shack for hardline coax. We swept antenna and hardline before and after installation. Basicaly no reflected power, so we know a mismatch is not the issue. I know what RF at 150Mhz and below can do to ethernet data from experience over the years. I had the Netonix GUI open when we fired up the FM transmitter.  Everything looked good as far as ethernet sync, and stayed that way for the rest of the day. The next morning I found the 300 CSM was at 100Mbps. I peformed "Port Bounce Link" in the Netonix GUI, which put it back at 1Gbps, only for it to fail back to 100Mbps later, and continue to do so every time I put it back at 1Gbps. This was not a huge deal as this 300 CSM is for back up. I figured we could dig into issue when we had more time to do so.

Fast forward to early this past week. It was raining, we get calls that evening from customers on this tower that video was buffering, slow loads, etc. Go to Netonix GUI for this tower (it was already up) and see the PTP 550 is now at 100Mbps and the 300 CSM is at 10Mbps, yes 10Mbps. The 550 is saturating the 100Mbps ethernet, hence the phone calls. I try a few times to 'Port Bounce Link" for the 550 as well, and it keeps going to 100Mbps. Try the same with the 300 CSM, and it keeps going to 10Mbps. 

I ride out to tower that same night to verify it is the new FM translator. There is also a 900Mhz STL link on this tower to an AM array 5.5 miles away (same broadcasting format as the FM translator we also installed). We installed the 900Mhz STL the same time as the FM transmitter. I flip the STL off, attemp "port bounce" on both the 550 and 300, and no luck, both go back to 100Mbps and 10Mbps respectively. Drop power on FM transmitter down to 250w and boom, 550 will sync back to 1Gbps and hold. I can not remember if the 300 went to 100Mbps or not, my main concern then was keeping the FM transmitter on air and getting our 550 back to 1Gbps. I do know the next morning after rain had stopped and everything was semi-dry, the 300 would sycn back to 100Mbps. I go back to tower and by this time the 300 will sync to 1Gbps. I turn FM transmitter power back to 700w and the 300 immediately synced back to 100Mbps. 

I come back to office and look through different ferrite core makers (Palomar, Fair-Rite, etc). There are formulas you can use to find cores that are just for the frequency of the FM transmitter (92.5, 101.3, etc). Those cores are special order many times. This FM transmitter is not a high powered station so I ordered these cores.  They cover up to 250Mhz with the sweet spot around 100Mhz (FM broadcast band). 

Cores came in Thursday afternoon. Was pouring rain all day (both units downgraded their sync once again). Yesterday we had damaging winds, so no go up the tower. Will go up Monday and get these installed as more rain expected next week. You can wind Cat5e with braid and foil 5 times around these cores (will need to file down some plastic tabs sticking into core hole). We plan to use 2 cores at ethernet input of 550 and 300. 

I am 99% sure this will fix our issues. I have been in the communications field for over 20 years. We have had to use ferrite cores at 911 centers, communications towers, CCTV ,and SCADA systems, solar systems with solar controller inside radio shack. We have used them in police cruisers to stop two-way radio from ripping analog dash cam video a new one, to turning on warning lights on/off, changing lights pattern, etc. We have used these anywhere cabling ran near two-way radio antennas and we had issue with RF getting into other equipment. These baby's nearly always fixed our issues with RF bleeding into cabling of other systems. Sometimes we had to also upgrade grounding of effected equipment, wrap HVAC tape (foil tape) on sections of unshielded cabling of effected equipment to act as a shield, and add ferrite cores. 

The 300 CSM at the other end of this link has no issues wth sync at all. But, it is on the catwalk of a water tower, connected to a Netonix switch with 20' of ethernet cable. We also have many 300-16/25's at customer locations with no sync issues. 

I would almost bet money that if everything is properly installed and grounded (both yours and the broadcast equipment), that if you add some ferrite cores at input of radio RJ45 jacks, it will fix your problem. 

1 Like
11

@MW_WISP Yes we are using a good Superior Essex double shielded ethernet cable with thixotropic filling compound and grounded connectors. If I am not mistaken it should be this one.

@CWB thank you for the detailed text. Do you use them only on the radio side or also the rack side? You use just one before the radio or more along the cable? I will give a try to these ferrites and let you know how it goes. 

12
@someone wrote:

@CWB thank you for the detailed text. Do you use them only on the radio side or also the rack side? You use just one before the radio or more along the cable? I will give a try to these ferrites and let you know how it goes. 

It all depends. I am pretty positive our problem is at the radio jack. I did buy 10 of these just in case. We put 2 on each ethernet cable as close as possible to the RJ45 jack of the radio.  I have the rest at the office. I will see tomorrow if I can take a pic of with some ethernet cable and cores the way we did ours. 

13

ferrite_cores.jpg
ferrite_cores.jpg2268×4032 344 KB

5 loops through 2 cores. I pushed our cores close together and then wrapped them with rubber splicing tape to hold them together and keep plastic housings from desentegrating in sun light. Just wrapped outer part of cores, leaving middle holes open so water does not get trapped inside cores. 

1 Like
14

So I tried your solution but again no success. I did excactly as you show in the picture, I even added about 5 of these still nothing. I tried one every 2 meters of the cable, nothing. I tried to install them also down on the rack side again nothing changes. I tried with 3 different 3000L all exactly the same behaviour. Yesterday we had a light rain and it went completely down, no link at all. After the rain stoped it came back by itself at 100Mbps HALF Duplex. 

I don't know if we are missing something or doing something wrong but putting an 1000 there has no problem at all.  We are kind of desperate with the 3000L...

15

It worked for us, but we have no 3000L's on this tower. 

Do you have the ground lug on the 3000L grounded? If so, the ground wire needs be as short as possible or it can become an antenna bringing RF onto the radio chassis. You may want to even strip the braid and foil away at the radio end of your ethernet cable, or use a non-shielded (not grounded) RJ45. Keep ethernet grounded at bottom of tower.

Also, are you using a GPS puck? Could be possible the puck coax is bringing in RF messing with the ethernet circuitry. May need to either disconnect the puck or wrap its coax around a choke.

How is everything grounded inside the tower building? Single point ground, as in each device has a direct ground to a ground bar (no daisy chain grounding from device, to device, to device, then to ground)?

Have you tried another switch or router?

It could be combo of all above, RF interference/overload can be a realy headache, and sometimes only fixed by trial and error.

Also, does the 1000's hold 1Gbps sync on this tower in the rain, or do they go to something like 100Mbps full? If they hold 1Gbps, and the 3000L's are shutting down (especially if it is a direct replacement of just the radio) I would open a tick with Cambium. Probably not a lot they can do, but it would bring to their attention, and it would be logged, that the 3000L is possibly much more prone to RF interference on the ethernet interface. If there is enough complaints, Cambium will have to investigate this further, and possibly even modify the circuit in future productions, if they find the 3000L more susceptible than their other products. 

I just did a search  and there is some pretty good information on chokes for this issue.

In fact , this search puts a link back to this forum at the very top of the list. Newkirk shows with pics how his company killed sync issues on a several KW FM broadcast tower with ferrite cores. 

One last thing, if your ethernet is running up same leg as any of the transmission runs, I would definitely move it to another leg of the tower. Just throwing some ideas out. 

1 Like
16

Some info:

Rain on an rf cable will make the cable act like a capacitor to the tower. As we know capacitors allow ac frequencies through. 

My solution: your mixing FM with ethernet, just avoid the twisted pair issue all together. A pair of comtrend sfp to ethernet and a dc to dc power injector with a couple capacitors across the buss and a couple to tower ground. Run fiber up the tower, its not overly expensive anymore. 

Some things to try:

run your cable up the inside of the tower leg away from the FM cable. The tower will act like a faraday cage to some extent and provide critical shielding.

This will sound as bad advice, but read all the way through before you try anything or dismiss any part of it.

Contrary to best practices, ground only the bottom of your data cables and do not ground the radio case. The data cable and antenna provides this protection, dont create ground loops which can act as antenna coils.

Make sure the FM antennas are not directly below your 3000L. No amount of vertical separation will help, make sure they are 180deg opposite side of the tower, maybe also install a ice guard to act as an rf deflector.

Make sure your antenna is protected with a lightning aerial that is at least 48" above your antenna. This is less to do with the FM problem but to protect from another problem.

Ferrites are near a must when colocating with an FM station, but you can add all the ferrites you want it wont help if your cable is cut to a length which is a near harmonic of the broadcast frequency. Just because the tower is 235' doesnt mean your cable is 235'.

Does your data cable have any mid-span shield grounds? If so remove them. Same with mid-span LPUs. 

Once you are able to keep the ethernet link stable, add the radio case ground. If all is good, add back the next piece until you find what is causing the link to falter. 

Good luck!