Uneven Chains

The time has come to sort this issue out once and for all. Please tell me if you experience uneven chains when looking at e-align with epmp products.

Right now if I look at a bunch of F300-25 radios connected to a ePMP3000L I see massive differences in the chains of most subscribers. Its not always like this and sometimes the chains are even.

I’m nnot sure if it’s a reporting issue or real as throughput to these subs is good but, some are so bad, like -64 and -74 etc. I’d expect terrible performance with this but maybe it can handle, I just don;t know.

What I do know is that sometimes is like this and sometimes it’s not. It’s not interference so please don’t … and even if there was it wouldnt explain it, it’s not an SNR reading, it’s RSSI.

This has been happening for a long time on and off. Ive had a couple of ptp links using f300-25 that have done the same. They would self reboot too. I removed them and put Ligowaves in.

I have noticed it’s mainly f300-25 that show uneven chains and hardly any f300-16. I doubt this is due to beam width and it’s definitely not poor alignment.

What gives?

Cheers!

1 Like

Hello Riddle,

You are correct in your assessment - unbalanced chains are not caused by mis-alignment, or certainly not interference. On any RF system that uses MIMO - where we have data sent from a transmitter to a receiver in 2 or more streams - we always have some amount of interaction between the streams. They are only separated by the cross-polarization isolation provided by the receive antenna. In an ideal RF world, the MIMO streams are separated by a perfect 90 degrees 0 either V to H polarity, or -45 to + 45 slant polarity - depending on the RF system. On a great link, the unbalance between the chains is the difference in propagation loss between the 2 polarities that make up the MIMO stream.

Anytime we have multiple RF streams transmitted, we have some kind of interaction between the streams. The signal is transmitted as close to perfect as possible at the antenna, but stuff starts to happen right there. An RF link is always affected by its environment (for example, the tower steel, ground effects (the earth always has some effect on a path - in the PTP world, that interaction is called dispersive fading), clutter effects such as the signal passing over trees, along the sides of buildings, etc. In our fixed wireless world, we have 2 predominant wireless modes called MIMO A and MIMO B. When the link has high enough SNR, it operates in MIMO B, where each stream has different data, effectively doubling the capacity of the RF system. When the link quality is poor, it operates in MIMO A, where both streams are carrying the same data, increasing the sensitivity of the receiver to improve the link as much as possible by combining both streams. Stream imbalance is caused by some interaction with the environment on one polarity, but the effect is less on the other polarity. The effect is that the link operates at the lower quality level of the 2 streams, so in your example RSSI case, the link will behave from an error rate / retransmission / modulation coding scheme perspective as the -74 dBm RSSI, not the -64 dBm RSSI. And in fact, if you do a throughput test on the link, the lowest performance will be the same as though the RSSI were less that -74 dBm, but the best performance will be something less than -64 dBm performance you’d expect from the throughput tables in the spec sheet.

Now the question is what to do about it? From 20 years experience in MIMO based systems, here’s my recommendations:

  1. Make sure that the AP antenna is mounted away from the tower steel by the length of a few odd half wave multiples. I know that means some math, but this will make sure the interaction with the tower is minimal.
  2. Make sure the SM antenna is mounted clear of the building roof. It should be the highest thing installed on the building, with the mounting pipe extended above the SM. Ideally, the SM should be in an imaginary 60 degree shadow created by the end of the mounting pipe above the SM.
  3. If you are NLOS, find the location for the SM on the roof that has the least effect from the NLOS element. This means that the really nice mounting on the roof at one end of the building may not be the best practice install location. I call this ‘installer diversity’.
  4. Make sure both the AP antenna and SM are installed with vertical being vertical. Twisted mounts cause polarity mismatch, which can lead to imbalance.
  5. Don’t install the SM so that the path runs along the building, but make sure you are as close to perpendicular to the building as you can be. Polarity twist occurs when the path is influenced by the building, or any other ground clutter, and that also contributes to stream mismatch.
  6. Finally, be very critical about installing SM’s with good performance. Set a minimum acceptable performance level, and do your best to never connect an SM that is below that threshold. Then, when you do have imbalance, the effect is minimized. This is rule number 1 on the best performing networks!

If you watch the link quality to your cellphone, as you move along carrying the phone, you’ll see your SNR go from spectacular to 0 in just a few steps. We accept that level of performance on our cell. In the fixed world, though, we have to provide some performance guarantees, and we can’t move our subscribers residence or business location, so we have to make sure the AP’s and SM’s are installed as perfect as possible - then we have to learn to live with the environment we install them into.

Sorry for the length of my response, but I hope this helps some of you!

2 Likes

Yes!! I have the same issue… one day customers chains are good, next day the chains are off… I thought I was the only one and all I got for an answer was its interference and I do not believe this is the case. Just curious, what firmware are you using?

Thanks for reply Dave but i’m aware of all this stuff.

There’s more than one issue at play here, I think three.

  1. I have replaced a radio with uneven chains (8dbm) and seen the issue completely resolved. This is a faulty radio.

  2. I have seen an AP have this issue with most of it’s SMs (25 on this particular tower), but performance still be fine, then completely gone the next time I check in on it. Must be AP issues surely.

  3. I have seen a ptp link with the issue and replacing the radios resolved it. Faulty radio.

So, two of the scenarios were caused by faulty radios and one was caused by who knows what. This is the one I need to reslove. I have openend a ticket with Cambium and will report back any useful stuff that comes from it.

When I link test these radios (scenario #2) the results are not horrible. I have manually set MAX MCS at CPE end to reduce retransmits (0% for every CPE) so only get 50-170Mbps to each SM on a 40Mhz channel. I’d expect to see worse effects with a [sometimes] 11dmb difference in chains.

Keen to hear from others experiencing an imbalance in chains of more than 5dbm using the ePMP line with fw 4.6.1 or newer.

Cheers