Another tale from the the land downunder. This time for all you RF geeks. I apologise in advance if I use dB, dBm, and dBi incorrectly, I tend to use them interchangeably at work.
One of those random things I have to do is support wireless gear that our ISP sells on the side to system integrators for point to point wireless between buildings.
It is fairly easy work, we over engineer the links to perform better than the system integrators expect. This is a story about how the original engineer over engineered the link too much.
The link was installed about 6 years ago and from what I understand hasn’t performed as expected.
In the office, at my desk working on how one of our transit providers fudged up their route map and was advertising our address space back to us, a story for another time maybe.
phone rings
Me: G’day slazer speaking.
Customer: Hi, its Customer from [redacted], we bought a wireless link from your firm few years ago and it has been working mostly well till last week when it fell over and we haven’t been able to get it back.
Me: Ooooookk, let me grab your details and I will give it a crack.
Customer: The box in the rack says Redline AN50E and the link light is off.
Me: all right, do you still have management access to the radio?
Customer: I do on this side, not the remote site obviously.
Me: Makes sense. on the status page what are the RSSI and SNR values
Customer: RSSI says -86 dbm for all 3 values and SNR is 0 dBm
Me: Is the other end powered on?
Customer: Yes, the guys in the other office can login to the management as well.
Me: That’s good, can they tell you the values on that side too?
hold music starts
Customer: They are seeing the same values.
damn
Me: Do you mind if we come down and have a look?
Customer: No worries mate, just ask for me at reception.
I make a list of kit we will need for the job and “delegate” it to my minion to load into the van and we head out.
We get to site and Customer shows us around the master end of the link. I spot the first of many problems. The ethernet is running in half duplex mode (may account for their poor performance.) and the radio is running at 20dB transmit power.
I turn to Customer.
Me: have you played with any of these settings?
Customer: When it was originally installed the tech said if we have any problems with the link we should turn the transmit power up to 20.
I stare blankly at him for a few seconds before double checking I’m not going insane. I make note of the usual misconfiguration suspects, frequency, channel size, encryption enabled, correct encryption key and drop the transmit power down to 1 dB. We head over to the slave end.
Most of the settings are correct, with the exception of transmit power, again it is running at 20dB. I drop it back to 1dB and see the SNR come up above zero for a few seconds before disappearing.
We do a test on the indoor coax cable going to the roof and see no RF coming back down the cable. Damn a faulty outdoor unit. So we head up to the roof and see what we can do about the outdoor unit.
I let my minion and Customer go up the ladder first and as I pop my head out of the roof access hole I see a disaster.
The original tech installed a 60cm panel for a rf link which is no more than 50M. Rf geeks will know why this is a disaster. 20dB of transmit power along with a 28dBi antenna, no way that is legal in Australia.
We swap out the outdoor unit on the slave site, because we were on that side, and as soon as we plugged in the new outdoor unit it started chirping away with its alignment buzzer saying it has the maximum modulation.
Me: That’s not good.
Minion: What do you mean? The link is working with this new outdoor unit, so we found the faulty part.
Me: Yea, but where is the antenna connected at the moment?
Minion: In the faulty unit.
Me: Yes, so with 1 dB transmit power on both end and only one 30cm panel on the master side we are forming a link.
Minion: So?
Me: What do you think will happen when we attach the 60 cm panel and put the transmit power back to 20dB?
Minion: It will get saturated and the link will fail.
Me: Yes, so all the drop outs they are talking about is because the link was overengineered too much.
We reattached the panel and looked at the management RSSI -36dB, SNR 30dB.
Me: That has sorted it.
phone rings, Customer comes up on caller ID
Me: Hi mate, we got it back up, how is it looking?
Customer: The link light is on, but I cant ping across the link.
damn it Rf is up and talking but no traffic is passing, the encryption key must be wrong. I get him to correct the encryption key and his traffic starts flowing again.
I confirm the modulation and transmit power are ok and head back over to the master end to talk with Customer.
Me: The outdoor unit is most likely to have burnt out because the RF levels were too strong.
Customer: I notice now when I put the transmit power to 20dB the link goes offline.
Me: Never change that value to above 1 ever
Customer: Ok then. The speeds are better, before it was running between 6 and 12 Mb/s now it is saying 54Mb/s
Me: Yes, because of RF magic we turned the signal power down to get a better signal.
Customer: I’ll accept that.
And with that done, Minion and I went back to the office.
Context for those who aren’t in the RF world. Imagine having a conversation with someone across an alleyway with one person shouting at the top of their lungs and the other using a megaphone. At some point hearing damage kicks in.
The original tech installed a 60cm panel for a rf link which is no more than 50M.
In case anyone else has this minor confusion, this is a radio link between two buildings which are 50 meters apart. And a square, directional flat-panel antenna that is 60 centimeters on a side is grossly overkill for the short distance involved.
Why does the power increase kill the signal?
Ever tried having a conversation with someone when they are shouting in your ear in a quiet room?
Sure you heard them, but did you understand what they said?Thats what is happening here. Normally you would pre calculate the required transmit power to get your preferred wireless speed.
Normally, increased RF power is helpful to overcome interference or noise. Just like how normally, an automobile or airplane will be easier to operate if it has a bigger engine with more power.
At some point, though, the extra RF or engine power is no longer beneficial but also isn’t harmful. And if you go significantly beyond that, then you end up in a region where the extra power is downright harmful and is actively working against you.
Consider an automobile driving in a rainstorm. Having more power is bad, because the tires can lose grip more easily, leading to a crash. In an airplane that has gotten into a stall, applying power is the wrong solution and just aggravates the stall, which is not good.
Here, adding more RF power is just cooking the other receiver like it’s a Thanksgiving turkey. The extra power is no longer helpful for making communications, and may be physically damaging the receiver.
So in this case when they switched to the smaller 30cm, they reduced the ability for the high power signal to damage the components past the 30cm dish?
Almost. By virtue of being a smaller antenna, the 30 cm panel does not focus its energy as tightly as a larger (eg 60 cm) antenna. Likewise, a smaller antenna does not pick up (ie receive) as much energy as a larger antenna does. Thus, by using a 30 cm panel, less of the high energy from the opposite radio will reach the receiver, and that keeps the receiver from being damaged.
In RF engineering, there is a careful balance between output power, antenna size/shape, environmental conditions, and desired link quality. Whoever built the radio link originally did not apparently perform the necessary calculation. I’m not an RF engineer, but for spanning a mere 50 meters, this 20 cm antenna with built-in radio should be more than sufficient for a basic link.
Thank you for all the information. Very well written.