Question for Jerry on RF balancing

I just finished reading your posting on balancing RF and it was great. Thanks for all your contributions. One question poped up after thinking about it for a while. I understand the principle where a radio with a strong signal may interfere with another with a weaker signal, but, how is that a problem in a system where everything is in sync? Just trying to have a better understanding. We have seen problems in our 900 Mhz system that clearly seems like is related to self interference as described in your posting.

Thank you

Sebastian

“rf balancing” is really more of an issue for systems with no polling mechanism. as long as you have enough control slots allocated, it really shouldn’t be necessary on a canopy network. eh?

Each AP has one SM transmitting to it at a time, however in a network with multiple AP’s, there are multiple SM’s transmitting at the same time.

By balancing the RF levels, you ensure that every AP can hear it’s SM.

Thank you… :smiley:

The technical reasoning behind “RF balancing” is simple-

In 802.11, you have potential problems where one station can transmit at a much higher signal strength than other stations. The one station can transmit nearly continuously, and drown out other stations for long periods of time (long can be milliseconds, or from user perception it can last for seconds). RTS/CTS is supposed to fix this but the 802.11 committee used the wrong timings so it didn’t work all that well.

Motorola never had this problem, because the radios already use a polling mechanism during the control slot time. Add more control slots if the radios don’t have enough time to request transmit slots… “RF balancing” should not really be necessary if you have enough control slots.

However there is still one other possible reason that “RF balancing” helps:

If the signal is too strong, the AGC (Automatic Gain Control) circuit in the RF front end on the Canopy (AP/SM) will turn down the signal so that the demodulator can properly interpret it. The same thing happens on pretty much any other CMOS radio (like 802.11 products) because everyone uses the same basic design.

If the AGC turns down the signal then it turns down everything else the radio hears, too. So it turns down the signal from other customers and they become harder to hear, quite literally.

The AGC shouldn’t turn down the gain for very long, only for as long as the AGC circuit detects the strong input signal, With enough control slots, this should also not be a problem. But it’s a possibility that a customer keeps the AGC engaged long enough to make weaker customers hard to hear.

If you don’t have enough control slots and a lot of SM traffic, then Canopy acts just like overloaded 802.11. It has been very, very hard for me to get an AP to that point. And since I can’t read the registers from the FPGA’s virtual radio hardware (and I have no documentation to tell me what the registers mean anyways) then this is all just a bunch of guesswork and hypothesis. Oh yeah we don’t have any documentation on the Canopy MAC layer either, which makes any more detailed analysis impossible.

I understand the control slot part. Great info. I went to a Moto training a few years ago (Advantage or HW scheduling didn’t exist yet) and I remember them teaching something about the AP listening to more than one SM at a time and making up 1 uplink frame with it. I don’t know if this is still the case or if I even remeber it rigth :?

Bottom line, balancing RF should not be necessary, but under certain situtations it could turn out to be a problem. I sitll feel that I have more control if I keep it balanced, the problem is my field techs don’t see it that way. I also guess that this is probably more of a problem in a 900 mhz system that it will ever be on a 5.7 or higher freq. is that rigth?

One of the issues we see with our 900 is that some SMs (about 10 out of 200) will not come back online after rebooted. When they are online, they are fine and stable but as soon as you reboot the SM it could take anywhere from 20 min to 1 hour to re-register. However, during this period, you can reboot the APs at any time, and those SMs will come back. Guaranteed. I haven’t even bother to call support because I know what their answer is. INTERFERENCE. The bad part about that is they will not give me any suggestions on what to do or how to test it. Anyway, my point is, if it was interference, why would it come back by just rebooting the APs? Why wouldn’t go down by itself? because interference will knock the SM out anytime, not just when the SM is rebooted. I can not blame anything else other than may be this RF balancing problem.

Would love to hear more

I had to go through and balance all of my SM power levels when we started having SMs registering to the wrong APs - I also implemented color codes about the same time.

My levels were so different from high to low I had one customer who connected to an AP 18 miles away, instead of the one 3 miles from him. After balancing everything out, he was able to reconnect to the 3 mile distanced AP without issue - I also saw a lot of other improvements with signal levels.

I would say RF balancing, although not “necessary” should be good practice anyway.

Typically there are more than 1 AP per tower.

While one SM is talking to one AP at a time, the RF energy from multiple SM’s is still hitting the same tower. If one SM is hitting one AP on the tower with -50dB and another SM is hitting an adjacent AP with -78dB, I can guarantee you that the weaker SM is going to be harder to hear.

By balancing the power to within a 10dB window you will improve the overall performance and sensitivity of all of the AP’s on the tower. Additionally, 900 operates at pretty impressive distances. By balancing the power on every tower, you will improve performance network wide.

I can testify that it matters, as I have tested it. I have customers ranging from 3 miles to 17 miles across 3 AP’s on the same tower. Before balancing I was having difficulty getting good links at 14 miles. After balancing I get rock solid links at 17 miles.

Yeah, multiple APs can hear the same SM, much like multiple SMs can hear the same AP.

If your antennas have a reasonable front/back ratio then you really shouldn’t have to worry about this. After all, the APs that point in the same/similar direction are on different frequencies. So the fact that they both hear an SM is irrelevant since they should NEVER overlap in frequencies (or you will have other major problems anyways)

And so your APs that are on the SAME frequency are back-to-back, that is, the SM would be heard by the far AP but at a very weak level because the signal would literally be coming in the back of the antenna. If your antennas don’t have a high F/B ratio, then this might not work as well. And you’re using the wrong antennas for this type of deployment. Putting two APs on the same tower with the same frequency is truly a novelty and it requires special considerations :slight_smile:

Also, Jerry, I’m not disputing that this technique improves your network. I’m just trying to analyze the complexities with the limited information I know about the technology. I have several AP sites set to 20 miles, and customers within 15-20 miles are usually stable on my network unless there are interference isuses. It’s very complicated and fun at the same time until everyone starts calling when something goes down… Note that I don’t do the ‘rf balancing’ trick except for customers who are very close into an AP…I don’t want one customer at -40 and another at -80 on the same AP, that’s definitely outside of my threshold of pain.

I have a few SM’s pretty close to my tower… 1000 feet or so as well as having SM upwards of 3 miles away through heavy, wet trees. This condition gives me an RF power spread on my sms of -40 to -80.

In my testing of RF balancing, I set up an SM about 500 feet away from the tower and started ratcheting down the TX power on the SM. At 20 dBm, I was still getting -54/-55 at the tower. As soon as I took the SM TX power to 19 dBm, it completely disconnected and would not reconnect until I put it back to 20dBm. So I left it there. The next morning I checked on the SM and found that it had dropped off. We had a foggy morning and the water in the air was enough to attenuate the test SM to AP connection. So I set the SM TX pwr to 21 and it reconnected through the fog… the AP was seeing the SM at -55.

The difference between setting the SM TX pwr between 19 dBm and 21 dBm seems to be quite large while the AP never saw much change in the signal power it saw from the test SM. This adjustment seems rather course.

Has anyone on the forum tested the output power from a 900 SM with a callibrated power meter…?? I am curious as to how accurate this setting actually is.

Thanks in advance.