We have had a PTP 550 connectorised link up for several months now using a 2ft KP Performance dish. We have been using 5240 and 5745, both 20Mhz wide. Works like a champ with MCS 9 all around with our lowest percentage at MCS 9 coming in at 98.6%. Link is using 4.3.2.1.
We just ordered another PTP550 link. We set thes up in a test scenario. We used similar frequencies; one frequency in the 5.2 band and one in the 5.8 band, MCS 9 all around at 100% again. This link is a short 1.8 mile shot. We decided to use DFS channels for this, or at least test them. We used 5.265 and 5.300. Both were extremely clean channels. Set transmit power to 3dbm since we are using a 27db dish. RSSI was around -60. Ran some wireless link tests using the web GUI. Only getting around 170Mbps down and 45Mbps up instead or our regular 230down 65up. Went and looked at performance page, and found modulation rates varying from MCS 6-9 for both radios. MCS 9 was only around 60%. Ran transmit power up 10db for some more wireless link tests. Throughputs were worse, and modulations percentages were down.
At this point it hit me that each radio must be using it's own internal clock even with channel bonding enabled for TDD timing. If that was the case, this would deifinitetly explain the low throughput and MCS percentages scattered all over the place. I left the first radio on 5.265, and put second radio on 5.550, power levels the same. Throughput was around 200Mbps down and 50+Mbps up. MCS 9 was now close to 70%. The frequency spacing definitely helped. Changed radio 2 back the 5.800 channel I was first using, and bam..everything back to 100% MCS 9 up and down for both radios.
I did not not perform individual radio tests, as my time was limited. Specturm was extremely clean. The frequencies we did not change during the test (5.265) had more throughput and MCS 9 percentage got higher and higher as we increased frequency spacing. And when we used 5.800 for the second radio, the 5.265 radio had 100% modulation at MCS 9, so there was no other interference on that frequency.
It seems to me the only fix is to use external sync such as cnPulse when using channel bonding so both radios transmit and receive at the same time. I have looked through docs for the PTP550 and rewatched videos pertaining to the PTP550. I see nothing that says if you want to run channel bonding with frequencies fairly close to each other, you must use a sync source. It may be there and I have missed it. My tests seem to point to you must. I am guessing this is because even with channel bonding enabled both radios are using their own internal timing source for TDD operation???? If that is the case they are transmitting and receiving at different times, causing self interference due to using the same antenna.
Can someone from Cambium please chime in on this??
We have had a PTP 550 connectorised link up for several months now using a 2ft KP Performance dish. We have been using 5240 and 5745, both 20Mhz wide. Works like a champ with MCS 9 all around with our lowest percentage at MCS 9 coming in at 98.6%. Link is using 4.3.2.1.
We just ordered another PTP550 link. We set thes up in a test scenario. We used similar frequencies; one frequency in the 5.2 band and one in the 5.8 band, MCS 9 all around at 100% again. This link is a short 1.8 mile shot. We decided to use DFS channels for this, or at least test them. We used 5.265 and 5.300. Both were extremely clean channels. Set transmit power to 3dbm since we are using a 27db dish. RSSI was around -60. Ran some wireless link tests using the web GUI. Only getting around 170Mbps down and 45Mbps up instead or our regular 230down 65up. Went and looked at performance page, and found modulation rates varying from MCS 6-9 for both radios. MCS 9 was only around 60%. Ran transmit power up 10db for some more wireless link tests. Throughputs were worse, and modulations percentages were down.
At this point it hit me that each radio must be using it's own internal clock even with channel bonding enabled for TDD timing. If that was the case, this would deifinitetly explain the low throughput and MCS percentages scattered all over the place. I left the first radio on 5.265, and put second radio on 5.550, power levels the same. Throughput was around 200Mbps down and 50+Mbps up. MCS 9 was now close to 70%. The frequency spacing definitely helped. Changed radio 2 back the 5.800 channel I was first using, and bam..everything back to 100% MCS 9 up and down for both radios.
I did not not perform individual radio tests, as my time was limited. Specturm was extremely clean. The frequencies we did not change during the test (5.265) had more throughput and MCS 9 percentage got higher and higher as we increased frequency spacing. And when we used 5.800 for the second radio, the 5.265 radio had 100% modulation at MCS 9, so there was no other interference on that frequency.
It seems to me the only fix is to use external sync such as cnPulse when using channel bonding so both radios transmit and receive at the same time. I have looked through docs for the PTP550 and rewatched videos pertaining to the PTP550. I see nothing that says if you want to run channel bonding with frequencies fairly close to each other, you must use a sync source. It may be there and I have missed it. My tests seem to point to you must. I am guessing this is because even with channel bonding enabled both radios are using their own internal timing source for TDD operation???? If that is the case they are transmitting and receiving at different times, causing self interference due to using the same antenna.
Can someone from Cambium please chime in on this??
There is some level of self-interference happening in this situation when the bonded channels are very close to each other. The two radios are internally sync'd in software using a TDD system but uses the same clock source. An external clock source is not necessarily the answer. We recommend about 100MHz of separation between the two radios please. We will update our documentation accordingly.
There is some level of self-interference happening in this situation when the bonded channels are very close to each other. The two radios are internally sync'd in software using a TDD system but uses the same clock source. An external clock source is not necessarily the answer. We recommend about 100MHz of separation between the two radios please. We will update our documentation accordingly.
Thanks for your reply. You answered all my questions. We are seeing around 300Mhz of separation before what I would consider stable enough modulation rates to not have to lock a certain MCS down. We ordered a cnPulse first thing this morning to see if it helped. We will use it somehwere if not here.