I am very new with Cambium radios, so I apologize when I say something ignorant...
I am working with a co-op that is testing an ePMP 1000 GPS radio on a tower along side a Ubiquiti Rocket5AC (different channels...) We are evaluating which solution to go with for a PtMP deployment. The subscribers are all 1.5 to 3 miles from the AP. The problem is, we are trying to get 75+ Mbps to the subscriber.
The good news is, subscriber numbers per AP will be low (10 - 15 per AP). We are currently using the integrated radio as the CPE. May try a Force 180 soon...
When I put the AP in 75/25 or 50/50 mode, the download speeds to the CPE are in the 30 - 50 Mbps range. When I go to "Flexible", they jump to 100 - 120 Mbps. In fact, upload speeds increase from ~5Mpbs in either of the ratio modes to ~30Mbps in flexible.
I know I loose GPS sync in flexible mode - that makes total sense. However, I only have 2 APs on this tower so I am not too worried about frequency re-use.
What am I missing that will hurt me down the road if I run in flexible mode? Are there other ways to maximize the speeds to the CPE in 50/50 mode (which is what we want ideally) to get 75 - 100 Mbps down?
These tests have been done in channel widths from 10 to 40 Mhz. We have stuck with 40 for best constant results.
Our Ubiquiti tests are mixed. Our problem with Ubiquiti is I can't get them in the 5.2 range (hardware restricted) which is where we want to be due to noise in 5.7 and 5.8.
Based on your notes here there is really no downside to running your sectors in flex mode. If you are looking for additional capacity you can also try running them with 5MS frames. I've noticed some improvement in total speed on GPS mode as well with the larger fame.
Running only two APs on a tower, you have limited use cases for GPS. With ePMP 1000 you can operate up to 3 APs on a tower in the UNI 1 band giving each a 5MHZ guard band from each other with 20 Mhz channels. This is one of the big improvements you will see with ePMP over any of the UBNT M, or AC products. In our testing we have found UBNT products to need more like a 20Mhz guard band between APs, and ePMP APs do really well with only 5Mhz even in flexible frame mode. As for possible pain down the road, there really won't be much. Worst case is if you need to start moving to a GPS Sycn mode for additional tower capacity you can just split your subscribers off on additional sectors for added capacity. As for the best channel size, I would highly advise considering keeping it at 20Mhz, with flexible mode you will see capacity around 90-100 Mbps with good signal to noise ratios.
As for your testing with ePMP radios using GPS, with 20 Mhz channels we get near 80Mbps down, and about 20 Mbps up with near 30 SNR on clients in production.
On your shorter shots the ePMP Force 180 radios will be a good choice, I would look at the Force 110 or the upcoming force 200 5ghz radios for your shots closer to three miles.
I'm having possible SNR issues. Having a hard time getting much better than 23/25. I am not too happy with the spectrum analyzer in Cambium. Hate the fact it takes the radio off line. Since I am testing this side by side with Ubiquiti gear, I can use their spectrum tools, which I like. Noise floor is down around -95 at both the AP and client sites inthe 5.2 spectrum. My SNR on UBNT gear is -30 or better.
If I pop my ePMP gear up to 5.8 on a 20 Mhz channel, throughput drops considerably in both directions. Noise floor is around -85 at this freq. First guess would be noise. However, I put a Rocket 5AC on that same channel and I get great throughput. Again, I am thinking noise. But why such a difference? For now, the client CPEs are ePMP 1000 integrated radios. I have a Force 180 on order to test.
Just logged into the AP to drop to a 20 Mhz and test. This is how it greeted me (screenshot). Nothing I do seems to fix it. I'm a bit afraid to request a reboot since the radio is way up on a tower on top of a mountain.... Does this screen make sense to anyone familiar with these? it is still operating fine, I just can't configure it at all.
Clear your web browsers cache and re-open the page. The web UI seems sometimes buggy if you do a firmware upgrade. I've never had an issue after rebooting a radio, so I would be shocked if you did.
I would HIGHLY consider running Force110, the current dish design is a bit of a pain to assemble but the 25DB dish will give you much improved performance, they normally don't cost much more either. You can pick up the Force110 Radios for around $112.50 plus shipping, most vendors like to sell them in packs of 4. What sector size are you running? The standard 90 Degree? What size of sector would be required to cover your ideal area?
Yeah, rebooting is a drag. We actually often times try to deploy an extra radio on our towers to keep one in spectrum analyzer mode.
With regards to your UBNT radios, what type of sector are you running, and what are your client antennas?
I've tried three separate browsers and all load the same page, or lack there of. I'll try a reboot when I know my test clients are away, just in case.
I am running a 120 sector (RF Elements) on this radio. I am testing a TwistPort 60 sector soon. The reason I have a 120 degree is not for the horizontal but for the vertical beam size. Our antenna is 1700' above the town with clients at 1 mile to 3 miles. Need the vertical width. Would much prefer two 60's side by side but I loose too much vertical. Really hoping the TwistPorts work!
The Ubiquiti I am next to is a 60 degree shielded sector (Ubiquiti brand). Clients I am testing are on the small NanoBeam-AC-19 CPE's. I have one PowerBeam 300 ISO CPE that works great - as I would expect.
Looking forward to the Force 200 5Ghz radios. Would like to see some isolation built into the domes...
I may pick up a Force 110. The harder sell will be to mount that on some of the client homes.... Some are in a "ritzy" neighborhood. You know how that goes. :-)
We have 4 twist ports deployed, at 30 degrees. We are extremely happy, feeding data back to RF-Elements on the Cambium Twist Ports later this week. I would advise in a high noise environment to look at the 30 Degree TwistPort antennas, and running more APs. That's where we have been moving, before that we had been running the 90 Degree cambium provided antennas. Based on your description of the tower site, it sounds like the Twist Ports will really make a big difference in your signal levels! Are you also running a RF-Elements beta for ePMP Twist ports?
I've actually also heard good things about the Rocket AC antennas with ePMP radios, but based on your tower angle it still sounds like the twist ports will be the magic you need. I'm guessing the Force 200 Radios will cost slightly more, but the savings in time will be will worth the extra few dollars.
I am in the beta for TwistPorts (I think). Been talking to one of the prinicpals at RF Elements and he is sending me an adapter for the ePMP as well as the Rocket. I bought two 60 degree horns but based on suggestions, may downgrade to 30's. That means I need 4 of them to cover our town and I don't have enough clear frequencies for that, though... 5.8 is a bit crowded on our mountain (some radio stations using it for studio links). Would you re-use a frequency without GPS sync 30 degrees away from another antenna at the same freq?
Have you tried the TwistPort in the 5.2 range? They are not listed below 5.4 but RF Elements gave me the impression people were testing them in 5.2 with good results.
The integrated units have too little gain for your further out clients. I wouldn't use them for much past a mile PtMP. The 180 will be a slight improvement but you really need the added gain of the Force 110 for your 3 mile shots.
The sell to your subscribers on the bigger dish is that performance is better and reliability is improved with the added SNR you will see. The small CPE will come back to haunt you with poor performance in the end when the noise floor inevitably gets worse.
Funny you should ask, I have 3 twist ports running in 5.2 and they seem to be working great. Yeah, I'm working with the same person. Small world. As for frequency re-use, no you would want to use GPS, and identicle settings on all APs. Same Max distance, same frame size. You could likely get re-use with as little as 60 degrees off each AP. Right now we have two APs pointed 80 degrees off each other from center, and SMs in oppsite area can hear the primary AP at something around -87, so fairly close to the noise floor in our area. We haven't yet enabled GPS Sync and actually started stacking APs on same channels, but we do plan on testing that soon. There is a slighy loss of AP capacity (5 or so %) going from Flex to 75/25.
You can get the rocket twistports from StreakWave, so we just picked those up locally. At this time we have not deployed any TwistPorts on UBNT, mainly due to a lack of time.
I'm guessing once Force 200 hits in 5Ghz they want all the Force110's to be cleaned out. I'm trying to get my hands on another 20 pack my self. I'll let you know if I happen to find any Monday.
Streakwave did on Monday or Tuesday but they seem to be gone now. Double Radius also had some last week, I'm not sure about now.
It would be annoying if Cambium really did stop producing the 110 even after the new model hits distributors, a little overlap is nice for having spares of each on hand.
Something must be going on. No one has any Force 110's in stock (that I can find)....
The channel is being flushed of 110's to make way for the 200's... should be coming this way on a boat from China... The 180's are available and work very well. If you're making short 1-5 mile shots I think you'll be very pleased with the Force 180 over the original integrated model.
I spent a couple hours today taking my AP through all frequencies in 5.1 -> 5.8 at 20 Mhz channels. I have a couple radios out at 2 miles I used for speed tests back to the AP.
With a 20 Mhz channel, I can not break 50 down and 15 up using the 1000 integrated CPE (I have a Force 110 coming).
What concerns me more is the SNR. My SNR is consistently 24/21 with an MCS usually of 11/11
I've attached a screen shot from a UBNT radio pointing exactly the same direcion as the ePMP sector from the same tower that shows the RF environment. It's dead quiet in the 5.1 - 5.2 bands as well as DFS. If I pick some clean holes in 5.7 or 5.8 (of which there are a few), nothing really improves. I've re-pointed the CPEs and have the best signal I can get (-66).
Limitation of the 1000 integrated CPE? Both CPEs sitting in dead lobes of the sector (they are 1 mile apart horizontally)?? I should have better performance, IMO.
I wouldn't expect much more than that from the integrated unit at that distance, it only has 13dbi gain and a wide beamwidth, it's pretty susceptible to noise.
Also, the spectrum analyzer on the UBNT AC radios is VERY optimistic. Try running the same using the ePMP sectrum analyzer or a UBNT M series. I would guess that things look much noisier.
I spent a couple hours today taking my AP through all frequencies in 5.1 -> 5.8 at 20 Mhz channels. I have a couple radios out at 2 miles I used for speed tests back to the AP.
With a 20 Mhz channel, I can not break 50 down and 15 up using the 1000 integrated CPE (I have a Force 110 coming).
What concerns me more is the SNR. My SNR is consistently 24/21 with an MCS usually of 11/11
I've attached a screen shot from a UBNT radio pointing exactly the same direcion as the ePMP sector from the same tower that shows the RF environment. It's dead quiet in the 5.1 - 5.2 bands as well as DFS. If I pick some clean holes in 5.7 or 5.8 (of which there are a few), nothing really improves. I've re-pointed the CPEs and have the best signal I can get (-66).
Limitation of the 1000 integrated CPE? Both CPEs sitting in dead lobes of the sector (they are 1 mile apart horizontally)?? I should have better performance, IMO.
A few things -
We are working on a web based built in spectrum analyzer that does not take the radio offline. This is high on the list of requests from our customer base. In the meantime, please note that ePMP spectrum analyzer shows the noise floor relative to the channel bandwidth and what the SM sees. Other products show the noise floor at a sub carrier level which is native to wifi chipsets.
In terms of Force 110s, yes, they are a bit tight on the NA market but we are trying to assist with that situation. Last but not least, the Force 180s are a 15 degree azimuth beamwidth and has at least 3dBi gain improvement over the ePMP1000 which is sitting at 13 and a 30 degree azimuth. Your tests with UBNT might also be not quite apples to apples in terms of gain. Since you are a new customer trying to test ePMP, I could send you a sample of Force180 that you could try at one of your sites.
We are working on a web based built in spectrum analyzer that does not take the radio offline. This is high on the list of requests from our customer base. In the meantime, please note that ePMP spectrum analyzer shows the noise floor relative to the channel bandwidth and what the SM sees. Other products show the noise floor at a sub carrier level which is native to wifi chipsets.
In terms of Force 110s, yes, they are a bit tight on the NA market but we are trying to assist with that situation. Last but not least, the Force 180s are a 15 degree azimuth beamwidth and has at least 3dBi gain improvement over the ePMP1000 which is sitting at 13 and a 30 degree azimuth. Your tests with UBNT might also be not quite apples to apples in terms of gain. Since you are a new customer trying to test ePMP, I could send you a sample of Force180 that you could try at one of your sites.
Sakid
I'd love to try out a couple Force 180's. I am needing to make a decision in the next few weeks if we want to go with ePMP or Rocket5AC gear. If I get get reliably 75Mbps inbound with ePMP on a 20 Mhz channel, that might be all I need. I am getting good speeds with UBNT gear most of the time but their gear is so tempermental, especially when it comes to firmware. I would love to take a Force180 and put it at all the spots we tested the UBNT gear at and do a real apples to apples test. I am also testing the new TwistPort horns against the traditional sectors.
We are working on a web based built in spectrum analyzer that does not take the radio offline. This is high on the list of requests from our customer base. In the meantime, please note that ePMP spectrum analyzer shows the noise floor relative to the channel bandwidth and what the SM sees. Other products show the noise floor at a sub carrier level which is native to wifi chipsets.
In terms of Force 110s, yes, they are a bit tight on the NA market but we are trying to assist with that situation. Last but not least, the Force 180s are a 15 degree azimuth beamwidth and has at least 3dBi gain improvement over the ePMP1000 which is sitting at 13 and a 30 degree azimuth. Your tests with UBNT might also be not quite apples to apples in terms of gain. Since you are a new customer trying to test ePMP, I could send you a sample of Force180 that you could try at one of your sites.
Sakid
I'd love to try out a couple Force 180's. I am needing to make a decision in the next few weeks if we want to go with ePMP or Rocket5AC gear. If I get get reliably 75Mbps inbound with ePMP on a 20 Mhz channel, that might be all I need. I am getting good speeds with UBNT gear most of the time but their gear is so tempermental, especially when it comes to firmware. I would love to take a Force180 and put it at all the spots we tested the UBNT gear at and do a real apples to apples test. I am also testing the new TwistPort horns against the traditional sectors.
How can I get a couple demos?
Send me your address offline to sakid.ahmed@cambiumnetworks.com please.
