Hi all,
I was wondering if anybody here have used an ePMP 1000 NON-GPS connectorized radio together with a Lanbowan 34dBi Dual-Pol Dish Antenna for a PTP link?
I am thinking about using an ePMP 1000 connectorized radio with this 34 dBi dish from Lanbowan for a 25 KM PTP link?
Any suggestions and feedback will highly be appreciated.
Regards,
Sanay
Hi. Well, I've used LOTS of the Lanbowan dishes, although not speficially with the ePMP's for backhaul. We use lots of their DulPol wire grids in 2.4 and 5Ghz as SM antennas with the connectorised SM's, they are 22-24 dBm and are ideal for places where the Force200's don't quite have the gain you want.
As for the larger dishes, we've used them with AirFiberX Backhauls and with our older 802.11n gear, and they work just fine. So - I see no reason you wouldn't be pleased with that setup for an ePMP Backhaul. :)
Hi,
Thanks for your reply and feedback.
If the line of sight is clear and all other factors are normal, what kind of realistic throughput do you think we can acheive on this 25 KM PTP link?
Regards,
Sanay
Hi. I'd recommend using Cambium's LInkPlanner software. It'll do the math way better than I can. :)
http://www.cambiumnetworks.com/products/software-tools/linkplanner/
Ninedd, is there a us based dealer for their antennas? I want to try one of their 2.4ghz dishes…
@jakkwb wrote:
Ninedd, is there a us based dealer for their antennas? I want to try one of their 2.4ghz dishes...
I'm not sure if there is nor not. I'm in Canada, and we buy direct from them in China. We actually get them FedEx most of the time, and they are pretty affordable, so that even with the high shipping prices, they are still landed in-line with what you'd expect to pay. So - if you talk to Anne <anne (at) lanbowan.com> and mention me (Todd), she'll give you a good deal. :)
We use theiir 22 dBi DualPol Grids (http://www.lanbowan.com/contents/214/446.html) and they work pretty good for the price. They are about $40 to buy, and shipping will add another 50% probably, so they probably end up about $60 landed. So, 50% shipping on a small order feels like lots, but the way we look at it, if they were $60 with free shipping, we'd think that was an awesome deal. :)
And they make a 5 Ghz version of that same sized dish, which is about 28dBi then, and that's a pretty good higher gain SM antenna for the dollar too.
Dear All,
I have established a 25 KM LOS link using a pair of ePMP 1000 connectorized radios (NON-GPS) together with a pair of 34 dBi lanbowan Dish using a 20 MHz Channel Bandwidth.
The link has a approximate Downlink RSSI about -60 dBm and Downlink SNR of 35 dB.
With the Wireless Link Test tool, the maximum throughput I got was 54 Mbps Downlink and 27 Mbps Uplink.
Is the the normal throughput I should expect out of this current link?
What suggestions would you give to further fine tune this link to achieve more throughput?
Thannks,
Sanay
Seems low, assuming you're using ePTP mode and that there is no traffic on the link. Take a look at eAlign, is the RSSI on both chains about the same? What MCS are your DL and UL running at? (Monitor --> Performance, and see where the bulk of the frames are at.)
Hi,
Yes I am using ePTP mode and the RSSI values on the eAlign tools shows roughly about the same RSSI value of -60 to -61 dBm on both sides.
Regarding the MCS on DL and UL, I see the bulk of traffic (over 60%) on MCS 10 - QPSK 3/4, MCS 11 - 16-QAM 1/2, MCS 12 - 16-QAM 3/4 and MCS 13 - 64-QAM 2/3.
Thanks,
Sanay
It is likely a noisy path in that case. You might get better overall performance by using TDD mode, though you will gain a bit of latency.
You could also try ePMP 2000 radios at both sides, others have noted improvements in noise filtering with them. Other manufacturers radios in the same price range may or may not deal with noise better but may be worth testing in your environment.
I've had good luck with ePMP 1000 radios on a link of similar distance until noise degraded their performance beyond what I could accept. Radios from another vendor performed better in that environment otherwise I would have left the Cambium units in place.
Hi Jacob,
Thank you for your suggestions.
I will try TTD mode and check the results this weekend.
You think a Mimosa B5C will work better in this case?
Regards,
Sanay
It may or may not. This is may be one of those situations where you have to try a few approaches and settle on the one that works best.
Hi Jacob,
When you mentioned using TTD mode, are you refering to AP Driver Mode "TDD" or "TDD PTP"?
Thanks,
Sanay
Hello,
Since this is a PTP link, use "TDD PTP" mode in the AP and "TDD" mode in the SM. The only difference between "TDD" and TDD PTP" mode in the AP is that "TDD PTP" mode will only allow one SM to connect, either the first SM to complete registration successfully or the SM with the MAC address configured at the AP.
@Me Sanay wrote:Hi Jacob,
Thank you for your suggestions.
I will try TTD mode and check the results this weekend.
You think a Mimosa B5C will work better in this case?
Regards,
Sanay
in my experience, the epmp have done the best in noise. (for the money) of course sometimes a different radio will win.
if you are able, use the spectrum analyser and use a different channel.
you want the best possible modulation, with the lowest possible re-transmission rate. in the Ealign, you want to see both your receive chains matching or only off by 1 or 2 points. if they don't match up, you've got a bad alignment, path obstruction, or broken part.
at 60dbm, no reason your can't get your modulations to MCS15 if you can find a clear channel. only actually need - 68 when there isn't any noise to fight.
another thing to consider is the quality of the antenna. I don't know much about lanbow, they may have wonderful characteristics, i don't know on that. I know they are great for being so low cost, but if your FTB is 25 db, and you have a powerful noise source thats 30 degrees off of your center beam. or your center beam is 10 degrees wide, and you've got a lot of little noise sources within the spread, a better quality pair of antennas will work magic. when you are looking at backhauling, the more front to back the better, the narrower and smaller the beam the better. I'd compare the antenna radiation graph against one from radiowaves, or a lower cost unit like UBNT to get a feel for the characteristics.
in a noisy environment, the antenna can make or break the link.
@Me Sanay wrote:Hi,
Yes I am using ePTP mode and the RSSI values on the eAlign tools shows roughly about the same RSSI value of -60 to -61 dBm on both sides.
Regarding the MCS on DL and UL, I see the bulk of traffic (over 60%) on MCS 10 - QPSK 3/4, MCS 11 - 16-QAM 1/2, MCS 12 - 16-QAM 3/4 and MCS 13 - 64-QAM 2/3.
Thanks,
Sanay
But what's the RSSI of each chain (ie. ch0 and ch1)? Are they the same?
Dear all,
I have realigned the antennas and now we are getting roughly around -61 RSSI at both sides using a 20 MHz channel.
This is a very noisy environment where we have another service provider with several of their APs (5 GHz) installed in the same tower as us.
Also there are 4 other service providers (5 GHz) on a nearby tower about 100 meters from us with their own devices installed.
I have already chosen the best frequency I can find at the moment.
Using the Wireless Test Link, I am now getting around 75/75 mbps (Downlink / Uplink).
Doing some TCP test, I achieved an average about 40/45 mbps (Downlink /Uplink) with peaks about 50/60 mbps.
Why am I getting 25% less throughput than what is being reported in the Wireless Test Link tool?
Now, if I switch to a 40 MHz channel, Wireless Test Link tool averages 90/90 mbps and in a few cases peaked at 170/105 mbps.
Doing some TCP test, I achieved an average about 50/60 mbps (Downlink/Uplink) on the 40 Mhz channel with peaks about 65/75 mbps.
Again, the actual test is some 25% less than what is reported by the Wireless Test Link tool?
However, I observed that using a 40 MHz channel in this environment is susceptible with the connection link getting dropped every 5-10 minutes with the RSSI dropping from -62 dBm to -90 dBm every minute or so.
Will it really be possible in reality to achieve a 75/75 mbps throughput on a 20 MHz channel as reported by the Wireless Link Test tool once I start connecting users behind this backhaul at it’s current state?
Also will replacing the ePMP 1000 radio with an ePMP 2000 radio help?
Any suggestions will be appreciated.
Some Information for the current 20 MHz channel link:
Downlink Packets Per MCS :-
Uplink Packets Per MCS:
AP Side Current RSSI: - 59 dBm (Ch0: -62 dBm, Ch1: -62 dBm)
SM Side Current RSSI: -61 dBm (Ch0: -66 dBm, Ch1: -63 dBm)
Wireless Link Test:
Downlink: 77.997 Mbps
Uplink: 74.785 Mbps
SM Distance: 25.482 KM
RSSI (dBm) Downlink/Uplink: -57/-61
SNR (dB) Downlink/Uplink: 33/34
MCS Downlink/Uplink: 12/14
Downlink Quality: 100%
Downlink Capacity: 80%
Thanks,
Sanay
@Me Sanay wrote:Also will replacing the ePMP 1000 radio with an ePMP 2000 radio help?
The ePMP2000 has several design advantages over the ePMP1000. In a PTP situation, what people are seeing quite substantial improvements from is the ePMP2000's filter, which is designed to help eliminate interference from interferers on adjacent channels.
So, if you're using a 40 Mhz wide channel, and you have an un-synchronized uncoordinated interferer - the normal recommendation would be that from the bottom or top of your channel edge, that their closest edge be 80Mhz away. That's ''twice the guard band as your carrier''. Of course, that depends on a number of other factors - if their antenna is 2' away and if you hear them at a -30 signal, then that's more of an impact than if they are 100' away and well shielded and you hear them at a -80. But, generally speaking - the rules are ''twice the guard band as your carrier'' on non-synchronized interferers.
However, with the ePMP2000 there is a filter which makes nice sharp edges on your channel use - so it'll eliminate interference from signals that are strong and near your channel. If they are right on top of you (overlapping on the same frequencies) then it might not accomplish much, but if they are adjacent to you and too close and loud, then the ePMP2000's filter can dramatically help.
Additionally, that filter is 2 way - so it'll also help you not interfere with them in the same scenario, where they are close to your frequency and physically close (loud) to you. Your signal will now be sharper drop off at the edge of your broadcast frequencies too - so that'll make you not interfere with others that are using channels near to your chosen frequencies to. That may allow them to use a 20 Mhz channel rather than a 40 Mhz channel and get their desired throughput to - further allowing both of you to reduce interference, and to up your spectral efficiency.
@Me Sanay wrote:Dear all,
I have realigned the antennas and now we are getting roughly around -61 RSSI at both sides using a 20 MHz channel.
This is a very noisy environment where we have another service provider with several of their APs (5 GHz) installed in the same tower as us.
Also there are 4 other service providers (5 GHz) on a nearby tower about 100 meters from us with their own devices installed.
I have already chosen the best frequency I can find at the moment.
Using the Wireless Test Link, I am now getting around 75/75 mbps (Downlink / Uplink).
Doing some TCP test, I achieved an average about 40/45 mbps (Downlink /Uplink) with peaks about 50/60 mbps.
Why am I getting 25% less throughput than what is being reported in the Wireless Test Link tool?
Now, if I switch to a 40 MHz channel, Wireless Test Link tool averages 90/90 mbps and in a few cases peaked at 170/105 mbps.
Doing some TCP test, I achieved an average about 50/60 mbps (Downlink/Uplink) on the 40 Mhz channel with peaks about 65/75 mbps.
Again, the actual test is some 25% less than what is reported by the Wireless Test Link tool?
However, I observed that using a 40 MHz channel in this environment is susceptible with the connection link getting dropped every 5-10 minutes with the RSSI dropping from -62 dBm to -90 dBm every minute or so.
Will it really be possible in reality to achieve a 75/75 mbps throughput on a 20 MHz channel as reported by the Wireless Link Test tool once I start connecting users behind this backhaul at it’s current state?
Also will replacing the ePMP 1000 radio with an ePMP 2000 radio help?
Any suggestions will be appreciated.
Some Information for the current 20 MHz channel link:
Downlink Packets Per MCS :-
MCS 14 - 64-QAM 3/4 (65%)MCS 13 - 64-QAM 2/3 (29.9%)Uplink Packets Per MCS:
MCS 14 - 64-QAM 3/4 (31%)MCS 13 - 64-QAM 2/3 (57.2%)AP Side Current RSSI: - 59 dBm (Ch0: -62 dBm, Ch1: -62 dBm)
SM Side Current RSSI: -61 dBm (Ch0: -66 dBm, Ch1: -63 dBm)
Wireless Link Test:
Downlink: 77.997 Mbps
Uplink: 74.785 Mbps
SM Distance: 25.482 KM
RSSI (dBm) Downlink/Uplink: -57/-61
SNR (dB) Downlink/Uplink: 33/34
MCS Downlink/Uplink: 12/14
Downlink Quality: 100%
Downlink Capacity: 80%
Thanks,
Sanay
retransmitted packets will have a negative effect on your TCP performance if you block your MCS state to 13, you'll see somewhat of an improvement in TCP, and a minor drop in OTA link test.
monitor > performance will give you an idea of your retransmitted packets. if your over 10%, for it down a modulation to get gain overall throughput.
the default cambium method is usually the best, but is very bursty sudden noise changes setting your max modulation to your most carried modulation (your case 13) you'll see an overall improvement in the TCP area.
I 2nd the epmp2000 suggestion. they will get rid of your co-band noise. all you need to worry about is on channel noise with those units.
I had a unit that wouldn't come up from MCS12 because of co-band, after changing it to 2k, it runs at MCS15.
If your radio is suffering its code state from ONLY CO-BAND you'll see a similar improvement, if CO-channel is your problem, you won't see much improvement.
if is perfectly possible to get 75 mpbs on 20mhz.
Dear All,
I have now got the best performance using ePTP Master/Slave using a 20 MHz channel.
I am now testing a downlink capacity of 70 mbps which I think could be further improved with more accurate alignment.
This will work fine for me for the next 6 months. After that, I will probably have to replace the ePMP 1000 with either the ePMP 2000 or the Mimosa B5C.
Thank you for your suggestions and help.
Best regards,
Sanay
