Hi all

I have setup a new link for one of our clients that need high throughput (approx 18Mbps in one direction). Here is my layout:

RB333(2 x R53-H) <===> Nstreme Dual <===> RB333(2 x R52-H)

I have set the band to 5Ghz-turbo and the frequencies are 220Mhz apart. We are also using new dual feed grids.

The strange thing that I found was that on TCP I was only getting about 6Mpbs via the Nstreme in one direction, but if I change the link to a normal AP bridge and station, I was getting up to 28Mbps.

Any help? Connection tracking is off and I have also disabled everything that I don't need.

After further testing I found that on UDP, I get approx 30Mbps in one direction while using Nstreme Dual.

If I switch it over to AP Bridge and Station(with Nstreme enabled), I get in excess of 60Mbps over the same distance in the same direction.

Any ideas anyone?

You may not be getting enough isolation on the dual feed grids. As a result, you may be saturating the receivers on the R52Hs when transmitting on the other polarity.

A check would be to temporarily move to separate antennas with a decent amount of vertical separation.

We have seen similar problems with dual-feed 4' dishes...

George

Ghmorris is right on that - probably not enough isolation with dual-pol antennas to use high power cards.
Try turning down the TX power, it may interfere less with the adjacent receiver.

Regards

how can you turn the power down? It is already on 0. Can you make it a negative number?

I am experiencing the same issue you are reporting with two rb333 running rc6.

With nstream1 everything look ok, but using nstream dual the performance are very poor.

I don't think it is an isolation issue, i use a dual pol mti at 25dbi gain and cm9s, becuase running two separe wireless interfaces on those antennas there are no problems.

regards
Ros

CM9's dont have such good near-adjacent channel filtering compared to more recent radios.
Try newer ones -

what do you advice?

Regards
Ros

Well if it is a inter-radio noise issue, why did MT make the RB333's slots so close? And in the same breath, why would they then make the R52-H? I chose the R52 range because they are optimized for MT and I needed all the throughput I could get.

Same exact problem for us on RC6...
Have not tried any other version with this yet.

SMA

i tested with compex 5423 cards but the result are identical to what i have with cm9.

As soon as possible i will test two separated antennas.

Regards

Please...

if the results are the same with a dual feed as with separated antenna's we will then know that it has to be something else. Hopefully if the problem is found it can be fixed in the next release of v3.0