Link Capacity Test vs Speed Test-

I have a 450i set up in a lab environment. (1 AP and 1SM) I do realize that the lab environment is not the ideal place to test (signals bouncing off the walls, the  ground and my head .... I swear I can feel it power up).

My question is... as dirty as the set up is, how should the speeds shown on the Link Capacity Test compair to the speeds shown on end user speed tests? 

The LCT shows 95down and 17 up (Efficiency is 97%dnown and 93%up)

But Speedtests (ookla, ATT, getmyspeed) all show right around 45/14. 

AP set to

20MHz channel

SM target level -65db

Xmit power 0 dBm 

75%/25% split 

SM Rec Level is -54dBm

Good S/R ratio

AP & SM are 8-10 feet off the ground (I did move them outside the office. About 100ft apart)

Link Quality

Down 25% on all Modulations

Up 33% on all but 256....256 is <1% 

Are you running the speed test with the SM pulling the IP or is the SM bridged to a CPE? Also how are you handling WAN IP assignment, DHCP, PPPoE, Static, any NATing?

SM is bridged. Using PPPoE. DHCP. 

For this test Im using broadband connection strait out of laptop. In the field I have the customers router do the PPPoE challenge.

No NATing.

Some similar reports on this post here:

The speeds you see are spot on with what we see.  Nothing wrong with your setup.

to my knowledge the current SM's cap at like 60mbit of actual throughput on the SM side, the AP side should be able to handle more, have you tried 2 sm's with 1 ap and seen how far you can push the total throughput of the ap? or 3 sm's with 3 laptops.

I was told by someone from cambium that they had fixed the issue that was restricting the sm's to 60mbit in 14.x i think, but to my knowledge i've never seen above that on a single SM, for actual TCP throughput.

Would you be able to run the same tests with everything the same except PPPoE is disabled and post the results here?

 PPPoE is disabled in the SM for the tests. The customer router handled PPPoE challenge. I did run the test with straight bridge (no PPPoE) and got the same results.

On a "loaded" AP the differences are greater.  Due to network load. But I expected to see the numbers be the same with no other traffic on the AP.

I have seen this for years.  SMs will never get more than around 55mb of real throughput.  Real world and lab testing I can never achieve more than this on a signle sm.  Obviusly I can max the AP out in the lab with multiple SMs and it works as expected.  I have been trying to get cambium to just admit this in a public setting, but I can never get them to.  All testing is done with no NAT or PPOE, just bridge mode.  If I am doing something wrong I would love to know.  I assume this will not be an issue when the new SMs come to market.