Hello. I’m using a RB911G-5HPacD and a Metal G-52SHPacn to establish a point-to-point connection inside a rotating machine and need to obtain the maximum bandwidth possible to send information from the RB911G-5HPacD (rotating) to the Metal G-52SHPacn (fixed)
The RB911G-5HPacD is connected to a radiant cable that goes all around the rotating part of the machine.
The Metal G-52SHPacn is connected to the antenna that comes inside the box and it’s placed at about 20 cm from the rotating radiant cable (therefore, when the machine rotates, the distance between radiant cable and fixed antenna doesn´t change)
When the machine is stationary, the bandwidth is about 160 Mbs (measured with winbox’s bandwidth test, TCP, SEND, from rotating to fixed device)
As soon as the machine starts to rotate, the bandwidth drops abruptly, to about 20 Mbs at 50 RPM and 2 Mbs at 100 RPM.
What can I do to improve the bandwidth as the machine rotates?
Fixed device configuration:
11 test1.rsc (1.07 KB)
Rotating device configuration:
13 test1.rsc (1.09 KB)
Thanks in advance.
I can think of self-inflicted interferrnce due to rapid movement of antenna (the rotating leaky cable) relative to stationary AP.
I’d expect a better behaviour if you used a simple dipole antenna, mounted in the center of rotation and directed across the axis of rotation … if the antenna was really a omnidirectional antenna with a nicely circular radiation pattern, then it would appear as stationary to the (stationary) AP.
But perhaps I’m not getting how your setup looks like. Can you post a photograph of your apparatus?
Hi mkx, thanks for the quick turnaround.
Please see the pictures in the attached file, I hope it’s clear.
I didn´t mention before but the signals levels (as reported by winbox on the registration tab) are between -40 and -50 db, maybe too high?
I cannot install an antenna at the center of rotation because objects go through the machine while it rotates.
Thanks!
Maybe the issue is connected to polarization?
Your radiant cable antenna should essentially be equivalent to a loop antenna:
https://en.wikipedia.org/wiki/Loop_antenna
and the polarization should depend on the position of the feed:
https://www.tutorialspoint.com/antenna_theory/antenna_theory_loop.htm
which in your case is rotating, and pretty fast at 50 or 100 RPM, the wave transmission in the center of the loop will be absorbed by the machine drum, no idea what happens on the outside in the same plane of the antenna.
While geometry of leaky feeder forming “antenna” definitely definitely makes things more interesting, it’s the observation of greatly reduced throughput when antenna starts to rotate that makes me think that rapid “antenna” revolution is making problems here.
I’m thinking: leaky feeder makes a “bent dipole” or “halo antenna”. But when rotor starts to rotate, it’s similar to having normal dipole quickly move in tangential direction … relative to reception antenna (which is placed orthogonally which makes polarization completely wrong but that’s another issue here). And that may mean quite large Doppler frequency shift. It should be possible to calculate the shift magnitude, but one needs to know the diameter of the rotating “thing” (the drum) … and RPM which was already mentioned.
Regarding Rx signal level: values between -40dBm and -50dBm should be fine.
Hello and thanks for the answers. Just one clarification: the fixed antenna is placed, not as shown on the picture, but parallel to the radiant cable. Anyway, I agree that it’s not the main issue because, when the machine is stationary, we see a good signal and bandwidth.
The perimeter of the drum (and also the lenght of the radiant cable) is about 4.5 m. Which, running at 100 RPM, gives a tangential speed of 7.5 m/sec that would be the relative speed between the fixed antenna and any point in the radiant cable, It’s about 27 Km/h, not very fast (I’m thinking comparing to a cel phone inside a car traveling at normal road speed)
But a car is not a panjandrum 
:
https://en.wikipedia.org/wiki/Panjandrum
The antenna on a car moves more or less on a linear path, maintaining the same orientation, i.e. at time n+1 it is parallel to itself at time n.
I have seen that there are (obviously paywalled) papers on the specific topic, so it is likely to be not easy-peasy to find a solution.
Thanks jaclaz. Could you point me to one of these papers?
Sure, though no idea if actually related/useful:
https://ieeexplore.ieee.org/abstract/document/7446823
https://www.academia.edu/95246126/Study_and_improvement_in_the_radio_communication_quality_for_rotating_electrical_machine
Hello. Thinking about something similar to the doppler effect… is there any possibility that the packet size the Mikrotiks are using has any influence in the bandwidth drop? Is it possible to reduce this value to a minimum so that I can test if it has any impact? What is the parameter I should change in the configuration?
Thanks in advance.