You are not logged in.
My greetings to this fine and respectable forum.
My name is Faye, I work at a wireless & sensor networks lab here in Greece, and I desperately want to find a way to keep our two WARPs V3 "phase-aligned" at all times.
So, after many sync-experimentation I have resolved that the most logical thing to do is to have my receiver node to clock-sync and trigger the transmitter node. I know it is the opposite compared to the sync examples on warplab website but it is logical, isn't it? I trigger the receiver to open its RFs, be ready to receive whatever will be transmitted, and then through the twisted pair cable it triggers the transmitter to wake up and send the message. This way I observed that the delay of the receiver picking up a signal is constantly 40 samples (40x25ns -> 1000ns -> 1ms). No more, no less. Everytime, the received signal starts at sample 41.
So no need to use xcorrelate since the delay at the receiver is the same all the time.
Each time I run the same experiment though, which includes re-initializing the warps while they are still on and sending a brief sinusoid, the phase offset is different.
IF i run the same experiment without clearing up the workspace nor re-initializing the nodes, no matter how many times I run it, I get almost exactly the same phase offset everytime.
But there is always phase offset. It is merely a matter of luck to get the waves of the sinusoid aligned between Tx and Rx after every new initialization. And I don't want to have an offset to handle at the start of every session of experimenting.
So, how am I supposed to tackle this obstacle?
If someone with experience would like to take some of their time and help me out please do so.
- Node 2 does indeed feed node 1's clock because if I turn on node 1 first it doesn't start up as usual but it gets configured only after I turn on node 2.
- Pin 8 (out_d0) of node 2 triggers pin 15 (in_d3) of node 1.
- I use RFAs of both nodes.
- The MMCX cables are 30 cm long.
- Twisted pair cable is 25 cm twisted as it is from jumper's edge to jumper's edge.
- Ethernet cables connecting them to switch are exactly the same but I lack any information about their precise length at the moment.
- Antenna extention cables (yes we use these to put the longest distance we can between antennas sometimes) are the same too, 5 meters each.
- Antennas are the exact same too, omni-directional, with 130cm extension cable each.
I have no idea if these info will help at all in helping with the phase offset issue but I trust that they will at least offer a more complete view of our WARPs assembly here.
Thank you in advance!
Offline
Each time I run the same experiment though, which includes re-initializing the warps while they are still on and sending a brief sinusoid, the phase offset is different.
IF i run the same experiment without clearing up the workspace nor re-initializing the nodes, no matter how many times I run it, I get almost exactly the same phase offset everytime.
This is the expected behavior. Each RF interface has its own PLL, inside the MAX2829 transceiver. The transceivers share a reference clock so their carrier frequencies will be the same. But each PLL will lock at a slightly different time, leading to an inter-carrier phase offset. This offset will remain fixed as long as the PLLs remain locked. The PLLs are reset whenever the MAX2829 is reset, which occurs in the init_nodes() process. To achieve a deterministic phase offset among transceivers, run init_nodes() once, then avoid any further reset of the RF interfaces.
Offline
Thank you thank you for this immediate reply.
So if I understand correctly, there is no way to set a specific phase offset at the start of an experimentation session. I need to run a simple experiment and see what is the offset, which means that I need to find a type of experiment to measure the phase offset as accurately as possible and then, without initializing the nodes again, to use that number to fix the phase of the rest of the experiments. Is this it or is there a better way?
Offline
So if I understand correctly, there is no way to set a specific phase offset at the start of an experimentation session. I need to run a simple experiment and see what is the offset, which means that I need to find a type of experiment to measure the phase offset as accurately as possible and then, without initializing the nodes again, to use that number to fix the phase of the rest of the experiments. Is this it or is there a better way?
Right- if you need to establish a specific inter-antenna phase you must estimate the per-PLL-lock phase offset explicitly then calibrate it away in your baseband signal.
Offline