Project

General

Profile

Actions

Feature #5698

closed

ms-sdr needs a test setup

Added by Hoernchen 4 months ago. Updated 3 months ago.

Status:
Resolved
Priority:
High
Assignee:
Category:
-
Target version:
-
Start date:
10/10/2022
Due date:
% Done:

100%

Resolution:
fixed
Spec Reference:
Tags:

Description

As originally discussed quite some time ago the ms-sdr needs a test setup:

  • one raspberry pi 4 that as far as I can remember should have arrived at some point last July, running https://ubuntu.com/download/raspberry-pi/thank-you?version=22.04.1&architecture=server-arm64+raspi - this is supposed to mirror my setup here
    • this is the ms side, so just setting it up for ssh access is all that needs to be done, I'll finish the setup
    • this side should be connected to the bladerf xa4/xa5 that is already at the office
  • one apu or some other platform that can do gsm+gprs ( no edge) @ 4sps
    • this should run our basic 1 arfcn gsm/gprs cni setup, no special config required, also ready for ssh/vpn access
    • this side should be connected to a b2x0 , see SYS#5964 for that, might arrive at the office tomorrow.
  • one side is the ms, the other is the network, so they neeed to be cross connected (rx->tx) with 30db of attenuation in each path
  • both devices need a 10mhz gps refclock, careful here: the proper u.fl connector on the bladerf that has to be used is marked REFIN

Checklist

  • fix Rx/Tx label on B210 face plate
  • move the setup to the usual racks
Actions #1

Updated by roh 4 months ago

i guess it makes sense to mount all in BGT carriers, since it needs to go in the rack for getting clocks.

so we need each one time:
Actions #2

Updated by roh 4 months ago

found:
  • rpi4-4g, and a metal cooling case (joy it 'block')
  • bladeRF 2.0 xA4 in acryllic case
Actions #3

Updated by laforge 4 months ago

Hi @jsteiger,

excellent idea about the BGT.

On Mon, Oct 10, 2022 at 03:17:42PM +0000, roh wrote:

I have swapped the B210 that gets to the office with my BGT mounted prototype.

That fully assembled unit will also arrive at the office tomorrow.

Please send the non-BGT RPI4 from AM93 to me in return.

  • some case for the bladerf (sadly there seem to be only useless acryllic ones from the vendor

You could just use a

  • some pc (apu are in extremely short supply)

yes and no. We can still order them at (for us) reasonable pricing with some lead time.

Actions #4

Updated by roh 4 months ago

i put the assembled (passive cooling case) rpi4 onto maikes desk with a post it '-> harald'

Actions #5

Updated by roh 4 months ago

  • % Done changed from 0 to 50

not sure which nutjob is maintaining ubuntu nowadays, but i had to remove some packages from this 'supposed server' image to make it not shitty and breaking for us:
purge snap, unattended-upgrades, cloud*, netplan.io, ubuntu-advantage-tools, open-vm-tools and everything which depends on em.

then one can use 'normal' network config again, and not some may-work-sometimes-when-the-mac-is-even-and-its-thursday crud.
i went for /etc/systemd/network/10-eth0.network

no clue why it still uses eth0.

please lets use maintainable and not broken-by-default images as base in the future.

ssh password auth is disabled (not default on ubuntu!)
keys for hoernchen,laforge,osmith are installed for user ubuntu (has sudo)

the apu was no issue. user is root, keys like above.

the plan is to put it in the top end of the testrack on ports 19 and 20 of the upper switch (testrack1)
for now i have placed it in the lab on a desk for ease of debugging.

  • 10.9.25.150 mssdr-bts (apu)
  • 10.9.25.151 mssdr-ms (rpi4)

i have put a 6V psu on the usrp, but i'm not sure it needs it. for me they were flaykey sometimes when only powered by usb. no clue what bladerf needs or does, since it seems to depend on the order of powerup of devices?! please advice what worked for you.

clock is just a simple splitter, when its in the other rack its the proper 10mhz distribution amp.

please log in and give it a try.

for rf its only simple flex and 30db attentuators between RX1 and TX1 of both devices. if you'd like a rfdsatt here, give me a shout.

Actions #6

Updated by roh 4 months ago

i have fixed the networking issue with the apu and the power-rail-thinko about the gpsdo - my fault.

i've also added the video-amp as 10mhz distrib.

the clock seems rectangle on input and sinusoidal on output. levels is ~0.7Vpp - should be fine, but eric seems to have issues.

Actions #7

Updated by roh 4 months ago

removed the amp again - back to simple T as that one is working for eric in his current setup.

Actions #8

Updated by Hoernchen 4 months ago

Blade still can't lock, which is weird, brought a working tested blade that works for me to the office and gave it a try, still fails...

Actions #9

Updated by Hoernchen 4 months ago

Brought my whole setup to the office only to discover that it now locks immediately, I am not aware of any changes that would explain why this is now the case, but I guess it's fine as long as the issue does not show up again.

Actions #10

Updated by laforge 4 months ago

  • Checklist item fix Rx/Tx label on B210 face plate added
  • Status changed from New to In Progress

After some investigation and remote debugging for Hoernchen, I discovered that roh had labelled the Rx/Tx of the USRP wrong. The label for Port A/1 were (and still is) swapped. So you ended up connecting Tx to Tx and Rx to Rx.

Before figuring that out, I already had tried to
  • swap the RG178 cables with rigid-flex RG402
  • remove the 30dB attenuators
  • measure the signal levels with an RfExplorer
    • the RfExplorer showed the carriers something like 400-500 kHz above the expected frequency. Is it off that much?

I re-introduced the 30dB attenuators and kept the rigid-flex connections.

roh, please make sure to swap the label to propely indicate Rx/Tx. Next time I would hope something trivial like that doesn't need my intervention...

Actions #11

Updated by Hoernchen 4 months ago

  • % Done changed from 50 to 80

The setup finally appears to work as expected, after stumbling into https://osmocom.org/issues/4832 due to my copied stp config not working for some reason until I just dumped the actual running config instead of relying upon the dynamic auto setup, there was "no" msc available, and this led to opening and then immediately closing the dl channel during LU attempts, which caused tons of bit errors and obviously looked like additional rf issues in the DL..

roh feel free to move the test setup to a more permanent location - just mention it here first, so I don't end up wondering if it crashed..

Actions #12

Updated by Hoernchen 4 months ago

  • Checklist item move the setup to the usual racks added
Actions #13

Updated by roh 4 months ago

  • Checklist item fix Rx/Tx label on B210 face plate set to Done
  • Checklist item move the setup to the usual racks set to Done
Actions #14

Updated by roh 4 months ago

  • Status changed from In Progress to Resolved
  • % Done changed from 80 to 100
  • Resolution set to fixed

setup is moved to the rack.

i think both sdr are happy, but please give it a thorough checkover - had to swap some cables over due to length constraints.

it now lives on the right side of the BGT frame, the apu is a bit lower in the rack, there was simply no other space.

the labels are fixed. it seems i assumed the same layout for both channels when printing those, not a 'mechanical mirrored one' - my bad.

Actions #15

Updated by laforge 3 months ago

  • Tags set to ARDC
Actions

Also available in: Atom PDF

Add picture from clipboard (Maximum size: 48.8 MB)