OAI and LimeSDR-Mini

@dchard Thanks for all those informations, what is LimeSuite build commit that you used ? please can you share enb band 3 rf conf file ?!

what kind of OCXO did you use!! any reference or any link for buying it please?

Hey there!
Following @bastienbaranoff tutorial, I get most of the time the same error as many of you -> Failed to find ue context associated with eNB ue s1a I did managed to connect my phone (samsung galaxy s5) but I never got a stable internet connection.
Like @dchard said,

The reference signal power and TX/RX gains will need to be fine tuned for sure, as none did that for the Mini yet.

Nevertheless, @bastienbaranoff managed to find 127 for the tx_gain and 160 for rx_gain which seems to work in his case. Any idea where those 2 values comes from?
@bastienbaranoff, do you have any link to share ? how did you figure out it was the maximal values?

Thank you very much

1 Like

@Niels So you solved UE attach problem by changing the phone and using Samsung Galaxy S4 right ?

yes, the attachement is possible with samsung s5
and alcatel pop but with samsung s6 , i can’t see the network even! you
could download even you have failed to find ue context assiciated with
eNB. It’s not an error but warning, https://github.com/chunyeow/openairinterface5g/blob/master/openair-cn/S1AP/s1ap_eNB_context_management_procedures.c . Like @dchard said, it due to some many reasons :
quality of radio bad (due to misconfiguration of tx and rx gain because not yet tuned with vna or enodeB analyzer)when you use limesdrmini longtime with oai, the chip need cooling
technic, it’s why the UE could connect and download or upload file but
reject by network after some minute! it’s also due to of the bad quality
of modulation if the ci is hot (temperature)the oscillator stabilization ! using ocxo like @dhcard said , i don’t know more information about it!

I use a galaxy s5. And yes I get to the phone attached once is a while but it is really unstable and the connection stop after few seconds. Never managed to get a proper internet connection.

like he said, he use the maximum configuration of tx and rx!! i’m not testing yet, how about if we report the limesrd configuration for this tx and rx?

I see, thanks for the info! So if I cannot reach internet, it must be from somewhere else. I’ll keep digging.

after some minutes , did the enodeB reject your UE!

I try to answer all of the questions:

@Bourascom

I used the prebuilt packages on Ubuntu 18.04.1 LTS 64bits. Did not compiled it myself.
Will share the config file a bit later.

@dast

Not used any yet, the picture I shared is without an OCXO. But I can also see a huge freq error. Will use my 10MHz Trimble OCXO which I believe outputs a sinusoid frequency. For testing I will also use my vector analyzers internal OCXO, so the system is freq. and phase locked for the measurement.

No, I did not get rejected, the phone can stay connected for a long time, but if I push traffic, disconnects can occur.

Will get back with more info when I have a bit of time.

i have amplifier for band8 , if you have also config for it! it’s will be help me! i would like to ask you if you have error “ailed to find ue context associated with eNB ue s1a” on connecting too the nodeB?

Will take some time, as my primary interest is in band3, band7 and band 42.

The important thing here is that every band and every bandwidth(!) configuration needs its TX/RX calibration along with proper ref sig power configuration. If this is not the case, then all kind of weird stability and performance errors will arise.

The RX is not that critical as power control can instruct the UE to increase/decrease its power, but still the TX + ref sig needs to be precise. Imagine that the if the phone is incorrectly detecting the reference signals, it can lead to incorrect channel estimation, incorrect meas. reports which leads to incorrect scheduling and eventually bad performance.

Thank for you clear explanation, so the configuration should be for
every band for every bandwith or for every frequency ref for every
bandwith! that will be more work! for example for band 3, did the
2680000Hz and 2685000 hz not have the same configuration rx and tx?

Not for “every” frequency, but at least every band + bandwidth (5,10,15,20MHz). Within a band there are no such variations, that would require a calibration for each possible frequency.

Example: you have a calibration for 1820MHz center with a 5MHz bandwidth. If you want to move to lets say 1860MHz with the same 5MHz bandwidth, than you dont need new cal data. In other words you can use the same vTX, RX gains and same ref sig power.

@dchard Did you already manage to get a proper configuration for Tx, Rx and sig power for the band 7 for instance?

Not yet, there is an amp failed in my analyzer, so I need to fix that first. Not too bad, but it measures the amplitude 14dB less than it should be.

what kind of analyzer did you use !

For configuring with vector signal analyzer, what kind of parmaters should be changed?is it on the enb.banbxx.tm1.25PRB.lmssdr.conf ? or all the .ini file?

It will be in the enodeb conf file, but obviously for each band and bandwidth a new example file will need to be created. We can also measure the output purity with different TX gain values to see if the lime mini distorts at max TX gain or not. We can also measure frequency error with and without an OCXO to see if the mini can meet the advertised specs.

I am using an Agilent 89600S vector signal analyzer.

I compare for example this two configuration file : https://gitlab.eurecom.fr/oai/openairinterface5g/blob/cd8cc1a35c8d07b2782ef79865b12200e75dde0b/targets/PROJECTS/GENERIC-LTE-EPC/CONF/enb.band7.tm1.rrh.lmssdr.conf and https://gitlab.eurecom.fr/oai/openairinterface5g/blob/cd8cc1a35c8d07b2782ef79865b12200e75dde0b/targets/PROJECTS/GENERIC-LTE-EPC/CONF/enb.band13.tm1.lmssdr.conf . The parameter changing are : rx, tx,pdsch_referenceSignalPower , pusch_p0_Nominal. May be the pusch_p0_Nominal is also a parameter that should be consider to other band?