Good morning guys, i’m new on mikrotik forum and i hope don’t write on the wrong section.
I want optimize my lhgg lte modem performace and i found the top configuration between the band aggregation.
I see that in the dl-modulation the lte modem use bpsk instead of better technology like 16qam or 64qam. i think that with this parameters bpsk is very poor.
i have routerboard v7.10.1 and modem updated.
other thing i just lock the band, but i want to lock primary band, is there any way?
Is the posted screenshot during active download or while modem was mostly idle? Modulation can change every TTI (which is 1ms in LTE) and can theoretically be different for each resource block (resource block occupies 180kHz of bandwidth) and definitely different for each CA frequency, so single screenshot can be very vague … If this is in (almost) idle state, then nothing to worry, mobile technologies tend to throtzle down to most robust modulation able to carry required amount of payload (on LTE that would be QPSK without MIMO).
ty for your time. When I took the screenshot it was mostly idle, so I started a test while downloading a program from sourceforge.net which has reliable servers, created a new monitoring session but saw the results were the same. bpsk covered the test 90% of the time, I rarely saw 16qam but only for a few seconds. Another thing is that the download was only at 500/600 kb/s, very poor as compared to the speedtest result (70-90 mbit/s).
p.s. I don’t have giga cap instead of my sim and I have company apn with public ip.
Are you saying that while using very same setup you can get pretty decent throughput numbers by running speedtest?
yes, in the meantime i download at 500kb/s , the speedtest result are 70-90 mbit/s
very strange 
One thing is certain: LTE radio as layer 1 in networking doesn’t know anything about L7 payload … so the difference you’re seeing is for sure not due to radio modulation.
More likely your MNO / ISP is playing funny games: priotitizing speedtest to make customers feel they are getting subscribed service but failing to provide it for actual connections … to hide the fact that their network is largely under-dimensioned for the number of subscribers…
You’re right. In this morning i try first to download tensorflow from github: https://github.com/tensorflow/tensorflow , it goes between 2,8/3,4 Mb/s . It use every time bpsk , so i create a speedtest right after the download. it gave me some real parameters.
And after that, if i’ll try another speedtest, you can see the little difference 
Today the primary band is the 1 and not the 7, i don’t know if change something but the sinr is bad.
I create a little log file when i download from github
log (2).txt (4.92 KB)
When using CA, primary band doesn’t matter much for DL, user payload will spread over all CA bands regardless (the only “exclusive” use is signalling between modem and network which always takes place on primary band). It only matters for UL as older CA-capable modems only did CA in DL and would always use primary band for UL.
Keep in mind that radio measurements are separate for each CA band and can be wildly different. I’m assuming values shown are for primary band, those for secondary bands are likely worse (modems tend to “camp” on bands with best radio signal, but MNO can affect selection via OTA parameters and it’s really hard to override it from modem side).