Important update regarding LTE modem updates

[tuftec]

@robert-hh no. Band 3 is required for operation in Australia with Vodafone.

Eventually is a long time.

Thanks

[robert-hh]

@tuftec OK. Band 3 is accepted by the API already. The actual set is 3, 4, 12, 13, 20 and 28. But that only means, it's forwarded to the Sequans firmware.

[Xykon]

I believe I found the issue with the modem not attaching. It’s a problem in the MicroPython firmware. Unfortunately I have to travel today and I’m currently in the train on the way to the airport.

If anyone with capabilities to modify & build the firmware can try this: in esp32/mods/modlte.c
Find && (strlen(pos) >= 31) && pos[30] == '7') {

Replace with && (strlen(pos) >= 31) && (pos[30] == '7' || pos[30] == '9')) {

Let me know if this helps I’ll try to build new firmware packages at the airport and at least try to make them available for manual flashing.

I’ve only tested this with the latest NB-IoT firmware.

As for the start-up commands, it seems Sequans has adapted the NB-IoT firmware to use largely the same startup code as CAT-M1.

So you can use the attach method as shown here https://forum.pycom.io/topic/3476/orange-belgium-gpy/7 or you can do the attach manually by sending the necessary AT commands.

For this, in a possible correction to my earlier post (I’m not entirely sure if it makes a difference), please try and replace the following two lines:

lte.send_at_cmd('AT!="addscanfreq band=20 dl-earfcn=6300"')
lte.send_at_cmd('AT!="zsp0:npc 1"')

with these:

lte.send_at_cmd('AT!="RRC::addScanBand band=20'")
lte.send_at_cmd('AT!="disablelog 1"')

However this should only be necessary if your band is not currently implemented in the firmware.

[robert-hh]

@xykon OK. I added that, w/o a difference. I did not expect any change, because the response to AT+CEREG? is always either 2,0 or 2,2. I'm monitoring the link between the ESP32 and the Sequancs chip, and there are unsolicited messages of alternating +CEREG: 0and +CEREG: 2 every 30 seconds. Other people may be more successful, which after the lte.send_at_cmd("AT+CEREG?") command receive a string, starting with +CEREG: 2,1 or +CEREG: 2,5

[Xykon]

@robert-hh Thanks for trying... for everyone else I have uploaded test firmware packages for GPy and FiPy at [LINK NO LONGER AVAILABLE]

I'll be travelling for the next 16 hours without access to Internet.

[tuftec]

@robert-hh Not sure what I was thinking in my earlier reply.
For clarification - BAND 8 is needed for Vodafone Australia!!!!!
Sorry about that.

[tuftec]

@xykon thanks.
Look forward to receiving further instructions.

[Xykon]

Another thing to try is sending the command lte.send_at_cmd('AT+CEMODE=0') before calling the attach method or lte.send_at_cmd('AT+CFUN=1').

This will configure the modem to use EPS attach instead of combined EPS/IMSI attach.

[stsiafak]

@tuftec Yes you are right. I have tried this and I am getting the message you describe. Seems there is a bug in the libraries because the hardcore commands (addScanBand, addScanFreqRange, addScanFreq) return no error in case using the value of 8(?). Have you tried something more robust like:

send_at_cmd_pretty('AT!="addScanFreqRange band=8 dl-earfcn-min=3450 dl-earfcn-max=3799"')

[tuftec]

@stsiafak I have not tried the pretty command at this point. Everything else I have tried does not seem to work.
I have not got the time to just keep trying different suggestions.
We need a clear direction from Pycom/Sequans to get this working.

[Xykon]

In regards to band 8, I was informed now that, while the hardware supports all bands (1,2,3,4,5,8,12,13,18,19,20,28), the Sequans firmware currently only supports a maximum of 6 bands so we chose the ones that gave us the best support at the time of launch. I have a call with Sequans later today to discuss this limitation and provide a firmware that can be used with band 8 as soon as possible. I hope to have an update on that within 24 hours.

[tuftec]

@xykon thanks

[tuftec]

@xykon I have been advised that for future compatibility with Vodafone's Australian network, all devices should support bands 5, 8 & 28. Currently only band 8 is supported.
If there is only a limitation in the modem firmware for only 6 bands, then maybe there should be a special version for Australia that will provide support for Vodafone, Telstra & Optus/Singtel.

[robert-hh]

@xykon How do I have to understand the answer. I read, that the set of bands the firmware can support is 6, but which bands are in this set of 6 can be freely chosen. If that is true, then there should be not problem, unless in a certain region more than 6 bands are used simultaneously.

[mfalkvidd]

@robert-hh with 12 bands to choose from, covering all possible combinations of 6 bands would require pycom to make 665,280 different firmwares available (12 * 11 * 10 * 9 * 8 * 7). I think that would be troublesome.

[robert-hh]

@mfalkvidd That is not required. Since lte.attach w/o parameters only send AT commands like:
AT!="RRC::addScanBand band=20"
to the device for the 6 supported bands, an user could also manually send these commands, or would call lte.attach(band=xx). It's only important, that the Sequans firmware could deal with that. And obviously, the should be a link the the docs to a table, where users could find the frequencies of interest.

[kevin]

@robert-hh The problem is that the underlying modem firmware from Sequans (e.g., NB1-38729) only supports bands 3/4/12/13/20/28. Operations involving other bands silently fail at the modem level, with no effect and no error messages.
Pycom was aware of this from the beginning, but it is still not resolved:
https://forum.pycom.io/topic/2496/gpy-readiness/8
Not sure if it is a lack of support from Sequans, or concerns over certification, or a deliberate strategy to control roll-out with selected partners. In any case it goes against the principle of "open development" that motivated many of us to choose Pycom/Sequans.

[kevin]

@xykon Pycom was looking into a "test" firmware from Sequans, supporting all 12 bands:
https://forum.pycom.io/topic/2496/gpy-readiness/18
That was 8 months ago, I guess it didn't work out? Can you look into that option again? Even if not fully calibrated/certified, it would work for lab testing.

[robert-hh]

@kevin So it seems that the decision for Sequans was not a lucky choice.

[Redferne]

Just tried upgrading a FiPy to Stable v1.18.1.r1 and LTE firmware NB1-38729. Seems to attach fine using Telia NB-IoT, however it will not let me connect due to this:

>>> lte.send_at_cmd("at+cops?")
'\r\n+COPS: 0,2,"24001",9\r\n\r\nOK\r\n'
>>> lte.send_at_cmd("at+cereg?")
'\r\n+CEREG: 2,5,"03F4","01A7561C",9\r\n\r\nOK\r\n'
>>> lte.isattached()
False
>>> lte.connect()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
OSError: modem not attached

Sequans

+CEREG: <n>,<stat>[,[<tac>],[<ci>],[<AcT>[,<cause_type>,<reject_cause>]]]

Without digging in the code, is <stat> 5 properly handled as registered, roaming?