A few initial EM7565 observations

Bjørn Mork bjorn at mork.no
Thu Dec 14 10:01:57 UTC 2017

This is based on a device running firmware 'SWI9X50C_01.00.02.00' and
is likely not applicable to any later firmware revisions...

1) shows up as USB version '2.00' instead of '2.10' when connected by

 Bus 001 Device 074: ID 1199:9091 Sierra Wireless, Inc. 
 Device Descriptor:
   bLength                18
   bDescriptorType         1
   bcdUSB               2.00

2) as noted by Sebastian: does not seem to aggregate more than 2
 cells. Both AT cli and QMI shows only 1 active or configured secondary
 cell, even when downloading at max capacity (or limited by USB2)

3) as noted earlier: LTE band numbers are off by one in the QMI CA
 responses.  This bug has been verified using the Sierra QMI SDK as
 well, and has been reported using their forum.

4) as noted by both Reinhard and Sebastian: USBCOMP defaults to
  '1,3,50D' which should enable both rmnet0 and rmnet1, but the
  firmware will show only rmnet0 (USB interface number 8).

5) switching to MBIM using '1,3,100D' appears to work initially, but has
 serious issues related to power management.  The USB device dies
 completely if the host (auto-)suspends the port.  It is currently
 necessary to disable autosuspend if you want to use MBIM

6) Reinhard noted that the NMEA port needs 'sendsetup'.

7) the unexpected USB version number means that the 'set DTR' quirk is
 disabled when connecting this modem by USB2. It does however still work,
 although I have a feeling there often are QMI timeouts when the QMI
 control channel has been inactive for a while.

Based on these observations, I am guessing that there are lots of
changes to the modem power management stuff (as expected), and that
there still are a few rough edges in that area.  Looking forward to the
next firmware update :-)


More information about the libqmi-devel mailing list