I've solved the problem by setting the voxl-imu-server.conf to
"imu0_enable": true,
"imu0_sample_rate_hz": 200,
"imu0_lp_cutoff_freq_hz": 92,
"imu0_rotate_common_frame": true,
"imu0_fifo_poll_rate_hz": 200,
...
Tested on 2 different VOXL2 boards, this reduces the number of lost messages sequences from 60% to <0.5% of all messages. Information on why this is the case would be interesting, as it is stated here that the default config is optimized for VIO.
@Moderator Hi! I'm using :
system-image: 1.5.5-M0054-14.1a-perf
kernel: #1 SMP PREEMPT Fri Mar 24 00:48:55 UTC 2023 4.19.125
------------------------------------------------------------------------------
hw version: M0054
------------------------------------------------------------------------------
voxl-suite: 0.9.5
------------------------------------------------------------------------------
Packages:
Repo: http://voxl-packages.modalai.com/ ./dists/qrb5165/sdk-0.9/binary-arm64/
Last Updated: 2023-08-24 23:20:47
When experiencing the message dropout, this is the output of voxl-inspect-cpu :
Hi! I'm working with a ModalAI Sentinel right now equipped with the VOXL2, and I've noticed after running a few test flights that the recorded imu_apps data is very patchy, with bursts of messages followed by entire seconds of silence. This puts alot of stress on my VIO. Am I missing something or is this a known issue, and if so how do I fix it?
my voxl-imu-server.conf params are the default.
Thank you in advance!
Hi! Is any dev available to help out with this?
Hi! I'm passing an external estimator input (VICON or custom kalman filter) to the (Sentinel's) VOXL2's EKF2 over a MAVROS topic, but I'm not able to get the /mavros/local_position/pose to match the vicon pose.
I use launch the mavros and remap the VICON Pose to the following topics with the frame_id = 'map'.
If I open rviz I get the and look the the two poses :
I see that first of all the VICON pose does not get accepted by the Flight Controller (EKF2) and also that even without feeding the PX4 with the VICON pose the yaw of the Sentinel (mavros/local_position/pose) is off.
I suspect one of the following to be the problem :
Any help would be much appreciated, thank you!
@tanguy-sagere nvm was a permission issue, thanks!
@Eric-Katzfey Hi, well now the voxl2 doesnt even show up on adb devices anymore (so I cant adb shell). However, I'm still able to connect over wifi (softap), so the voxl is working ... I've tried multiple usb cables too. After resetting the .bashrc to the default, I'am able to access shell via ssh however (as im still able to connect over wifi). the adb issue is persistent over multiple computers too
Hi!
I have a sentinel with a voxl2 that had been working fine since it was recently delivered. However, after connecting the voxl to my wifi and ssh'ing to it, whenever doing "adb shell" I am brought to "root@m0054" (adb root) rather than "voxl2: $" (adb shell).
Assistance would be welcome!
Best,
Tanguy
@modaltb Hey! Thank you for the reply ahaha, I understand its a busy quarter for everyone 
We are connecting the IMU to the I2C3 pins of the J9 6-pin pinout of the 5G Modem. On other devices (such as the jetson AGX orin) we are able to get a data rate of 928 hz , but on the VOXL its stuck at around 628 Hz in ROS. For the ORIN we were able to increase the rate by modifying a bus_clk_rate file in /sys/bus/i2c/devices/i2c-1 but there appears to be no equivalent here. I found a voxl_i2c_write function in your docs that looks like it could increase I2C bus rate, but unsure how to access and use it. Also tried changing the device tree file to increase the rate ceiling and making a .patch to change it and add changes to kernel_patches but that dropped the rate to 120-ish, and never increased it. Quite confusing! Thanks for your time and have a nice day!
Hi! Anyone have an idea how to fix this?
Hello! I'd like to permanently change the I2C frequency of the data from my lsm9ds1 IMU to a higher setting (currently reads at around 638 Hz on the VOXL2 using rostopic hz /imu_lsm9ds1/imu , but I find no bus_clk_rate file in /sys/bus/i2c/devices/i2c-1 (nor in the other devices for that matter) to open and write the desired frequency to (id like it as close to 1000Hz as possible).
Is doing so possible; and if so, is there a file to be edited or command to be used for this?
Thanks!
Yep that's what I was seeing! Ok I'm using a custom OS here with a debug kernel already included so flashing that again solved the issue. I'll make sure to have everything disconnected when I reboot it from now on. Is there any other way to prevent this from happening in the future? I mean we`d like to have the VOXL2 on a drone with all cables attached and be able to just toggle the power supply without having to dettach and reattach everything everytime. Thanks!
@tanguy-sagere said in "Error battery voltage" on VOXL2:
For the sake of completeness, I flashed the modalai platform voxl2_platform_1.3.1-0.8.tar.gz on the device with the debug board disconnected (direct connection with USB-C). The flashing process completed successfully with install.sh but ;
The device appears in dmesg as "qcom Kona" which is correct.
sudo adb shell does not work.
sudo adb devices lists no devices.
I still see no output from the debug board serial debug monitor.
ethtool still says "no link detected" when connecting Ethernet to the debug board.
Quick recap :
Im starting to think that the debug board failed mechanically for whatever reason. Is there a software fix for this or should I contact support personally to have the debug board replaced as per warranty?
For information : I am using the "5G Model Carrier Board, USB HUB, I/O Breakout for VOXL 2 Carrier Board"
I followed the video you recommended, forcing into fastboot with USB-C disconnected at first and then flashing system image. Flashing goes well :
=====================================================
| Done installing voxl-suite for QRB5165 |
...
[INFO] Done installing voxl-suite
----Installation Completed----
But then when I restart the VOXL2 in normal boot and screen into it using the debug board serial debug monitor to observe the booting process; this is all that appears :
Decompressing kernel image done: 3156 ms
Dtbo hdr magic mismatch 0, with D7B7AB1E
Best match DTB tags 455/0000000B/0x00000003/20001/3001E/20020/20027/20021/(offset)0x9A781B37/(size)0x0008B3AB
Using pmic info 0x3001E/0x20020/0x20027/0x20021 for device 0x3001E/0x20020/0x20027/0x20021
Memory Base Address: 0x80000000
Cmdline: noinitrd earlycon=msm_geni_serial,0xa90000 androidboot.hardware=qcom androidboot.console=ttyMSM0 androidboot.memcg=1 lpm_levels.sleep_disabled=1 video=vfb:640x400,bpp=32,memsize=3072000 msm_rtb.filter=0x237 service_locator.enable=1 androidboot.u
RAM Partitions
Add Base: 0x0000000080000000 Available Length: 0x000000003B800000
Add Base: 0x0000000180000000 Available Length: 0x0000000100000000
Add Base: 0x00000000C0000000 Available Length: 0x00000000C0000000
ERROR: Could not get splash memory region node
kaslr-Seed is added to chosen node
Shutting Down UEFI Boot Services: 3227 ms
Start EBS [ 3228]
BDS: LogFs sync skipped, Unsupported
App Log Flush : 34 ms
Exit EBS [ 3280] UEFI End
Is there a hardware defect with the VOXL2 or with the Debug board? Was this also part of the bug you encountered?
Thanks for your time and help
For the sake of completeness, I flashed the modalai platform voxl2_platform_1.3.1-0.8.tar.gz on the device with the debug board disconnected (direct connection with USB-C). The flashing process completed successfully with install.sh but ;
I should also mention that I have been working with an ESD-bracelet and at no point was the VOXL2 subjected to contact with a conductive surface other than my hand, nor dropped or otherwise physically threatened.
Hi! I was setting up internet access to the VOXL2 (rather than sharing my laptop internet with it), working in root after ssh'ing into it from my laptop via my laptop Ethernet-> USB-A to JST adapter -> debug board USB2 Host-> VOXL2 when I suddenly couldnt connect anymore to the VOXL2 from my network manager (and so ssh wouldnt work).
As that seemed weird, I rebooted the chip into normal mode and screened into it using the debug console (laptop connected to debug board with micro USB) but nothing would appear on screen. So I tried the same thing in fastboot mode; and the following happened (and keeps happening after changing cables, power socket, etc...) :
This shortly pops up when screening the chip after fastbooting it
Handling Cmd: getvar:version Error battery voltage: 0 Requireed voltage: 3600, can flash: 0
Then this loops continuously :
LmhIsenseInit Post GPU TIME 2ms
Entering in func1 of NPU TIME 2ms NPU IDDQ 41 LmhIsenseSubSysEntry Post SubSysEntryCb TIME 0ms
LmhIsenseSubSysEntry Post
LmhIsenseSubSysInit TIME 0ms
LmhIsenseInitSubSys Post SubSysEntry TIME 0ms
LmhIsenseInitSubSys Post SubSysTrim TIME 0ms
LmhIsenseInitSubSys Post SubSysExit TIME 0ms
LmhIsenseInit Post NPU TIME 2ms
isense TIME 5ms ISENSE TOTAL TIME 5ms
Disp init wait [ 1670]
-----------------------------
Platform Init [ 1687] BDS
UEFI Ver : 5.0.220324.BOOT.XF.3.2-00354-SM8250-
1
Platform : QRD
Chip Name : QRB5165
Chip Ver : 2.1
Chip Serial Number : 0x5A93EBEB
Silver cluster Core 0 Freq: 1516 MHz
-----------------------------
UEFI Total : 524 ms
POST Time [ 1717] OS Loader
Loader Build Info: Feb 18 2022 17:53:29
VB: Non-secure device: Security State: (0xF7F)
VB: RWDeviceState: Succeed using devinfo!
KeyPress:2, BootReason:0
Fastboot=1, Recovery:0
GetVmData: making ScmCall to get HypInfo
GetVmData: No Vm data present! Status = (0x3)
VM Hyp calls not present
Launching fastboot
Fastboot Build Info: Feb 18 2022 17:53:28
usb_shared_hs_phy_init: hs phy cfg size: 12
usb_shared_ss_phy_init: ss phy cfg size: 143
ssusb_phy_init_success_lane_A: 0
Fastboot: Initializing...
VB: Non-secure device: Security State: (0xF7F)
Fastboot: Processing commands
Failed to get the graphics output protocol.
Failed to get the graphics output protocol.
Failed to get width or height Failed to get the graphics output protocol.
Failed to get the graphics output protocol.
Unable to show fastboot menu on screen: Out of Resources
Dev_Common_Speed: Bus Speed: Super
Dev_Common_Speed: Bus Speed: Super Handling
Cmd: getvar:product Handling Cmd:
getvar:version
Upon ssh'ing to it with the debug board I get the following message :
ssh: connect to host xx.xx.x.xxx port 22: Network is unreachable
It should be noted that the chip still pops up in "fastboot devices" as well as in "dmesg -w", so its still alive. Whats going on and how can I fix this?
@Hector-Nevarez Problem solved, was hardware-related. Thank you for your help!