iCubGenova09 – Magnetometer of the IMU in the strain2 constantly streams zero #1485

GiulioRomualdi · 2023-01-18T16:04:57Z

Device name 🤖

iCubGenova09

Request/Failure description

Reading the magnetometer from IMU mounted on the strain2 through the yarp port I got a constant vector of zero elements.

Detailed context

I'm trying to read the Magnetometer measurement from the IMU mounted on the FTs sensor of the iCubGenova09.
The multipleanalogserver device is used to stream the data.
I checked the IMU readout, and by reading the yarp port, I noticed that the magnetometer value is constant and equal to zero.

This behavior is observed in all iCub FT imus. Please notice that iCubGenova09 is currently mounting the FT45 in the feet and the FT58 in the legs.

This is the left_leg imu (FT58).

icub@icub-console-xps:~$ yarp read ... /icub/left_leg/imu/measures:o 
[INFO] |yarp.os.Port|/tmp/port/1| Port /tmp/port/1 active at tcp://10.0.0.150:10004/
[INFO] |yarp.os.impl.PortCoreInputUnit|/tmp/port/1| Receiving input from /icub/left_leg/imu/measures:o to /tmp/port/1 using tcp
(((-0.125 -0.4375 -0.5625) 1674054506.32432508469)) (((1.27000000000000001776 -0.340000000000000024425 -9.6300000000000007816) 1674054506.32432484627)) (((0.0 0.0 0.0) 1674054506.32432508469)) (((-178.0625 -7.125 100.875) 1674054506.32432508469)) () () () () () ()

while this is the left_foot imu (FT45) (currently on iCubGenova09 there are two FTs connected to the same ems for the foot)

icub@icub-console-xps:~$ yarp read ... /icub/left_foot/imu/measures:o 
[INFO] |yarp.os.Port|/tmp/port/1| Port /tmp/port/1 active at tcp://10.0.0.150:10004/
[INFO] |yarp.os.impl.PortCoreInputUnit|/tmp/port/1| Receiving input from /icub/left_foot/imu/measures:o to /tmp/port/1 using tcp
(((-0.125 -0.125 0.0) 1674054603.7856593132) ((-0.25 0.0 -0.1875) 1674054603.78365850449)) (((0.530000000000000026645 -1.1399999999999999023 9.69999999999999928946) 1674054603.7856593132) ((0.770000000000000017764 -1.14999999999999991118 9.64000000000000056843) 1674054603.78365850449)) (((0.0 0.0 0.0) 1674054603.78766036034) ((0.0 0.0 0.0) 1674054603.78365850449)) (((-6.125 -2.875 186.6875) 1674054603.7856593132) ((-6.3125 -4.1875 169.25) 1674054603.7836587429)) () () () () () ()

Accordingly to the multipleAnalogSensorsSerializations.thrift, the magnetometer is the third element of the yarp bottle.
So for the foot FTs (FT45) we have

(((0.0 0.0 0.0) 1674054603.78766036034) ((0.0 0.0 0.0) 1674054603.78365850449))

Similarly for the leg (FT58) we have

(((0.0 0.0 0.0) 1674054506.32432508469))

I also tried to check the magnetometer data coming from the IMU mounted on the iCub head and in this case the value changes as shown in the following plot

Here the outcome of the head inertials port

icub@icub-console-xps:~$ yarp read ... /icub/head/inertials/measures:o 
[INFO] |yarp.os.Port|/tmp/port/1| Port /tmp/port/1 active at tcp://10.0.0.150:10004/
[INFO] |yarp.os.impl.PortCoreInputUnit|/tmp/port/1| Receiving input from /icub/head/inertials/measures:o to /tmp/port/1 using tcp
(((-0.1875 0.1875 0.0) 1674057757.53589510918)) (((0.359999999999999986677 0.800000000000000044409 -9.74000000000000021316) 1674057757.53589510918)) (((3.750000000000000095e-06 3.67499999999999988134e-05 4.89999999999999984179e-05) 1674057757.53589510918)) (((242.375 2.0 -175.3125) 1674057757.53589510918)) () () () () () ()

Additional context

I opened the issue here however I think this is not a problem related to the Hardware of the robot since is involving all the FTs.

@pattacini if you want we can move the discussion in a separate repo.

cc @traversaro and @G-Cervettini

How does it affect you?

No response

The text was updated successfully, but these errors were encountered:

pattacini · 2023-01-18T16:34:10Z

Thanks for reporting @GiulioRomualdi 👍🏻

I opened the issue here however I think this is not a problem related to the Hardware of the robot since is involving all the FTs.

@pattacini if you want we can move the discussion in a separate repo.

True that it might be unrelated to the HW but it's still a problem whose progress can be shared with the community herein.

We will plan for it soon.
Stay tuned.

AntonioConsilvio · 2023-01-24T13:09:05Z

Hi @GiulioRomualdi, we are organizing to address this issue in early February.

cc @maggia80 @Fabrizio69 @pattacini

pattacini · 2023-01-24T13:34:17Z

/remind on February 06

octo-reminder · 2023-01-24T13:34:19Z

⏰ Reminder
Monday, February 6, 2023 10:00 AM (GMT+01:00)

on

GiulioRomualdi · 2023-01-24T16:20:12Z

That's great! Thank you!

octo-reminder · 2023-02-06T09:00:04Z

🔔 @pattacini

on

pattacini · 2023-02-07T07:31:32Z

Just planned for this sprint!

Nicogene · 2023-02-14T09:01:18Z

We added icub-tech-iit/ft-sensors-simulink#11 the possibility to read the magnetometer and IMU_STATUS data, and although it states that the calibration of the magnetometer is good also with the Simulink model we have only 0 on the magnetometer data

pattacini · 2023-02-14T12:39:38Z

Next steps from @Nicogene:

Since the firmware that handles the imu is shared between the strain2 and rfe (they both mount imu bosch bno055), an interesting test would be to read the rfe magnetometer data and the eul angles.

My idea is that the magnetometer problems lead to jumps on the yaw.

We could try:

Instead of getting all the data we should try to get only the magnetometer

Wild guess: maybe it is a Faraday cage problem induced by the metal of the FT? The test on the rfe should answer to this question. (Tested with no success ❌)

Check if there is a problem in the electornics (see this similar problem) cc @icub-tech-iit/silo-electr

Nicogene · 2023-02-17T10:48:20Z

I tried to read the magnetometer data through the old imu we used to mount on icub-head, that has bno055 sensor as well.

I wrote a very simple executable that instantiates the multipleanalogsensorsclient and I got only magnetometer data, and in this case change and it is not 0.

imuBosch_BNO055 device reads 32 byte starting from the ACC register (0x08)
https://github.com/robotology/yarp/blob/e56e474fcfb574e39945a4c1dbf3256cfad202e6/src/devices/imuBosch_BNO055/imuBosch_BNO055.cpp#L622

The imu is configured in NDOF

In our firmware instead, we are reading 47 bytes, but I don't think this can affect somehow the magnetometer readings.

And also in the firmware we are setting the NDOF mode.

https://github.com/robotology/icub-firmware/blob/e858dc3ab33f6c3f8aeb1f24413f7ef3f50e90d8/emBODY/eBcode/arch-arm/embot/app/embot_app_application_theIMU.cpp#L530

The only difference seems to be the electronics boards.

pattacini · 2023-02-17T10:55:03Z

The next step will foresee configuring the FT-IMU in AMGEQ mode.

pattacini · 2023-02-23T09:07:18Z

Some interesting findings we came up with lately.

From @marcoaccame:

I have tested the acquisition w/ Set::AMGEQL rather than Set::FULL and nothing changes.

Moreover, as I spotted that the use of fusion may change the reading of the magnetometer, I did some tests on some boards I have:

on strain2: the magnetometer gives zero in both fusion and non fusion mode;

on mtb4c: sometimes the magnetometer appears but only if fusion is off.

on rfe: magnetometer ok if fusion is on, all zero if fusion is off.

Very strange.

Also, @AntonioConsilvio and @Nicogene reported that, among the three boards, the IMU unit is plugged with a different kit of resistances, which is worth investigating.

Nicogene · 2023-03-13T15:02:47Z

We repeated the tests on the 3 boards (strain2, rfe, mtb4) that mount the bno055 imu, and we checked also the sw version of the unit, and it is the same in all the 3 boards, it seems that the magnetometer works only on rfe, but it is not clear why.

	FUSION	NO-FUSION	Bn055 sw version
`mtb4`	❌	❌	swRev:785 blVer: 21
`rfe`	✅	✅	swRev:785 blVer: 21
`strain2`	❌	❌	swRev:785 blVer: 21

Fabrizio69 · 2023-03-16T08:11:34Z

Ciao @Nicogene can you recover code, revision and serial number for the boards under test?

pattacini · 2023-03-16T08:15:04Z

Also, @AntonioConsilvio and @Nicogene reported that, among the three boards, the IMU unit is plugged with a different kit of resistances, which is worth investigating.

Just for completeness, we verified that the differences were relative just to the pull-up stage.
The wiring and the connections are correct in the diagrams. We will be checking if we can spot differences in the boards that are not reported in the schemes.

maggia80 · 2023-03-16T08:19:17Z

since in the schematic, the only difference between RFE and the other two cards is in the selector pin BNO055_SELCOM
Can we check if we are controlling properly this pins setting it to 0? We can check with the oscilloscope to double check.

I have also a curiosity, are we using the configuration for using the external oscillator or the internal one? How the SYS_TRIGGER is configured?

Nicogene · 2023-03-16T08:47:56Z

I have also a curiosity, are we using the configuration for using the external oscillator or the internal one? How the SYS_TRIGGER is configured?

From this search it seems that on mtb4 and strain2 is configured the external oscillator, on rfe it doesn't (see here)
Maybe setting the external clock makes the magnetometer stop working? 🤔

cc @marcoaccame

maggia80 · 2023-03-16T09:19:43Z

I'll change to ~~internal~~ external also the RFE to see if this is the problem

marcoaccame · 2023-03-16T09:44:33Z

I have also a curiosity, are we using the configuration for using the external oscillator or the internal one? How the SYS_TRIGGER is configured?

From this search it seems that on mtb4 and strain2 is configured the external oscillator, on rfe it doesn't (see here)

Maybe setting the external clock makes the magnetometer stop working? 🤔

cc @marcoaccame

Hi, this is not the code running on the boards, rather the original HW test project. You have to look at the code inside the embot::hw::bno055 driver.

marcoaccame · 2023-03-16T10:34:35Z

I have also a curiosity, are we using the configuration for using the external oscillator or the internal one? How the SYS_TRIGGER is configured?

All boards use the default mode after bootstrap which is internal oscillator.

marcoaccame · 2023-03-16T10:55:24Z

since in the schematic, the only difference between RFE and the other two cards is in the selector pin BNO055_SELCOM

Can we check if we are controlling properly this pins setting it to 0? We can check with the oscilloscope to double check.

I believe this pin was added to change comm from i2c to usart.

The pin BNO055_SELCOM is not driven in the code. The symbol does not even appears in cube mx.

But the value of PS1, the associated pin on the bno055, must be zero because i2c communicates fine.

maggia80 · 2023-03-16T11:04:14Z

the default mode after bootstrap which is internal oscillator.

This is not the best choice. We should use the external one. The internal one drift quite a lot

marcoaccame · 2023-03-16T11:17:52Z

the default mode after bootstrap which is internal oscillator.

This is not the best choice. We should use the external one. The internal one drift quite a lot

It can be easily done.

@Nicogene, if you want I can pass you some code to place just after the reset phase in the driver.

marcoaccame · 2023-03-16T15:27:52Z

In embot_hw_bno055.h add register value 0x3F

...

        OPR_MODE        = 0x3D,
        PWR_MODE        = 0x3E,
        SYS_TRIGGER = 0x3F,
        AXIS_MAP_CONFIG = 0x41,
        AXIS_MAP_SIGN   = 0x42
    };

In it’s cpp write 0x80 to the above register inside the init() function just after the bootstrap of the chip.

...

s_powerON(s, PORtime);

// enable the external oscillator by writing 0x80 in reg 0x3F
embot::hw::bno055::write(s, embot::hw::bno055::Register::SYS_TRIGGER, 0x80, 5*embot::core::time1millisec);

...

That should be all.

pattacini · 2023-03-22T08:46:26Z

Hi there,

We have an update on this.

From @marcoaccame:

I spotted that the mtb4 runs at 16 MHz and the rfe runs at 80 MHz. So I clocked the mtb4 at 64 MHz and bingo!

Just WIP however because at 64 MHz CAN timing are not correct yets.

Stay tuned.

pattacini · 2023-03-23T10:11:14Z

@marcoaccame has churned out a patch in:

Removal of mtb4fap board icub-firmware#359

We need to thoroughly test it and be careful while merging as there are other pending PR's in the queue.

pattacini · 2023-03-29T15:15:09Z

@marcoaccame drafted a PR:

Fix of reading of IMU magnetometer in boards mtb4, mtb4c, strain2, strain2c icub-firmware#361

Nicogene · 2023-03-30T10:11:10Z

robotology/icub-firmware#361 fixed indeed the magnetometer, see the plots above

Thanks heap @marcoaccame!

We will proceed then releasing the binaries

Magnetometer

`strain2`

`mtb4`

cc @GiulioRomualdi

GiulioRomualdi · 2023-03-30T10:21:00Z

That's great! Thank you a lot for the effort!

marcoaccame · 2023-03-30T10:39:30Z

OK, thanks @Nicogene. I have just merged the PR for the sources:

Fix of reading of IMU magnetometer in boards mtb4, mtb4c, strain2, strain2c icub-firmware#361

As I understand you will take care of the PR for the binaries which will soon follow.

In the meantime, I attach in here the new binary v2.1.0 for the strain2 board: strain2-v2.1.0.zip

Nicogene · 2023-03-31T09:47:22Z

The fix has been delivered also in terms of binaries:

strain2 v2.1.0, strain2c v2.1.0, mtb4 v1.21.0, mtb4c v2.1.0 icub-firmware-build#77

I think we can close the issue 🚀

cc @GiulioRomualdi

GiulioRomualdi added the iCubGenova09 (iRonCub3) S/N:000 iCub 3 label Jan 18, 2023

GiulioRomualdi assigned Fabrizio69 and AntonioConsilvio Jan 18, 2023

GiulioRomualdi mentioned this issue Jan 19, 2023

iCubGenova09 – Jumps on the estimated yaw angle provided by the robot IMU #1487

Closed

pattacini assigned sgiraz and unassigned Fabrizio69 Feb 1, 2023

pattacini assigned Nicogene Feb 19, 2023

pattacini mentioned this issue Feb 19, 2023

Magnetometer streams zero when embObjIMU and multipleanalogsensorsserver is used robotology/icub-main#824

Closed

sgiraz changed the title ~~Magnetometer of the IMU in the strain2 constantly streams zero~~ iCubGenova09 – Magnetometer of the IMU in the strain2 constantly streams zero Mar 22, 2023

marcoaccame mentioned this issue Mar 30, 2023

Fix of reading of IMU magnetometer in boards mtb4, mtb4c, strain2, strain2c robotology/icub-firmware#361

Merged

Nicogene closed this as completed Mar 31, 2023

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

iCubGenova09 – Magnetometer of the IMU in the strain2 constantly streams zero #1485

iCubGenova09 – Magnetometer of the IMU in the strain2 constantly streams zero #1485

GiulioRomualdi commented Jan 18, 2023

pattacini commented Jan 18, 2023

AntonioConsilvio commented Jan 24, 2023

pattacini commented Jan 24, 2023

octo-reminder bot commented Jan 24, 2023

GiulioRomualdi commented Jan 24, 2023

octo-reminder bot commented Feb 6, 2023

pattacini commented Feb 7, 2023

Nicogene commented Feb 14, 2023

pattacini commented Feb 14, 2023

Nicogene commented Feb 17, 2023

pattacini commented Feb 17, 2023

pattacini commented Feb 23, 2023 •

edited

Loading

Nicogene commented Mar 13, 2023

Fabrizio69 commented Mar 16, 2023

pattacini commented Mar 16, 2023

maggia80 commented Mar 16, 2023

Nicogene commented Mar 16, 2023 •

edited

Loading

maggia80 commented Mar 16, 2023 •

edited

Loading

marcoaccame commented Mar 16, 2023

marcoaccame commented Mar 16, 2023

marcoaccame commented Mar 16, 2023

maggia80 commented Mar 16, 2023

marcoaccame commented Mar 16, 2023 •

edited

Loading

marcoaccame commented Mar 16, 2023 •

edited

Loading

pattacini commented Mar 22, 2023

pattacini commented Mar 23, 2023

pattacini commented Mar 29, 2023

Nicogene commented Mar 30, 2023 •

edited

Loading

GiulioRomualdi commented Mar 30, 2023

marcoaccame commented Mar 30, 2023

Nicogene commented Mar 31, 2023

iCubGenova09 – Magnetometer of the IMU in the strain2 constantly streams zero #1485

iCubGenova09 – Magnetometer of the IMU in the strain2 constantly streams zero #1485

Comments

GiulioRomualdi commented Jan 18, 2023

Device name 🤖

Request/Failure description

Detailed context

Additional context

How does it affect you?

pattacini commented Jan 18, 2023

AntonioConsilvio commented Jan 24, 2023

pattacini commented Jan 24, 2023

octo-reminder bot commented Jan 24, 2023

GiulioRomualdi commented Jan 24, 2023

octo-reminder bot commented Feb 6, 2023

pattacini commented Feb 7, 2023

Nicogene commented Feb 14, 2023

pattacini commented Feb 14, 2023

Nicogene commented Feb 17, 2023

pattacini commented Feb 17, 2023

pattacini commented Feb 23, 2023 • edited Loading

Nicogene commented Mar 13, 2023

Fabrizio69 commented Mar 16, 2023

pattacini commented Mar 16, 2023

maggia80 commented Mar 16, 2023

Nicogene commented Mar 16, 2023 • edited Loading

maggia80 commented Mar 16, 2023 • edited Loading

marcoaccame commented Mar 16, 2023

marcoaccame commented Mar 16, 2023

marcoaccame commented Mar 16, 2023

maggia80 commented Mar 16, 2023

marcoaccame commented Mar 16, 2023 • edited Loading

marcoaccame commented Mar 16, 2023 • edited Loading

pattacini commented Mar 22, 2023

pattacini commented Mar 23, 2023

pattacini commented Mar 29, 2023

Nicogene commented Mar 30, 2023 • edited Loading

Magnetometer

strain2

mtb4

GiulioRomualdi commented Mar 30, 2023

marcoaccame commented Mar 30, 2023

Nicogene commented Mar 31, 2023

pattacini commented Feb 23, 2023 •

edited

Loading

Nicogene commented Mar 16, 2023 •

edited

Loading

maggia80 commented Mar 16, 2023 •

edited

Loading

marcoaccame commented Mar 16, 2023 •

edited

Loading

marcoaccame commented Mar 16, 2023 •

edited

Loading

Nicogene commented Mar 30, 2023 •

edited

Loading

`strain2`

`mtb4`