Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Detected

Questions and comments about all other Robots go here.

Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Detected

Postby Hayden » January 22nd, 2018, 6:20 am

Hi all,

It appears I have a strong magnetic field emanating from certain areas of my apartment floor. I purchased a Botvac Connected D3 about a month ago and was disappointed when it appeared to have trouble navigating my apartment. It would stop right in the center of the floor and complain that its path was blocked.

I returned the Botvac, and saw that the new Xiaomi Roborock 2 was available, with pretty stellar reviews. After purchasing it and letting it loose in my apartment, however, I saw similar behavior. Except this time, once it stopped, it exclaimed "error 11, strong magnetic field detected."

Some investigating revealed that, indeed, there is a strong EM field in certain parts of my floor that confuse both the Neato Botvac and Xiaomi Roborock 2. Unfortunately, as I'm leasing this apartment, there's not much I can do to mitigate whatever's causing this magnetic field through my floor.

My question now is how to work around this problem. It seems the only thing these robots use magnetic sensors for is to detect boundary markers, of which I have no need for (I live in a small studio apartment). Given that, I wonder if there's any way to physically disconnect the magnet sensors, or otherwise bridge the wiring so that the sensor input is effectively ignored? Or maybe if I beg hard enough, Xiaomi might ship a feature to allow me to disable boundary marker behavior entirely? :)

Please advise!
Hayden
 
Posts: 10
Joined: January 22nd, 2018, 6:12 am
Location: Redmond, WA, USA

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby glnc222 » January 22nd, 2018, 12:31 pm

Specifications for the Xiaomi model do not show magnetic boundary strip sensors found on the Neato robots, so it is not clear where such sensors are installed inside the robot.
These sensors output an analog voltage around 2.5v normally and a higher level when detecting a magnetic field.
A resistor network can supply a substitute output which can fool the system, or they might work disconnected or grounded depending on software interpretation of the input.

When a boundary sensing robot encounters a mag strip with sensors in the front, mounted in visible protrusions on the bottom of the case, the robot will turn away or back up, similar to encountering a cliff drop with the optical proximity sensors, stairways and such. If a source of magnetic interference keeps the sensor activated when the robot tries to move away, the software will interpret the situation as though the robot is surrounded by barriers and stop with a blockage error. A similar situation occurs when the optical wall sensor on the side of a Neato is broken and finds the robot surrounded by walls.

The mag strip Hall Effect field sensors require a fairly strong field as these things go, from a permanent magnet, strong enough to visibly attract a piece of iron like a nail.

Some kind of heavy motor might produce such a field, as it is often observed computer disk drive data can be damaged by close proximity to heavy electric motors or the magnets in powerful loudspeakers.

The Hacking Forum here includes a thread on adding a Neato type mag sensor to a Samsung Powerbot, which shows the kind of sensor components used -- like a small transistor, often attached to a ferrite antenna bar for the needed sensitivity.

Strong A/C currents in wires or radio transmitters can produce induction of currents in circuits nearby, similar to a transformer -- like radio receivers getting over-ridden by strong transmitters nearby. Such induction could interfere with the internals of a microprocessor chip in a robot, though none has been reported. Exactly what signals the Xiaomi is using to issue the unusual error message is not known except to the company.

The strongest magnetic fields are produced by medical MRI scanning machines, with superconducting magnets, so strong that no metal objects can be allowed in the room as they will fly across the room with potential injuries -- some accidents have occurred with oxygen tanks doing that.
glnc222
Robot Master
 
Posts: 4752
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby Hayden » February 2nd, 2018, 5:18 am

So an update - I did some exploratory surgery on my Roborock to try to find and disconnect/bridge the magnetic sensor to make the bot blissfully ignorant of the EM field strangely present in my apartment.

Upon opening the bot, however, there was no obvious sensor I could find that seemed responsible for detecting magnetic fields - the closest hunch I had was this thin metal sheet that sat under the main board with three contacts. I went ahead and removed this sheet, re-assembled the robot, and became disappointed as it again exclaimed "strong magnetic field detected" on boot.

I've attached a photo of the metal sheet I removed along with all of the other obvious sensors that were connected to the main board. Any guesses or suggestions as to where I should look next? Appreciate any input before I start disconnecting things at random and praying.

Thanks much!

Photos: https://imgur.com/a/IQuRr
Hayden
 
Posts: 10
Joined: January 22nd, 2018, 6:12 am
Location: Redmond, WA, USA

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby mfortuna » February 2nd, 2018, 1:04 pm

The metal sheet is a shield either to keep EMI out or EMI in.

If you can narrow down the area causing the issue you could try using a piece of sheet metal and cover it with a rug.
Mike
Reds x 3, Dirt Dog, Disco (now a parts bot), Create, Scooba 350, and Security Dawg
Evolution Mint
Neato XV-11
Shark Ion 750
User avatar
mfortuna
Robot Master
 
Posts: 5819
Joined: February 5th, 2006, 10:35 am
Location: NH

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby Hayden » February 2nd, 2018, 5:39 pm

Unfortunately the EM field is widespread enough in my apartment that any sort of workaround to try to shield it would be too much of an obstacle or eyesore.

I have a feeling now that the magnetic sensor is hitching a ride in an enclosure with one of the optical sensors. Will be doing some more exploratory surgery this evening/weekend.

If anyone has any more ideas please share :)
Hayden
 
Posts: 10
Joined: January 22nd, 2018, 6:12 am
Location: Redmond, WA, USA

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby glnc222 » February 3rd, 2018, 12:40 pm

If the bot is equipped with magnetic strip sensors in the optical cliff sensor assemblies they might have wire leads to the cpu which can be observed. The IR proximity sensors come in two types, a three wire type with an analog voltage output signal besides 5v power and a ground. A volt meter can detect which wire by observing as a hand moves close and away from the sensor. Another type has four wires, separate signals to operate the emitter and detector, with pulsing supplied by the cpu. In contrast the magnetic sensor would deliver a constant voltage varied with proximity of a strong magnet, close to the sensor. Observing with a volt meter while moving a magnet close and away would identify the lead. See the hacking forum thread on adding mag strip sensor to the Samsung robot. Their sensor is mounted on the front cliff sensor with a separate little board. Mag sensors take power and ground like the IR sensors.

Mag strip sensors do not use metal plates. Small ferrite antenna rods may be attached. Computer equipment does not use metal plates for sensing interference.

It takes a strong field to activate mag strip sensors, strong enough to physically pull on iron objects in the vicinity. Strips must be very close to the sensors to be detected. Such strong fields are rare. The only known source of such conditions is medical: the large EMR scanner equipment in hospitals, with superconducting magnets so strong no metal can be allowed in the room, and oxygen tanks accidentally brought near have caused injuries. It is more likely a mag sensor in the robot, if present, is defective.

Xiaomi should be asked what their system is responding to, given they do not seem to sell mag strip boundary markers for the robot. It would not be surprising there is no magnetic field sensor. Strong radio signals in the area would seem to have effects on other equipment, like radio receivers. Perhaps the software in the robot is defective.

[edit[ The vacuum should be tested somewhere else to determine whether it is defective.

Computers operate at radio frequencies performing digital switching, the frequency often listed as ratings for the processor speed, now in giga hertz (an IR proximity sensor uses a kilo hertz oscillation of the emitter). Consequently digital equipment has metal shielding such as the plate mentioned, to prevent generating interference with radio communications in the vicinity, demanded by government regulations. A common source of radio interference is high voltage power lines in the vicinity. A car radio will usually not work when passing under such lines, drowned out by static noise. A radio transmitter in the vicinity of a receiver will over-ride the tuner and punch through to the output. Cell phones could be similarly disabled. Some equipment, maybe military, would he "hardened" or exceptionally shielded against interference. Solar flares and nuclear explosions (EMP weapons) are noted sources of interference which can even damage electrical equipment.
Last edited by glnc222 on February 3rd, 2018, 5:52 pm, edited 1 time in total.
glnc222
Robot Master
 
Posts: 4752
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby vic7767 » February 3rd, 2018, 4:16 pm

This robot may also be equipped with an IMU (inertia measurement unit) that could be at fault. Normally used to check for keeping level with the floor (tilt warning).
Roomba and Neato Mods, come visit: http://www.vic7767.com/
User avatar
vic7767
Robot Master
 
Posts: 14630
Joined: January 14th, 2006, 8:31 pm
Location: Haughton Louisiana - USA

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby glnc222 » February 3rd, 2018, 5:53 pm

The description of the error as magnetic fields might be a language error in the message software. Perhaps it means something else. Contact Xiaomi for information.
glnc222
Robot Master
 
Posts: 4752
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby Hayden » February 5th, 2018, 12:55 am

Thanks for your input!

I've finally found the offending sensors. Photos:
https://imgur.com/a/bXEPa

The sensors were mounted to the very bottom of the chassis, hidden underneath the bump sensor assemblies.

After finding the suspected culprits, I started to work trying to find creative workarounds to keep the robot from being scared of certain parts of my apartment due to the "magnetic field". I ran the lead for these sensors out the robot's head and mounted them facing up towards the ceiling. Lo and behold, the robot stopped screaming about "error 11" and started vacuuming my living room! However, there were still a few places he was afraid to venture over to - presumably the field was strong enough there that the sensors mounted to his head still reported it.

I learned that using the "remote control" feature from the app I could manually discover the areas the robot didn't like - when guiding the robot with the remote control, he will still refuse to cross into the "magnetic field" areas. I was also able to disconnect the sensor in the middle of the remote control session, enabling the robot to venture into the field unfazed. However, if I disconnect the sensor before the robot starts a cleaning session, an "internal error" will be reported, and the robot will refuse to start.

My new goal now is to figure out how to fake out these sensors to report no magnetic activity. There are 4 pins that connect the sensors to the main board, I've been able to deduce the power and ground leads, but the other two I'm unclear on - presumably some type of serial connection..? I plan to try to hook the sensor up to my Raspberry Pi GPIO ports and try to see if it outputs anything when given power.

I tried to "protocol sniff" the two signal leads when the sensor was hooked up to the bot by bridging them with my Pi's GPIO ports today, but I couldn't decode the resulting waveforms, and they seemed to change whenever one signal lead was connected versus both. I'm not a savvy electrical engineer, so I'm probably missing a whole lot there. A couple photos of the waveforms (signal1, power, signal2) - second photo is zoomed in on one of the "groups" from the first photo:
https://imgur.com/a/3SjLk

Not sure what to make of that. It doesn't look like one is transmit or one is receive. Nor does it look like a protocol like I2C. And I can't find a datasheet on whatever part these sensors are. So I'm a bit stuck here. Next steps are trying to power it from the Pi and see if it outputs anything.

Thanks again for your input :)
Hayden
 
Posts: 10
Joined: January 22nd, 2018, 6:12 am
Location: Redmond, WA, USA

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby Brett » February 5th, 2018, 9:12 pm

Have them pointing up (or any convenient location) but wrap them first with electrical tape, then with Alfoil?
Time flies like an arrow.
Fruit flies like a banana!
User avatar
Brett
Robot Addict
 
Posts: 346
Joined: November 2nd, 2009, 8:50 pm
Location: Perth, Western Australia

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby Hayden » February 5th, 2018, 11:42 pm

I actually have tried wrapping it in aluminum foil. Unfortunately that doesn't seem to stop it detecting a field in some places. :0
Hayden
 
Posts: 10
Joined: January 22nd, 2018, 6:12 am
Location: Redmond, WA, USA

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby glnc222 » February 6th, 2018, 12:59 am

Now it sounds like the bot is equipped for a possible introduction of mag strip boundary markers. Sensors for those must be on the bottom close to the floor, close to the strips for detection.

The Allegro Hall Effect field sensors used on Neato output an analog voltage rising with detection, no oscillation; three wires, power and signal. There may be internal oscillation operating the sensors, processed into a simple output (why a filter capacitor is used on the power lines) -- with the Xiaomi part maybe less internally processed, the way some optical proximity sensors rely on the cpu and others have internal computing.
Oscillation suggests a system which might resemble a metal detector, just speculating, not an expert.

It is understandable that disconnecting a sensor produces an error message, as robot systems try to detect faults in the hardware. Sometimes sensors are not examined until the robot is moving and the sensors activated only when needed (the case with Samsung).

When examining the sensor signals compare them with and without a magnet present. One signal might be a driver oscillation and another some pickup influenced by magnets. Maybe it is a transformer inductor influenced by fields. Such things other than Hall Effect sensors might be more subject to radio interference.

If the sensors can be moved, perhaps the kind of shielding needed is magnetic, not radio interference. Surrounding with iron or ferrite noise reduction material could be needed (used a lot on auto ignition wires, some USB cables -- ferrite rings etc.), not simply conductive material. Small steel mending plates are in hardware which might be used, cut to size. There are also paramagnetic materials I do not know, just wonder about -- deflect fields.
Possibly stacked ferrite square shape bars to make a plate. But ferrite antenna rods increase the sensitivity of Hall Effect sensors, gathering the field...

When surrounding with aluminum foil similar to cable shielding etc., the foil needs to be connected to the system power ground I think.
glnc222
Robot Master
 
Posts: 4752
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby glnc222 » February 6th, 2018, 1:19 am

When one wire is some kind of driving signal, the output signal might be disconnected from the board, and the driving signal patched into the output lead through a resistor to simulate a normal condition -- depending on the signals and operation to be determined.
glnc222
Robot Master
 
Posts: 4752
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby glnc222 » February 6th, 2018, 1:53 pm

I wonder if a small ferrite suppression ring, torroid, over the round spot apparently the detector on the board, might work. These capture fields within the ring I think, while rods are used for antennas, fields looping through air between the bar ends. Rings are used around data cables.
glnc222
Robot Master
 
Posts: 4752
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby Hayden » February 7th, 2018, 5:07 am

glnc222 wrote:When one wire is some kind of driving signal, the output signal might be disconnected from the board, and the driving signal patched into the output lead through a resistor to simulate a normal condition -- depending on the signals and operation to be determined.


That sounds about right actually given what I managed to capture.

Experimenting a bit more... I patched one signal lead through to the other, and that kept the bot happy for about 30 seconds before it realized something was wrong and promptly stopped cleaning, paused for a few seconds, and started again (repeat forever). So assuming what you theorized above, it looks like if you just patch through the "raw input" the bot becomes unhappy with it after some amount of time (presumably it's looking for some difference in the signal coming back). Now the question is how do I introduce a variance that makes the bot happy.

Since I had everything wired up to a breadboard at this point I decided to add some indicators to each lead. Video is below. Red LED is power, yellow LED is yellow signal lead, blue LED is white signal lead. Without the sensor connected, you can see the yellow lead getting voltage from the bot, and the blue lead getting a ping every so often, before the robot gives up and restarts the cleaning cycle. I'm not sure if this means the "return signal" (presumably blue) isn't strictly a variance of the "input signal" (yellow). Hrm...

Video: https://photos.app.goo.gl/gRykUGO5F3OmPrFp1
Hayden
 
Posts: 10
Joined: January 22nd, 2018, 6:12 am
Location: Redmond, WA, USA

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby Medtech » February 7th, 2018, 1:16 pm

To shield the magnetic field, it is necessary to have iron, Fe. this material is easy to find by a characteristic feature - a magnet adheres to it, a thin enough sheet from which it is necessary to make a closed box when a strong magnetic field is present. Good magnetic soft iron can be found in network transformers, where there are many copper wires.
2018-02-07_20-04-59.jpg
2018-02-07_20-04-59.jpg (25.26 KiB) Viewed 4188 times

2018-02-07_20-05-48.jpg
2018-02-07_20-05-48.jpg (18.83 KiB) Viewed 4188 times

2018-02-07_20-07-03.jpg
Thank you GOOGLE translator for the freedom of communication.
User avatar
Medtech
Robot Addict
 
Posts: 150
Joined: October 17th, 2014, 3:12 pm
Location: Moscow Russia

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby Medtech » February 7th, 2018, 1:19 pm

a similar problem has already been solved in the Russian forum, it turned out to be enough to screen the sensor only from the lower side with ordinary roofing iron, thin sheet of iron (жесть)..
Last edited by Medtech on February 7th, 2018, 1:23 pm, edited 1 time in total.
Thank you GOOGLE translator for the freedom of communication.
User avatar
Medtech
Robot Addict
 
Posts: 150
Joined: October 17th, 2014, 3:12 pm
Location: Moscow Russia

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby Hayden » February 7th, 2018, 1:22 pm

Medtech wrote:a similar problem has already been solved in the Russian forum, it turned out to be enough to screen the sensor only from the lower side with ordinary roofing tin.


Do you have a link to this thread? I'm interested to see the discussion!
Hayden
 
Posts: 10
Joined: January 22nd, 2018, 6:12 am
Location: Redmond, WA, USA

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby Medtech » February 7th, 2018, 1:27 pm

the beginning of the discussion is approximately here http://4pda.ru/forum/index.php?showtopi ... ry58320834
Thank you GOOGLE translator for the freedom of communication.
User avatar
Medtech
Robot Addict
 
Posts: 150
Joined: October 17th, 2014, 3:12 pm
Location: Moscow Russia

Re: Xiaomi Roborock 2 - Error 11, Strong Magnetic Field Dete

Postby Hayden » February 7th, 2018, 5:19 pm

Looks like I'll be hitting up the hardware store for some sheet iron to try this out!
Hayden
 
Posts: 10
Joined: January 22nd, 2018, 6:12 am
Location: Redmond, WA, USA

Next

Return to All Other Robots

Who is online

Users browsing this forum: aarolar, BCCISProf, cav1452, jdong, johnnyh, riggster, shsng, vic7767 and 224 guests