Samsung PowerBot Virtual Wall.

Inside the Roomba and Scooba and more, Cool mods, Repair and Upgrades - including the all new iRobot Create Kit. Let's void that warranty baby!

Samsung PowerBot Virtual Wall.

Postby Bandi94 » October 2nd, 2017, 4:32 pm

Hello,
I recently tried to make a virtual wall for a samsung vacuum cleaner (PowerBot Essential), i know that the carrier frequency is 38khz so i started experimenting with different signals. I found a bunch of projects for IRobot virtual wall's where they transmit a simple 1ms ON, 1ms OFF modulated signal so i tried to do the same thing for the samsung one but with no luck, the robot is just going right trough. After a bit of research i found out that the PowerBot series are using the two IR sensors located on his "cheeks" to pick up the remote controller and to search the base so i am almost 100% sure that those sensors are also used to detect the virtual wall which probably sends out some specific "IR code" to tell the robot that you shall not go through.

I don't own a samsung virtual wall because the new models are sold without them so you need to buy them separately. My question would be if someone who own's a samsung "virtual guard" as they call it did try to record the sent code or if he is willing to do it so i can finish my project and why not make it a public project. I have already the whole circuit done it also includes a charging circuit as i am using a li-ion cell, low battery indicating, charging with a simple usb cable, i am just missing the damn signal sent to the robot.

Thank you.
Bandi94
 
Posts: 14
Joined: October 2nd, 2017, 4:18 pm

Re: Samsung PowerBot Virtual Wall.

Postby glnc222 » October 2nd, 2017, 9:06 pm

[edit] Virtual Guard signals found later post below, after much finagling:
http://www.robotreviews.com/chat/viewtopic.php?p=145590#p145590

Proximity detector operation and "Point Cleaning" remote projection signals below at
http://www.robotreviews.com/chat/viewtopic.php?p=145567#p145567

I have looked into this but not sure how I will capture the signal lacking a high frequency storage scope. I have an Arduino, and a PIC data logger up to 1khz. I have a high frequency "simplescope" for PC from Ukraine, but not as good software with it for capture. Unfortunately as below I could not get the signal to examine but have suggestions.

Note the burst modulation on and off so many ms is just power saving, with higher frequency pulses within these intervals.

I did a number of experiments with Virtual Walls in Neato thread "optical boundary marking". These systems use the same standard tech as TV remotes, at least on some, and likely because receivers are made with integrated decoding logic and very sensitive detectors (once available Radio Shack; I have one will try). A standard format coding (certain timing and spacing, besides numeric) is used to both filter out unrelated devices and signals, and to distinguish one remote from another, and to specify which button is pushed on a remote. It appeared that Roomba did not use coding, just the carrier signal (though with expanded Light Houses, maybe some coding is inherently involved, with different operations).

The Data Sheet for one receiver TSOP4838 shows some info on the signals
http://www.vishay.com/docs/82489/tsop322.pdf

First tests show the VG did not receive on either a 38khz or 40khz receiver, though a TV remote was on 38khz. Unless the bursts are too infrequent to register average voltage output over time like the TV remote. These are very powerful over-driving emitters on short duty cycles for distance. The VG has two main emitters besides the top omni directional.
So a phototransistor must be tried next, or tap the emitter in the VG. This also depends on what wave length IR is being used -- needed to make your own emitter. No response was seen on avg voltage out of a 940nm and 540nm phototransistor, with 940 responding to TV remote. Unfortunately the VG could not be disassembled without possibly breaking it; removing the internal screws separates the bottom but some wiring is holding the top, and I'm not risking it.

What you could do is tap the receivers in the robot to see what signals the dock beacon and remote produce. You can also do the same on the remote and dock emitters.
This would give the basic format of the signal, with just some different code used by the VG. A tedious hacking procedure could be performed running all possible codes into it from an emitter to see if any trigger a response -- assuming there is any software for such a response, which is not likely is my guess.
Given how the response software for the VG may be missing I would suggest just adding on your own receiver as I did in the Neato forum "Optical Boundary Mod". You could then even use Roomba VW's. Or, given the reasons below for switching to mag strips, possibly reconsider the value of the optical system in the first place.

An odd specification in Samsung's manual is for a long "minimum" reach of the Virtual Wall, instead of more relevant maximum, which can actually be true because there is a problem with strong beams reflecting off walls for false indications on sensitive receivers. Roomba seems to use a deliberately damped system and only short reaches. In making my own system I severely reduced emitter power. Roomba VW's are detectable 30ft away with sensitive receivers.

I had to cover the omni directional self-protecting top emitter on the Samsung Virtual Guard because created too much interference, and take care over a short kitchen entrance to aim it past the opposite wall to avoid reflections. VR9000 series thread. It is possible Samsung has integrated decoding receivers but I have not opened the rear case, such a puzzle box don't want to break it (no service here on mine imported from Australia). Sending PM re service manual and the Guard I have, in the U.S. I did interface a kitchen timer to the dock, and add side bumper extensions, and disable the wheel extension sensors for furniture problems (Side Bumper Extension and Tilt Mod thread).

Customer service told me the new U.S. models with mag strips (still some Virtual Guard models in Europe), will not use the optical Guards. That is, that part of the software can be missing -- perfectly reasonable. That needs to be tested. You can tell encoding is used because as the manual states, the same receivers are used for the dock beacon, VG's, and the remote, and care is needed not to have two of these simultaneously detected as the signal will be overlapped destroying the digital encoding. Coding must be used to distinguish the sources.

Detectors will deliver the coding pulses for decoding by the main cpu, determining what remote button is pressed. There are also standard emitter encoders. Samsung might even manufacture such components themselves, that being the bulk of their business (major suppliers to the iPhones for instance, despite being into phones themselves). You need to examine the Powerbot receivers. Unlikely there is any UART involved as modern digital things minimize wiring with serial interfaces.
There are several different standard carrier frequencies used (though technically the "carrier" is light waves, and here is just sort of the baud rate of the asynchronus serial interface digitally). International industry standard for these devices.

Of course, you could always add your own optical boundary system to the Powerbot as I demonstrated on a Neato, interfacing it to the cliff sensors with an op amp mixing circuit. Standard IR proximity sensors tend to have an analog voltage output representing distance.

I think Samsung switched to mag strips because the optical system doesn't really work all that well what with reflections and all, lack of beam collimation etc., besides being expensive and inconvenient with batteries etc. I power mine with a 5v wall charger reduced by a string of diodes to 3v though 3v wall adapters are sold -- 5v being the ubiquitous USB (see VR9000 series thread).
Last edited by glnc222 on October 10th, 2017, 10:46 pm, edited 2 times in total.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Samsung PowerBot Virtual Wall.

Postby glnc222 » October 3rd, 2017, 2:30 am

Note some additions by editing made to the above.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Samsung PowerBot Virtual Wall.

Postby glnc222 » October 3rd, 2017, 2:46 am

If possible it would be nice to see a picture of the front system board showing how the mag strip sensors have been added, probably on the bottom of the board.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Samsung PowerBot Virtual Wall.

Postby Bandi94 » October 3rd, 2017, 3:19 am

I own the POWERbot Essential (VR20J9040WG/GE) the one with the blue suction motor cover. The user manual i got with it is very vague and it doesn't tell anything about this subject. I am am not sure if this model still uses VW or mag strips. The model i think it's not very new so it might use the VW. As battery life i don't bother, i made my VW clone with built in li-ion cell (2.2Ah) the total circuit power consumption is somewhere at 3-4mA ( it can go a bit higher depending what they actually send out and at what frequency) that means at least 400hours on one recharge ( aprox 15day's if it is left 24/24, 7/7) so a 3hour charging with a phone charger 2 times / month is not a problem for me.
I would not jump to open it up as it's only 1 week old and i don't wanna lose the warranty.

I successfully recorded the remote and the docking station protocol using TSOP at 38Khz, i will post some pictures when i get home from work. They are different as timing so it's not the same "protocol" it's some custom thing i think. If i remember correctly they have signals as 0.9ms most possibly logic 0 and 2.9ms for logic 1 (maybe 1ms and 3ms, i don't 100% trust my logic analyzer as it is some chines stuff), beside this i found other timing's to so it's some nasty stuff going on with 3 or possible even more signal lengths, it was late last night and i didn't have much time to study the recording, i will today.

I found a used but in working condition VW so i might go for it and try to record it without opening it up, if not then i will pull it apart to see what magic is going on it there.

The "cheek" sensors according to the service manual pictures are standard metal case TSOP sensors, the docking charger also has one, so some kind of communication takes place. Probably the docking station sends some low power signal to the robot and when the robot detects it it send's back some kind of message telling it that it is coming to dock and then the docking station wakes up and sends out some "smarter" stuff as it has two IR LED's in a 45 degree arrangement to guide the robot.
Bandi94
 
Posts: 14
Joined: October 2nd, 2017, 4:18 pm

Re: Samsung PowerBot Virtual Wall.

Postby glnc222 » October 3rd, 2017, 4:23 am

The Samsung website for your country -- or where you bought it -- will likely show specifications for the model sold there and whether it uses mag strips or Virtual Guards. If using Virtual Guards they are usually shown in the User Guide as well. Compare with the U.S. website model no's and specs at samsung.com. The U.S. 9040 is VR2AJ9040WG/AA. Different countries need different wall power plugs in the packages.

There is no mention of either Virtual Guards or Mag strips in the U.S. 9040 User Guide or specs. To determine whether your model might use either of these purchased separately, call the customer service number shown in the User Guide packed with your unit or chat on the website for the country involved. It is possible this lower end unit does not have any boundary marking feature.

Physical barriers can be made of plastic corner guards in paint depts. or wood moldings in some cases.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Samsung PowerBot Virtual Wall.

Postby glnc222 » October 3rd, 2017, 1:34 pm

U.S. customer service does not think the 9040 accepts either Virtual Guards or Mag strips, but does not have factory access. The mag sensors are a cost component, and the software would be needed for the VG. Still they are not listed as parts at SamsungParts.com. The German models with separately sold markers note that on the website. The European 9040 seems listed only in Poland, not German or French websites. I suspect it is the same. The model listings change over time. They are now pushing the 7000 series in many models.
I saw a VG at Amazon U.S. Check if there is one from local Amazon and return it if not usable.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Samsung PowerBot Virtual Wall.

Postby glnc222 » October 3rd, 2017, 3:25 pm

Virtual Guard documentation with models including them make no mention of any pass-through function similar to a Roomba lighthouse, just the simple virtual wall boundary function. So even were you to add your own optical boundary sensors, it would not perform the lighthouse function you wanted.
Also the lighthouse function does not involve any receiver in the lighthouse, and there is none in the Virtual Guard. The function is entirely inside the robot, deciding whether to ignore or respond to the lighthouse emission according to whether a room is completed. The lighthouse may send different codes to indicate what mode is desired by a switch on the lighthouse. There is no such function within the Powerbot software, except it does complete one room at a time in some cases, depending on layout and starting point; mine just happens to reach a doorway as the last section in a room. Not so with a bedroom closet.
As to the Powerbot dock, the specialized IR proximity sensors in the front might also be involved at close range, possibly responding to dock beacon emitters.

The Powerbot often does not dock properly on its first pass, and turns around for a second pass which works. I wonder if the two IR signal receivers form any kind of direction finder.

A very complicated addition with another microcontroller might detect when Powerbot has finished a room as the display shows message "END" when an entire run is completed, and it is returning to the dock.

Roomba devices did have a radio system by which the robot turns on all the lighthouses and VW's when starting to preserve batteries in the VW's when not in use.

On the main battery: adding more capacity cannot be guaranteed to increase the run time, especially if at the maximum 60 minute spec already. In Neato Robotics the run time limit is programmed in the software, and for a reason. Premature exhaustion of the battery is used to monitor aging of the battery and declining capacity, adjusting a model of the battery with run time limit imposed to allow a reserve for returning to base. Whether Samsung does anything similar is unknown. The newer lithium batteries behave differently than the NiMh found in older robots, before advances in the lithium battery chemistry, with LiNMC etc.
Neato includes menu options to reset the battery model, for instance, when installing new batteries.

The capacity of battery cells sold has changed over time, and by the time a Powerbot battery needs replacement, entirely new model Powerbot packs and individual cells may be sold. Battery tech is researched intensively both for electric vehicles and the widespread use of portable devices. A major advance in lithium chemistry yet to be marketed was announced by the original inventor of the lithium battery recently, promising much greater capacity. Maybe replacing lithium with aluminum, I forget, so many of these innovations proposed.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Samsung PowerBot Virtual Wall.

Postby Bandi94 » October 3rd, 2017, 3:36 pm

Thank you for the info, i also sent an email to the local samsung customer service, i am still waiting for an answer if they don't respond i will call in and ask them. It seem's a bit strange why they would remove this functionality.. A more older version sold here is the VR5011 at half of the price of the powerBot and it has both of the functionality's: fence and guard and the new model would get this functionality removed, at the end of the day is just a simple IR decoding. I found some used VW sold for 10-15$ maybe i will buy one and see if it is working. I didn't found new VW in the local stores nor the online one's in my country to be able to return it easily so after all if will not work it's not a very big investment for those used VW, i already spent more money to buy the parts needed for building them at home :).

The fence / guard functionality is described here for the SR8F51 model http://www.shoppingsquare.com.au/p_4090 ... er__SR8F51

This samsung vacuum cleaners are very poorly documented, the user manual don't even mention how to pair the powerbot's wifi with the smart home app, it just's say's that it can be used and that's all. Also if you check a photo with the VW it say's on the top: red - fence, green - guide. So it give's the impression that it has 2 functionality's
Bandi94
 
Posts: 14
Joined: October 2nd, 2017, 4:18 pm

Re: Samsung PowerBot Virtual Wall.

Postby glnc222 » October 3rd, 2017, 3:51 pm

It is marketing. They want to charge more for every feature. They are not the only ones to do this with merely software. Software is a major expense, so also a reason to charge for it. Too bad they simply won't sell software separately from hardware. Some software such as persistent mapping, "select a room" etc. might require more memory in the hardware, but not clear. Another example is the addition of three passes into corners specified on the 9350. No update for my 9250 ever included this (the first WiFi model offered). Four updates were made, I think just for complications of WiFi tech itself.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Samsung PowerBot Virtual Wall.

Postby glnc222 » October 5th, 2017, 1:26 am

DIY Roomba virtual wall thread http://www.robotreviews.com/chat/viewtopic.php?p=65658#p65658
Also PIC and Arduino projects you can Google. Yet used Roomba VW's could cost less than components for DIY.
Whether useful for Samsung unknown though doubtful without coding. With no response software in the Samsung anyway, how could you even tell whether the signal is processed? You could tell whether it gets through the TSOP receiver.
An optical system similar to the Neato forum project might be added to the Samsung, connected to the cliff sensors, maybe just the one in front.
A physical barrier such as plastic corner guards, portable, could be just as convenient if not more so. It would depend on how you planned to use it. Not all that different from mag strips.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Samsung PowerBot Virtual Wall.

Postby Bandi94 » October 5th, 2017, 3:13 am

I know those projects, well about cost i already have everything needed and also i have already one PCB done, total cost for building 2 DIY VW is near buying 1 original unit. Anyway it's not working for the samsung, i think they use some encoding and not only a simple 500hz on/off modulated signal because of the remote controller. Most of roomba's i think don't have a remote so they don't need to decode a bunch of signals so a simple 500hz modulated signal works. I still don't received a response from my local samsung customer service. I found the 9040WG model on a Hungarian online shop website where they mention that VW could be purchased as optional accessory for the robot. I am not sure how accurate that information is.

I also thought of some hacking for the VW, i could easily build a small PCB with smd PIC micro that i could hook-up the the existing TSOP receivers of the robot and decode a specific signal that i would transmit with the VW and when the PIC would found such a signal then it could 'fool' the front cliff sensor. I think i could fit all this inside the robot as a SO23-5 package PIC is very small, the 5 volt power supply could be taken from one of the sensors so no power management would be needed. The only problem is that i will loos the warranty for an almost new powerBot.

I will contact a seller next week who is selling a used VW to ask him if it is still available and if yes then i will buy it in order to test it on the powerbot to see if it will recognize the signal or not.
Bandi94
 
Posts: 14
Joined: October 2nd, 2017, 4:18 pm

Re: Samsung PowerBot Virtual Wall.

Postby glnc222 » October 5th, 2017, 3:05 pm

On Neato I used a cheap key-chain TV remote with added collimating lens from plastic rod
http://www.robotreviews.com/chat/viewtopic.php?f=20&t=17083&p=122722&hilit=optical+boundary#p122722
Instead of attaching additional TSOP receivers on the robot, you might find a signal passing the existing receivers, even though not processed by the cpu. Yet the remote control will also pass this receiver, and also the dock. The remote would also pass through any additional receiver you added. You could use a different baud rate and receiver, but than have to make your own emitter. TV remotes are high intensity using over-driven emitters on short duty cycles.
So it appears a microcontroller would be needed to exclude the other signals from the dock and remote.
With the TV remote a specific key is on in repeat mode, so a particular code is sent which can be filtered by the microcontroller. It might be possible to attach a UART instead of a microcontroller, and then a digital comparator on the output. A continuous, uncoded Roomba VW signal might be easier to decode for exclusion of others.

Behavior of the robot in response to cliff sensors has to be considered. One of the problems with the standard Virtual Guard is trapping the robot in the beam, because when turning, it has not pulled out of beam detection. This is why the beam needs to be collimated. There is a difference between random navigation Roomba's and systematic, guided robots. The guided robots will probe a boundary to map it, and check for any opening to pass through. The random robot just has the same behavior anywhere inside the beam, which is "go away", so it cannot be trapped.

If the Powerbot is started with the front over a table edge, it immediately pulls back and then turns. The software could have a complicated procedure for responding, remembering what it has done, not known.

You might get a Samsung Virtual Guard and return it if not usable. Instead of Samsung service try contacting the Hungarian vendor.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Samsung PowerBot Virtual Wall.

Postby glnc222 » October 5th, 2017, 3:16 pm

Try any real time Chat feature on the Samsung website. You may need to telephone them if they do not respond to email. Customer service may not have any more information than what is shown on their website.
U.S. customer service is good but will not have information on models in other countries.
The German website had models noted with Virtual Guards sold separately, only certain models, and does not show the Essential, blue model.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Samsung PowerBot Virtual Wall.

Postby glnc222 » October 5th, 2017, 3:27 pm

You could also build a mag strip sensor or get a used one for Neato and interface it to the cliff sensors. Trick is getting it close enough to the floor. Maybe cover with duct tape.
It is odd the Samsung side cliff sensors are mounted recessed behind the front, though the middle sensor is close to the front. Neato has just one on each front corner.
Certain black carpet materials will trigger cliff sensors, but most common black materials do not. See hacking forum thread "Optical Cliff Sensor Limitations" study.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Samsung PowerBot Virtual Wall.

Postby glnc222 » October 6th, 2017, 1:00 am

It appears I could access the Powerbot receivers after all, being in the front section, but very tedious because I would have to rebuild the side bumper extension mods made on the side of the case in order to open the front. Then I am not sure with the poorly controlled PC SimpleScope I can capture a screen which would show the coding. Whether I can tap the TSOP's is also a question even if I open it.
So i would really like to know how useful this would be, given how the software for using the input may not exist in that model.
It looks like there may be proximity sensors on the receiver board as well, with such a large cable attachment.
Image

The fact I do not get a signal out of a TV type TSOP from the VG but do with a TV remote makes me wonder about their signal format. I checked this with the SimpleScope and see nothing even in millivolts, while the TV remote shows clear pulses.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Samsung PowerBot Virtual Wall.

Postby Bandi94 » October 6th, 2017, 2:50 am

Hello, yesterday i got a response from Samsung that they forwarded my email for a more technical department to get an answer so in short time i should get some kind of answer.

Also thx for the picture and your time for getting the front cover of. If you have time you could try to capture first with a simple TSOP sensor the docking station and see if you get something on your SimpleScope, if not then the TSOP your using maybe has different frequency. I was able to capture with a 38Khz TSOP from the hardware store the signals sent by the docking station and the remote controller to so it would be odd to me that they would use some other signal in the VW. If you get a signal there then most possible they use different frequency for the VW to don't overlap with the base station and the remote so the next step would that i buy a VW and try it out, if it's recognized by the powerbot then i would take the VW apart and scope directly on the IR LED connection to see what frequency they are using there. I don't know how good you are in the electronics field so i would not ask you to solder some probes on the existing TSOP sensor having the chance to broke something, i don't think that it would worth the risk.

The proximity sensors may be used for Wall detection in order to clean straight along the walls that i noticed my powerbot does. First it cleans perpendicularly to the wall then it makes one more pass along the wall almost rubbing against it leaving just a small gap.

The custom VW detection could contain some logic that would 'fool' the cliff sensor for 1s knowing that the robot will turn around then shut the output off for couple of seconds even if the VW signal is still detected, but this would need a lot of testing and modifications for the existing circuits.
Bandi94
 
Posts: 14
Joined: October 2nd, 2017, 4:18 pm

Re: Samsung PowerBot Virtual Wall.

Postby glnc222 » October 6th, 2017, 12:50 pm

Apology the photo is from the Side Bumper Extension mod long ago, and I would have to redo the taping to open again. Will see over the weekend, and whether I can get another TSOP to check. Also test the Powerbot remote on TSOP on hand. Maybe Samsung will reply by then.
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Samsung PowerBot Virtual Wall.

Postby glnc222 » October 6th, 2017, 1:14 pm

glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Samsung PowerBot Virtual Wall.

Postby glnc222 » October 6th, 2017, 4:35 pm

Got the VG opened; top pries off, held by center screw inside. But all I read so far is an approximately 7khz signal on multimeter I think may be the A/C adapter noise, have to get batteries (adapt AA's).
No focus at all on dual emitters (not the higher ones on/off indicators), plus one in back.
The robot Remote shows up clearly on TV receiver and 940nm photo transistor, but not the Guard.
Maybe the Guard is detected by the IR proximity sensors, which can track a projected spot on the floor with the "point cleaning" feature on higher end models. That also suggests custom sensors, not simple proximity components.
Can't do more till tomorrow afternoon. Maybe the duty cycle is just too low.
VirtualGuard.jpg
glnc222
Robot Master
 
Posts: 4539
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Next

Return to Robotic Hacking

Who is online

Users browsing this forum: vic7767 and 347 guests