Roomba 550 Cliff Sensors -- New Fix

The very latest news and updates for the iRobot Roomba robotic vacuum, the Scooba Robotic Washer and the Dirt Dog workshop sweeper. All discussion and troubleshooting questions go here.

Re: Roomba 550 Cliff Sensors -- New Fix

Postby athyrio » November 15th, 2009, 5:50 pm

I'll add to the chorus of thanks for that amazing walk through! I love tearing things apart to fix or modify them, but this one took some serious cojones to try for the first time. Kudos!

I also want to add a couple of clarifications...

1) I had some difficulty finding all the tabs for removing the decorative top. I found a nice YouTube video of the process here:

[video]http://www.youtube.com/watch?v=EaZibdOIeD0[/video]

No secrets revealed - just a nice illustration of how much force to exert, and where.

2) In step 12, note that the top-middle screw is smaller than all the others. Make sure to get it back in the right hole!

3) Securing the detector and emitter directly to one another using opaque tape also works fine. There's no reason for them to be left free of one another. I used electrical tape, figuring it would be less likely to come undone.

4) Be careful during reassembly to make sure the cliff sensor wires don't get pinched in the semi-circular plastic attachment to the circuit board. I used some strips of electrical tape to make sure they stayed in their channels and slots during reassembly.
athyrio
 
Posts: 1
Joined: November 15th, 2009, 5:43 pm

Re: Roomba 550 Cliff Sensors -- New Fix

Postby benjamin78au » March 18th, 2010, 8:11 am

being IR, if you had a faulty one, could you not phot it with a digital camera and see which LED doesnt light up
benjamin78au
 
Posts: 10
Joined: October 31st, 2009, 9:20 pm

Re: Roomba 550 Cliff Sensors -- New Fix

Postby Gordon » March 18th, 2010, 11:07 am

benjamin78au wrote:being IR, if you had a faulty one, could you not phot it with a digital camera and see which LED doesnt light up
Ben, if you had taken time to read the OP, follow to the linked page provided there, you would have learned the purpose of this thread.

Having done that, you would have seen that this thread has nothing to do with non-operating IREDs.
Gordon
Robot Master
 
Posts: 4304
Joined: April 6th, 2005, 2:02 am
Location: Santa Ynez, CA USA

Re: Roomba 550 Cliff Sensors -- New Fix

Postby mfortuna » March 18th, 2010, 11:46 am

I have a dumb question. Scotch tape over the buttom of the sensor doesn't work? It works fine on 400 series bots.
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: 5831
Joined: February 5th, 2006, 10:35 am
Location: NH

Re: Roomba 550 Cliff Sensors -- New Fix

Postby Gordon » March 18th, 2010, 12:44 pm

mfortuna wrote:I have a dumb question. Scotch tape over the buttom of the sensor doesn't work? It works fine on 400 series bots.
That's not a "dumb" question! I asked myself that very question, clear back at the beginning of the 5XX's existence, when someone -- I think it was vic7767 -- reported that tape did not work as it did on the 4XXX. I'm talking about the Scotch or 3M paper-mending tape having a frosted surface you can write on.

That frosted surface will scatter the IRED's emitted energy back in the direction of its PT partner to trick the robot into believing a floor is always present.

The only difference that is externally obvious between the two robot series is the 5XX-series window that precludes debris accumulation in each cliff-sensors "pocket". So, it may be that the tape's adhesive optically "wets' the window surface, which then (somehow) perturbs the back-scatter magnitude.

One way of exploring that hypothesis would be to apply patches of the frosted-tape to each window with the frosted surface contacting the window, then hold those patches in place by slapping tape over them.

Even the aluminum-foil reflector was shown to be ineffective (as I recall). That may have been due to the foil's surface reflecting the IR energy too far off axis, partly due to its specular reflectance, and partly due to the foil not being on the floor where lines of "sight" of IRED and PT converge. IOW, the back-directed energy mostly missed hitting the PT.

That hypothesis could be tested by making the foil into a scattering type reflector. To do that, rip off a fair size piece of foil (thin foil), say 6" x 6", then crumple it into a tight ball. Unpack the ball enough to permit a second crumpling operation, after which, do the unpacking and notice the foil now exhibits thousands of tiny facets that reflect light in many directions. Now you are ready to cut out little window size pieces of that material and tape them in place over each cliff-window.

Then run dark-floor tests.
Gordon
Robot Master
 
Posts: 4304
Joined: April 6th, 2005, 2:02 am
Location: Santa Ynez, CA USA

Re: Roomba 550 Cliff Sensors -- New Fix

Postby mfortuna » March 18th, 2010, 4:08 pm

I get it. I wonder if you could remove the window and apply tape on the inside of the window, or remove the window all together since the tape would replace the window.
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: 5831
Joined: February 5th, 2006, 10:35 am
Location: NH

Re: Roomba 550 Cliff Sensors -- New Fix

Postby benjamin78au » March 18th, 2010, 5:37 pm

i have read thru the OP, was mainly curious , as it is designed around removing the cliff sensor. The question about the IR is more based on a few posts about people thinking they have a faulty sensor would it be a way of diagnosing with out disassembly?
benjamin78au
 
Posts: 10
Joined: October 31st, 2009, 9:20 pm

Re: Roomba 550 Cliff Sensors -- New Fix

Postby vic7767 » March 18th, 2010, 5:45 pm

benjamin78au wrote:The question about the IR is more based on a few posts about people thinking they have a faulty sensor would it be a way of diagnosing with out disassembly?
One could perform the tests described in the Roomba 500 Service Manual to determine pass/fail of the cliff sensors.
Roomba and Neato Mods, come visit: http://www.vic7767.com/
User avatar
vic7767
Robot Master
 
Posts: 14735
Joined: January 14th, 2006, 8:31 pm
Location: Haughton Louisiana - USA

Re: Roomba 550 Cliff Sensors -- New Fix

Postby Gordon » March 18th, 2010, 5:54 pm

mfortuna wrote:I get it. I wonder if you could remove the window and apply tape on the inside of the window, or remove the window all together since the tape would replace the window.
Mike, I was about to say "Conceptually, I would say yes.". Followed by: " But, to see the practical limitations of implementing either idea you should first review Dave's picture show -- the final five steps tell the story".
http://www.pottsland.com/roomba/Roomba_ ... ensor.html
... and I then followed my own advice. I have been through Dave's page about ten times, but never noticed what I just saw in a cliff-module's design (not that I've looked all that thoroughly, until today!). Today, I see the reason why slapping a back-scatter reflector right against the window, either outside or inside, will not reliably close the optical 'circuit'!

Mike, go there and study the pix associated with steps #21 & #22, and you will see what prevents that from taking place.

Hint: Also study the step #24 image.
Gordon
Robot Master
 
Posts: 4304
Joined: April 6th, 2005, 2:02 am
Location: Santa Ynez, CA USA

Re: Roomba 550 Cliff Sensors -- New Fix

Postby Gordon » March 19th, 2010, 4:05 pm

As I had said:
Gordon wrote:...I have been through Dave's page about ten times, but never noticed what I just saw in a cliff-module's design ... Today, I see the reason why slapping a back-scatter reflector right against the window, either outside or inside, will not reliably close the optical 'circuit'! ..., go there and study the pix associated with steps #21 & #22, and you will see what prevents that from taking place. ...Hint: Also study the step #24 image.
Upon doing that examination, the item that caught my attention is (what appears to be) an optical mask that has been placed between the cliff-sensor's IRED (IR-emitter) and its partner, the PT (photo-transistor detector of the 'reflected' IR energy). Here is the image (borrowing vignettes from Dave's 5XX work) that drew my attention:
cliff-moduleSK2.JPG

The optical barrier is the vertical, black bar (which I assume spans the front to back dimension of the cup-like 'window'-casting that the clam-shell holder (of IRED & PT devices) slips into). See Dave's htm to clarify that description.

There are two double-dashed lines to be noted in the above image. The lowest one represent the floor, while the one above it demarks the elevation of a mid-line through the two window apertures. These references are carried over into the next image which shows the guts of the cliff-module, after stripping away the clam-shell clamp and most of the window-casting (only the lowest window extremity shows).
cliff-moduleSK1.JPG
{You may want to click on this one, so features can be more easily seen.}
I have superimposed sketches of the IRED and its estimated beam-fan, the RED stuff at your RHS, and the PT, BLU, with an estimate of its view-angle, shown at your LHS.

With a little study of that image you will be able to tell that the IREDs radiance fan is delimited at right by the window's clear aperture, and at left by some gray plastic structure -- looking somewhat like a rectangular block.

In a similar fashion, the PT's FOV (field of view) is limited at far left by its window segment. and to its right-side by another gray block.

The black barrier slips up between those two gray-blocks at final assembly.

If you accept all the above as truth, it then becomes easy to visualize (since no aid will be found in the pix) that a reflecting surface (smooth aluminum foil, or scattering surface (white-paper, or crumpled and straightened foil) which is placed in close proximity to the IRED's window, either external or internal, will succeed in mostly re-directing the emitted IR beam-strength onto black & gray barrier surfaces, rather than onto the PT!

Application of (opaque) Al-foil over the entire module-window actually blocks the PT view altogether! That's why that material never succeeded!
---------------------------------------------------------

One may ask why it was so easy to defeat cliff-sensor operation with the 4XXX-Roomba design. To answer that, take a look at its cliff-module interior:
rt-rearCliffMdulApartLoRez.jpg

In this view you can see the two E-O devices and wiring at RHS. The dark-blue device is the PT, and the lower, water-clear part is the IRED.

At their left, the clam-shell clamp is shown separated, and the semi-cylindrical receptacle for each device is visible. As I recall the angular spread between device axes was 57-degrees and split symmetrically about the vertical. The only "barrier" between the two was created by setting them into short tunnels, so the PT had no direct view of the IRED's lens, and partly to reduce the IRED's beam-spread, and PT's FOV. Hence, by applying reflective trickery right at the two ports, there was good coupling between the IRED & PT.
---------------------------------------------------------

If you've been following all of that, you may come up with a question like: "Why did iRobot put all that optical-barrier stuff between the 5XX's IRED & PT?".

I'll bet this is what happened:
a) At the start of 5XX design, there was a specification to enclose cliff-sensors to minimize the amount of debilitating debris accrual.
b) Mechanical designer's said, "OK, we'll just slap a window in front each module. No sweat!"
c) They did (b), then (actually) did some cleaning runs with an Engineering Model 500! "Whoa! We have trouble with those cliff-windows -- once they get dust-coated, the dust scatters enough IR over to the PT that a fault occurs! Bummer! We better get Joe. the optics guy to give us a hand.", they say. Back to the drawing board with that window-design.
d) Joe suggests: "Look here, we mask the IRED's window section, by erecting this opaque barrier between the two, so the PT can't see that window half, or the dust on it!".

Q.E.D.

A few words explaining why vic7767's white-paper blinder has met with success in this thread:
viewtopic.php?f=1&t=7933
... are appropriate.

When people pay attention to the specific instructions, they generally find success. I think the critical portion of those instructions is the frontal wrap-around of the white paper. I also think its that small strip of paper across the front face of a cliff module that is the active portion of the paper-fix.

Inside the module, the IRED fully illuminates its window across the front to back span. The plane-parallel window facet is surrounded by a beaded frame (cast from the window material). IR irradiance incident at window's edge will likely pass into that "bead", transmit through it and be fanned out by the bead acting somewhat as a cylindrical lens. In Roomba's normal operation, such radiance is simply lost into Roomba's forward space. However, if something like vic7767's paper intercepts that forward directed IR energy (naturally there is a lateral component in the general direction of the PT, too), it becomes scattered back towards the cliff module. As the scattered IR enters the bead-frame on the PT side, a fraction of that energy will be deflected upward via internal reflection off the (interior) curved wall of the bead, and then detected by the PT. Obviously, that is a very lossy transfer of energy from one side to the other, but, keep in mind that cliff sensors have been designed to work with a very low amount of IR return.
Gordon
Robot Master
 
Posts: 4304
Joined: April 6th, 2005, 2:02 am
Location: Santa Ynez, CA USA

Re: Roomba 550 Cliff Sensors -- New Fix

Postby dura321 » March 20th, 2010, 4:05 pm

Nice job explaining it all
User avatar
dura321
Forum Moderator
 
Posts: 809
Joined: December 14th, 2005, 2:43 pm
Location: Massachusetts IROBOTS home town

Re: Roomba 550 Cliff Sensors -- New Fix

Postby vic7767 » March 20th, 2010, 5:12 pm

That makes sense Gordon given the fact that when testing the different solutions that were provided by others on the earlier vintage Roombas, none worked on the 5XX models.
Roomba and Neato Mods, come visit: http://www.vic7767.com/
User avatar
vic7767
Robot Master
 
Posts: 14735
Joined: January 14th, 2006, 8:31 pm
Location: Haughton Louisiana - USA

Re: Roomba 550 Cliff Sensors -- New Fix

Postby johnhst » May 5th, 2011, 7:55 am

While trying to disable the cliff sensors on my 530, I tried all the different white paper/tape tricks and all ended with a warning from my robot to move to a different area. I finally realized my machine had the sensors mounted level with the surrounding surface and there is a barrier between the LED and the receiver. (See photo 1) My tape and paper were closing off any path for the light to travel from the LED to the receiver and had nothing to do with reflectivity of the paper. My solution was to cut a small light channel from a pen cover (photo 2) and tape it over the sensor with aluminum duct tape (photo 3). Any tape will do but the aluminum duct tape should last longer than masking or scotch tape. This light channel allows the light to reflect from the LED to the receiver. Hope this help solve some of the cliff sensor disable problems.

Cliff Sensor copy.jpg
Light Chanel copy.jpg
Taped on Sensor copy.jpg
johnhst
 
Posts: 6
Joined: January 1st, 2011, 10:55 am

Re: Roomba 550 Cliff Sensors -- New Fix

Postby larrylwill » June 16th, 2011, 4:26 pm

I just bought a 560 and also have a some carpets that are either all dark or have dark areas and always had the problem with my Discovery and now my 560. I was about to take it apart when I found johnhst post. I had tried the tin foil method and found that it could not be flat and had to have some space between it and the sensor. So I took an old marker and cut 4 cup pieces as explained and used aluminum tape and it works as advertised and only took 15 minuets. If it lasts its a far faster solution than the disassembly solution. BTW I live in a ranch style house, no stairs, there should be a switch to disable those sensors. Thanks johnhst
larrylwill
 
Posts: 29
Joined: November 27th, 2006, 7:20 pm

Re: Roomba 550 Cliff Sensors -- New Fix

Postby larrylwill » June 26th, 2011, 3:34 pm

Well I tried the tape, I also used aluminum tape but one of them would get pushed off every few runs. So I decided to go the whole way and tape the sensors together. Took about a hour and now I shouldn't have to worry about it anymore. Thanks for the walk through. It works perfectly.
larrylwill
 
Posts: 29
Joined: November 27th, 2006, 7:20 pm

Re: Roomba 550 Cliff Sensors -- New Fix

Postby ichris » June 29th, 2011, 8:29 am

Dajiu - thanks for the awesome write-up. In the forth paragraph, you mention "The hardware fix listed above will cut the wires to these sensors and connect 1000 ohm between them..." Can you elaborate a little more on the "hardware fix"? http://www.pottsland.com/roomba/Roomba_5xx_Cliff_Sensor.html
ichris
 
Posts: 45
Joined: February 14th, 2009, 8:52 pm

Re: Roomba 550 Cliff Sensors -- New Fix

Postby Sagrat » October 30th, 2018, 4:50 pm

Hello,

Someone can tell me the specification of the IR LED in cliff sensor?

My Roomba 800 has one of these LEDs who has blown. Because of that, I have an error 6 and the Roomba refuses to move. I just need to change the defect LED, not all sensor.

Thanks for your help.
Sagrat
 
Posts: 5
Joined: October 30th, 2018, 4:32 pm

Re: Roomba 550 Cliff Sensors -- New Fix

Postby vic7767 » October 30th, 2018, 6:01 pm

Phototransistor Filtered, LTR-4206E

IR LED 940nm, EL-IR204-A
Roomba and Neato Mods, come visit: http://www.vic7767.com/
User avatar
vic7767
Robot Master
 
Posts: 14735
Joined: January 14th, 2006, 8:31 pm
Location: Haughton Louisiana - USA

Re: Roomba 550 Cliff Sensors -- New Fix

Postby Sagrat » October 31st, 2018, 3:18 pm

Thanks vic7767 !

Is this the good one ?

Other question :

Can I unlock error 6 without replacement diode ? If yes, how can I do ?
Sagrat
 
Posts: 5
Joined: October 30th, 2018, 4:32 pm

Re: Roomba 550 Cliff Sensors -- New Fix

Postby vic7767 » October 31st, 2018, 4:27 pm

Easiest solution is to replace the complete sensor harness that will contain the cliff sensors and two bumper sensors. Can be found on ebay.
Roomba and Neato Mods, come visit: http://www.vic7767.com/
User avatar
vic7767
Robot Master
 
Posts: 14735
Joined: January 14th, 2006, 8:31 pm
Location: Haughton Louisiana - USA

PreviousNext

Return to iRobot Roomba and Scooba Chat

Who is online

Users browsing this forum: Chonlameth, im1kissfan, johnnyh, KeldBroe, Miles267, vic7767 and 334 guests