roomba 530 spinning and stops with 9 beeps

[vic7767]

What else in roomba could be affected by temperature so consistently?

Consideration might be given to the solder joints in the electrical path on the printed circuit board inside the black plastic assembly bumper sensor. I don't know the behaviour of the individual diodes used in the bumper sensor and how they might behave to temperature. Possibly getting a spray can of "cold squirt" or other similar spray might help identify the component at fault.

[Gordon]

... Possibly getting a spray can of "cold squirt" or other similar spray might help identify the component at fault.

That's exactly what I was going to suggest to John about an hour ago as I started googling for long-extinct trade names of products, such as 'CircuitCool', or 'KoolIt' (neither of which might be correct). Those were Freon-12 based coolants, and probably disappeared when the Freon-12 = R-12 refrigerant got phased out of large volume production.

Interestingly, it took me three trips to MGChemicals before I found its test-coolant product -- this one:

Also of interest, vic7767, I searched for "cold squirt", but did not notice any hits! Things do come & go, faster than we know!

Back to John's search, he basically needs to arm himself with some such aerosol coolant (possibly, even the "Dust-It" product, sold at Staples, may give enough cooling), and a heat-gun (maybe a hair-dryer will do), then strip off the top-cover from the 5XX to give access to the Bump-switch modules, and to some of the main elex PWB asm.

John, if you have not gone inside your 5XX as yet, you might find Dave's instructions for doing so, very handy. See his page:
http://www.pottsland.com/roomba/Roomba_ ... ensor.html
...and follow steps down to about Step-16, then re-assemble the Bumper, battery, bottom-cover, etc., to allow the 5XX to run in Clean-mode. (I may be overlooking something, since I have not actually done quite this to my own 535 -- vic7767 will tell us what is missing).

With the 5XX running about in a small area, try heating/cooling each Bump-SW module to see if you can either initiate a nine-beeps fault, or delay the fault at will.

[mfortuna]

Also called "freeze spray" or "tech in a can"

[vic7767]

Like you Gordon, I was running from a dusty old memory when I watched my dad troubleshoot TV vertical hold problems back in the day of selenium stacks and 5U4 power tubes. The URL to disassembly is a good procedure but I will put a plug here. I have a 5XX disassembly DVD (home brew) that is available to order on my web site.

[ga2607]

Regarding the apparently heat-related nine-beep problem, I was the one who first posted this question, but I got busy last week and wasn't able to follow the discussion. I'm glad to see someone else has found that the problem also seems to be related to temperature inside the Roomba. One question raised in a previous post is whether I had made a side-by-side comparison of the bump sensor IREDs:

Doing an "A", then "B" viewing with a camera which has active exposure control can fool the investigator. Auto exposure must be set OFF, or the two IREDs need to be moved close together so they both appear in the same frame -- then, differential brightness can be discerned.

Just to clarify, I did move the two IREDs side by side so I could see them both in the camera viewer simultaneously, and they both appeared equally bright.

One other point that supports the heat buildup idea is that when I had everything disassembled this summer, i.e., cover off, bump sensors disassembled, etc., I put things back together just enough that I could run the Roomba using the wireless command center to steer, but with the circuit board open and bump sensors open. I ran it much longer than the run times I'd been getting when it was fully closed up, and never did get it to stop and give nine beeps.

I like the idea of testing using the hair dryer and cold spray. When I get some free time I'll try that. Right now it's cold enough that I'm getting sufficient run times to keep things relatively clean around here.

[richnorto]

I fixed it!!! I have a 560 with the bumper problem which I have cleaned out, verified led strength, everything several times. I decided to take it apart one last time to see if I could recognize a cold solder joint. Umm, most of them are dull so I guess my whole board and small boards could go at any time... Anyway, I remembered an early post where someone had found bent back transmitters inside their mechanical bumpers - mine looked perfectly lined up. I took a thin screwdriver and slightly pushed all 4 forward a bit. Not expecting any success, I put it back together and started it. It's now been running 25 minutes, happing bouncing off things and no weird backward dance. I'm not going to delude myself that this is a permanent fix - they may be degrading - but for the time being, It's working and when it quits, I will know the parts that need to be replaced!!! Thank you everyone for all the helpful advise offered on these boards!

[Gordon]

I fixed it!!! ... I took a thin screwdriver and slightly pushed all 4 forward a bit. ...

Rather than maximizing beam coupling, via the alignment attempts, its more likely you disturbed one or more of the poorly connected electro-optical device pin(s), and made it connect better to the circuit board.

Next time you get the error report, go in there and (flux each joint) then re-flow those dull solder fillets. Do all of them, while the iron is hot, and you have access! Hint: When soldering the pads, do one pad on one device, then shift to another device (allowing the first to cool), and come back later to finish the second pad. None of these semiconductors enjoy being overheated!

[richnorto]

Gordon, you are absolutely right! It ran for 45 min. and started the same behavior. At least I know where to work on the problem now - I don't have to buy a new sensor array etc... Is there a soldering for idiots page explaining this flux etc.? I did look up cold joints and, as I mentioned, most looked that way. I had bad luck on what I thought was an easy job - attaching a battery fuse - and am a little frustrated with it as I usually manage to make things work! Sorry to wear you out on my Roomba/electrical issues, but I don't know anyone with this type knowledge and on this forum, I have the help of you and several others with all the knowledge I don't have. I appreciate all the support!!!
Thank you, Laurie

[Gordon]

... Is there a soldering for idiots page explaining this flux etc.?...

I did find a beginner's page, years ago, but have lost the link. I was not overly pleased with the instructions given, but, I would have to say they were better than nothing! I'll do a little googling, to see what I come up with.

Here's such a link (although not what I had seen in the past):
http://blog.makezine.com/archive/2007/0 ... ake_v.html

[richnorto]

Thanks for your help with the solder tips! I finally pulled the Roomba apart and resoldered the points on the mechanical bumper where the sensors connect. I can report at least preliminary success! The 560 has been running strong so far with no back circles. That's a difficult job! Shaking hands, solder that wants to either stick to the gun or glop over...I was afraid I was going to fry the mini board. I'm so happy it seems to want to work now!

[Gordon]

... I finally pulled the Roomba apart and resoldered the points on the mechanical bumper where the sensors connect. I can report at least preliminary success!

Yahooo!! I've been wondering what you've been up to!

... That's a difficult job! Shaking hands, ...

Mind over matter is required. Emulate the cool headed OR surgeon at work!

... solder that wants to either stick to the gun or glop over...

That suggests the pad, and wire-lead through it, was not hot enough. When too much solder is on the tip, pull the iron out of the work zone and give it a quick whip, to shake off all that solder (you should be working in an area where you can sling molten solder on the table, or floor -- without concern), then do a quick wipe on the sponge and try to get the soldering done.

I was afraid I was going to fry the mini board.

Rather than harming the PCB substrate, the more likely item to damage is the pad, or trace to it. That sort of damage is called "pad lifting", and happens when the copper foil is heated so much, that it de-bonds from the board, and any mechanical scrubbing that may be happening at the same time simply slides the copper out of position. It is a mess, but, can often times be repaired well enough to function.

I'm so happy it seems to want to work now!

Keep fingers crossed!

[kriegerj]

Nick,

I also had the same issue with by roomba 500. After spending 5 days checking the mechanical side, it was clear that the issue was somewhere else.

You post help a lot to go straight to the modification. I changed the pull-up (24K, what I had available) and it reduced the unblocked voltage to 0.42 V which was enough to solve the problem.

Thanks again.

[fourth-horse]

Good info here! Now I just have to get the time to take the sucker apart and try one of the fixes suggested here.

Someone was asking how to confirm that you have the stuck bumper. Someone actually found a service manual in a Google cache and was able to get the built-in test instructions (see http://www.robotreviews.com/chat/viewto ... =1&t=12100).
One test shows how each bumper is responding by lighting up indicators. Here's how to get it started:

• Turn off Roomba by holding down Clean button
  Hold down Clean and Dock while pressing Spot six times.
  You'll know your Roomba is in test mode because it makes its own special sound and flashes all the lights.
  Press Dock twice if you have schedule buttons (day, hour, etc.), once if not.
  Now you're in bumper test mode. Spot should light for left bumper, Dock for right bumper.
  There are a lot more tests you can do, but you can just hold down Clean again to turn it off.

Unfortunately, my Dock button was stuck on, which is why I have to think about disassembling my robot friend.

[richnorto]

Oddly enough I just emailed iRobot with a 3rd 500 series Roomba going down with a 9 beep! Luckily, this one is under warranty! The weird thing is they had the 9 beep on the answer page as a wheel drop sensor isssue...? That is something I hadn't considered.

[daze99]

I suppose the hard part to this repair is knowing which sensor is bad. I just fixed my 580 after reading this thread. YEAA!!! Cured of the 9 beeps disease.

The bad part of this repair it that you pretty much have to completely dissemble the robot. 500 series is not has "modular" as iRobot had hoped. I'm not sure how common the 9 beeps problem is but since it is such a pain you would expect iRobot would make the components involved modular like the wheels, cleaning bin, etc.

Once I got the 580 taken apart (yes you have to pull on the face plate until it feels like it will break) I looked over the bumper sensor bank and thought. "It's got to be one of the mechanical sensors" since the optical sensors were pretty clean from my previous compressed air cleaning. I took apart the one that felt like it might be sticking and it looked pretty clean inside. I put it back together and discovered the rubber grommet seemed to be the cause of the lack of rebound of the sensor arm. I decided to remove the rubber grommets from the Bumper Actuating Arm/Optical Sensor set. I know the grommets were put there to keep dust and hair out of the sensor housing so I improvised with masking tape to cover the cracks in the sensor housings.

Re-assembly was quite a pain; a word of advice. Take a real good look at the sensor bank position before you take the under top panel off. It will fall out of position when you take everything apart and it is not easy to tell how it fits back together.

Once it got it back together after a few cycles of removing the top panel and then putting it back (to get the sensor bank in proper position) the robot worked without issue. I really believe the rubber grommets on the Bumper Actuating Arms are the most likely cause of this 9 beeps issue. They bind the arm with age.

Cheers and happy repair, I hope me telling my experience helps you with your's

-Jeff Wahaus-

[vic7767]

Thanks for the additional info Jeff. Now that it is known how to place the 5XX model Roomba into the Factory Built In Tests it has become very easy to determine which Bumper Sensor is failing. The rubber grommet may be an additional contributor to the 9 beep error but is certainly not the only cause.

[daze99]

Now that I'm about 4 days wiser I realize the problem was not the rubber grommets at all. Even though I had removed the grommets and increased the bumper spring force (by bending out front the springs a bit) the 9 beeps issue came back after the first cleaning cycle. After the Roomba recharged and I attempted a second cleaning cycle the issue came back with the same sensor.

Disassembly goes a lot faster the second time around. When I opened up the offending bumper sensor and pulled out the little circuit board that has the optical sensor pair I noticed it was sitting slightly crooked in the plastic holder that surrounds the board. It had slipped out of the tiny plastic tab that is supposed to hold it in place. This plastic holder also contains a vertical slit through which the sensors "look at each other". A slight misalignment causes the slit to block one sensor from the other. To repair I decided to hot-glue the little sensor board in place to keep it aligned. Additionally I used an exacto blade to shave the edges of the plastic slit to make it wider and less subject to misalignment. I estimate I increased the slit width by 50%. During reassembly I decided to put back on the rubber grommets since I no longer believed them to be the primary cause of the issue.

I've now run the Roomba through 5 cleaning/recharge cycles successfully and the 9 beeps issue seems to be fixed for good.

I expect a good number of faulty bumber sensors could be repaired in a similar manor instead of being replaced. I would bet that the majority of issues with this sensor are of a mechanical nature and not due to a failure of the electronic components.

-Jeff Wahaus-

[ga2607]

daze99, on December 23, 2009, wrote:

... To repair I decided to hot-glue the little sensor board in place to keep it aligned. Additionally I used an exacto blade to shave the edges of the plastic slit to make it wider and less subject to misalignment. I estimate I increased the slit width by 50%...I've now run the Roomba through 5 cleaning/recharge cycles successfully and the 9 beeps issue seems to be fixed for good...I would bet that the majority of issues with this sensor are of a mechanical nature and not due to a failure of the electronic components.

I'll be curious to see if this fix continues to work. I tried the same thing a while back to solve my 9-beep problem, but the problem returned. Admittedly, I wasn't as aggressive in the shaving, maybe increasing the width only 10%. I was worried that doing more would adversely affect the Roomba's ability to respond to collisions. Have you found any difference there?

My current fix is to set the Roomba outside (temps around freezing) before I use it, then I can get half-hour runs before the nine beeps. If I take it right off the charger (Roomba warm) I'm lucky to get one minute.

If the problem's mechanical, I wouldn't think the temperature would make that big a difference. I'm leaning toward an electrical problem. (BTW, I've checked iRED intensity and that doesn't seem to be the problem.) I haven't tried re-soldering the connections yet, but that's in my plans. Someone in a previous post also mentioned replacing a component that controls sensitivity to voltage output of the receiving sensor. That also seems like a good possibility, but I don't have the electrical knowledge to know what that component is or where it's located. If someone could post information enabling those of us who are electrically-challenged to make that replacement, I'd appreciate it very much.

[daze99]

an electrical problem. (BTW, I've checked iRED intensity and that doesn't seem to be the problem.) I haven't tried re-soldering the connections yet, but that's in my plans. Someone in a previous post also mentioned replacing a component that controls sensitivity to voltage output of the receiving sensor. That also seems like a good possibility

The best solution for a sensor level problem may be the software, they must store the expected threshold's in a table somewhere. Unless the sensor's output is binary. detect/no-detect If its an 8-bit level then a tweak to the sensor reading value may fix it. I don't know if access to the software would make such a change impossible.

I spent some time lining up the two sensors and pushed them a little closer together by bending them but keeping them as upright as possible. Basically the overall idea is to improve the detector's chance it will see it's companion, closer together and a larger slit will give a stronger signal.

My changes do not seem to have effected the "soft bump", the bumper still seems to be very sensitive and the Roomba is not hitting the walls any harder than before.

[vic7767]

The Roomba software is encrypted and cannot be accessed. The bumper sensor is binary (off or on). The Light Touch IR sensors are located behind an opaque band in the bumper and are not part of the bump sensor.