Brushes not spinning

rovermaster · Joined: 07 Apr 2005 Posts: 43 Location: Sacramento, CA

I'm hoping one of the more knowledgable users here has an answer to this one...

The brushes on my Discovery stopped spinning. The new bot has been RMA'd with iRobot, but I can't help tinkering with the little bugger (now that he's out of warranty).

I have gone into diagnostic mode and when I get to 2 long beeps and 1 short, nothing happens. Try that same diagnostic on my Roomba Red and the brush spins, as it should.

I have resoldered the connector to the main board thinking it was a bad solder connection there, still no spinning bushes in diagnostic mode (or any other mode). I pulled the motor out and bench tested it directly connected to the battery. It spun like it was new. Having seen other posts (I can't remember where) that the motor wears out and has little power, I tried installing the brushes. With the brushes installed and my hand offering resistance, I again connected the brush motor directly to the battery, and again the brushes spun like they were brand new.

I reinstalled everything back to the way it came from iRobot and ran it diagnostics again, still no movement on the brushes.

I am left to think that there is a problem on the main board.

My question to you knowledgable users out there is, what could be causing this problem? Have I overlooked something in my troubleshooting? Do I need to pull the main board and start tracing out circuits?

Your feedback is greatly appreciated.

Gordon · Posted: April 21, 2005, 11:54 am Post subject: Re: Brushes not spinning

rovermaster wrote: ...The brushes on my Discovery stopped spinning.... I have gone into diagnostic mode and when I get to 2 long beeps and 1 short, nothing happens.... I pulled the motor out and bench tested it directly connected to the battery. It spun like it was new. Having seen other posts ...that the motor wears out and has little power, I tried installing the brushes. ..., and... the brushes spun like they were brand new.
...
I am left to think that there is a problem on the main board.
... Have I overlooked something in my troubleshooting? No. You have shown everything to be good all the way back to the switch that supplies power to the motor. Most likely that "switch" -- a power transistor -- has a fault; but, the result is the same if it is not being 'commanded' ON -- so there could be a problem ahead of the xstr.
Quote: Do I need to pull the main board and start tracing out circuits?
You could try that. I have just examined that sort of task, and find it to be formidable. The PWB is multilayer, traces are hidden under various components, connectors, and daughter-boards. You might get there, but I envision having to remove those obstacles in order to deconvolve the as-built PWB-connections into a schematic.

EDIT: That work got done in 2007 for many sub-systems. See Schematic_6, for the Brush-Driver circuit.

There may be one chance for guidance. Looking at the PWB, I count nine power xstrs. Eight of them belong to the wheel-motor bridges. Hmmm, don't you suppose the one all by itself might be for the brush-motor?

rovermaster · Joined: 07 Apr 2005 Posts: 43 Location: Sacramento, CA

Gordon,

Thanks for the validation of my troubleshooting.

You mentioned the power xstrs. Are they easy to trace to each motor or are you going by what you have been able to trace out already? If you have drawings, pictures or schematics, I would really appreciate getting a look at them.

Thanks again.

Gordon · Posted: April 21, 2005, 9:06 pm Post subject:

rovermaster wrote: Gordon,...You mentioned the power xstrs. Are they easy to trace to each motor...(?) They are obvious (and have been discussed at roombacommunity). The bridge-quad for the LH motor is at Roomba's left end of the PWB and next to the 6-contact connector that feeds the LH motor. The bridge-quad that runs the RH motor is located just a little right of center, along the upper edge of the PWB; and its six-contact connector is the next one to the right.
Quote: ... or are you going by what you have been able to trace out already? I have traced nothing, so far. Doing so is not high on my list of things to do. Quote: If you have drawings, pictures or schematics, I would really appreciate getting a look at them. With respect to the Main PWB, the photos I have will give you no more information than what you have personally witnessed when you did the re-flow solder job. I have not photographed the forward side of a Disco PWB. {060221-edit: Deleted reference to a TO-220 "power-transistor", since the device is a 5-volt regulator (7805)}

The brush motor draws a bit less than an amp when running w/o load. {U11 is the power-control device for the BD-motor. "U11" is a low-power MOSFET. It is located near the LH-end of the PWB, just below J4.}

BTW, there have been a number of failed BD-motors due to one of more windings opening up. There are five coils that make up the armature winding. When one coil fails, the motor can still 'run'. Depending on where the armature stopped when the motor was shut off, upon re-powering, it may start by itself and run, or it may need a manual assist to get running; whereafter it develops enough torque to run itself and the unloaded brush pair. If you try to get work out of it, the motor will stall.

Such a forced stall can easily cause a second winding to open. When that happens, the motor cannot run.

rovermaster · Joined: 07 Apr 2005 Posts: 43 Location: Sacramento, CA

Thanks for the help Gordon. I finally got the courage to pull the board out of my disco and found a tiny, 6 leg SMT IC fried (and good). Since I don't have the ability to solder at that fine a level (and the "better half won't let me try the toaster oven soldering method), I have resigned myself to looking for a new board or just using this disco for parts.

Thanks again!

PS - I have a pic of the IC at the following link http://www.ghlane.com/Roomba/Fried_IC.jpg

The IC is just under the brush motor connector and the left wheel connector and right above the 2 big blue resistors.

Gordon · Posted: May 1, 2005, 10:16 pm Post subject:

Thanks for the pix, rovermaster. That is a mighty small device at U11; I can understand your hesitation at doing the R&R work. I thought I would give you some additional data.

Since another owner (over at the roomba-users board) suffered an identical failure, I've been doing a bit of sleuthing on that U11 device and its circuit, and want to tell you about my findings. Unfortunately there is no way to know exactly what part was placed there by iRobot, but I found a part that looks like a good fit. Rather than load the whole nine yards into a post, I uploaded it as a file into:
http://groups.yahoo.com/group/roomba-users/files/

If you go there and look near the bottom of the list for: "U11-discussion", you can read all about it. Like I say, its not definitive, but it may still be helpful.

Regards;
---Gordon
============================

Guest · Posted: May 2, 2005, 7:16 am Post subject:

Gordon, GREAT write-up!

In looking at the other case you referenced there, my disco's IC fried on the exact pin (#4, Ibelieve). As far as getting the number on the chip, I went back after reading your article and found my part number to differ from the other case quite a bit. The part I see is BDC8M, with the 8 underscored and the M barely readable through the burnt area.

I have done some www searching to find a way to inexpensively solder SMT ICs. The link below will take you there.

http://www.seattlerobotics.org/encoder/200006/oven_art.htm

Based on the description of how to solder SMTs in that link, I may soon be trying to convince the wife that we need to try some alternate means of preparing our meals. :D

rovermaster · Joined: 07 Apr 2005 Posts: 43 Location: Sacramento, CA

Oops, forgot to log in. That was me above.

Gordon · Posted: May 2, 2005, 12:27 pm Post subject:

rovermaster wrote: Gordon, GREAT write-up! Thanks -- it does give a possibility, but we're not yet at the end of the line!
Perhaps, if Case (from iRobot) monitors this board, he could help by confiming whether the MOSFET should be P or N-type.
Quote: ...The part I see is BDC8M, with the 8 underscored and the M barely readable through the burnt area. I now believe it will be rare for two owners to find the same code on that part, even though it seems reasonable that a couple characters should be constant!
Quote: I have done some www searching to find a way to inexpensively solder SMT ICs. Right! Thanks for that link, I found it to be interesting; but, I must tell you that I see several problems to overcome when doing such soldering by *general* heating. They can be summarized into a) items that may melt (jackets on electrolytic-caps & connector housings, and b) far-side components that may drop off during re-melt heating.

If I had to do the job... a) R15 would be temporarily removed; b) leads would be snipped from the bad part, then pads cleaned up; c) use a dot of Krazy-glue to position the new FET on the spot; and d) use a one-mm dia. tip to heat each connection. Solder-paste might work to advantage (if it can be applied that way); else, big-ole 0.015" wire solder might have to do.

One other caution about heating the chip: While looking at current-sensor ICs (the Maxim line), one of their data sheets specifically said hand-soldering their SOT-23 package is not to be done! Perhaps there is too great a thermal transient, or gradient across the device, when heated by a soldering tip(?).

I should also confess that the only chip I have soldered to a PWB is a DPAK MOSFET (.25-in. SQ, and big enough to put six SOT-23's on its back!). That part's DS also said not to mount it with an iron. Fortunately, the new part is still performing OK.

Good luck with your adventure!

rovermaster · Joined: 07 Apr 2005 Posts: 43 Location: Sacramento, CA

Gordon,

Your assistance has been invaluable! I have just received a "broken" Roomba Red today and swapped the board from that into the Discovery that had the bad chip. Note, it is far easier to replace board with power and serial ports, as desoldering the connection from the board itself is an excercise in patience.

Since there was nothing to lose on this I desoldered the chip from the old board. If you are interested, I managed to get some close-up pix of the chip and the board with the chip removed.

http://www.ghlane.com/Roomba/BCD8M.jpg
http://www.ghlane.com/Roomba/u11.jpg

rovermaster · Joined: 07 Apr 2005 Posts: 43 Location: Sacramento, CA

Sorry, try the chip here...

http://www.ghlane.com/Roomba/BDC8M.jpg

Gordon · Posted: May 4, 2005, 10:57 am Post subject:

I took a look at the pix. One thing I am curious about is where does that trace comming in from the left (under the "U11" label) connect (under the U11 part)? I can't make it out in your pix.

Here is a shot of that area (just in case others have any interest in this topic):
http://photos.groups.yahoo.com/group/roomba-users/vwp?.dir=/&.dnm=U11-area_crop%27d.jpg&.src=gr&.view=t&.hires=t

rovermaster · Joined: 07 Apr 2005 Posts: 43 Location: Sacramento, CA

Hey Gordon! Sorry fo rthe long delay in getting back to you. I had to put up all the Roombas for a while while dealing with other projects. I am back now. To answer your question, if it hasn't been answered already, the trace under the "U11" label does not appear to go anywhere. I checked and double checked, it appears to deadend just before the outline of the IC. It looks very similar to what it does to the right of the "J3" label (just above the "U11" label).

Now for the update... I had swwapped in a good board from a Roomba Red with a circle dance problem. I opted not to fix the circle dancer and just pull it's board to put into my Discovery. An hour after running with the new board, the brushes stopped spinning. I just opened the discovery up tonight to find the IC on the new board burnt in the same place as the old AND all the plastic around the brush motor has been melted.

So, new conclusion, based on my results so far...

If you have an IC for the brush motor burn up, replace the brush motor in addition to the main board. Either that or just fix your parts robot instead of gutting it and having two semi-usable boards.

Gordon · Posted: June 1, 2005, 2:11 pm Post subject:

Well, rovermaster, you tell a very sad tale. Whenever a part that has been providing good service fails, one must ask why -- what caused it to fail? I had mentioned something like that at the bottom of the U11-file (link is about 6-clicks up-thread). There are a few things that could kill U11 by permitting it to disspate too much power. Obvioulsy there was something wrong with your brush-motor; yet, it seems like there should have been some relief given by the over-current protective circuit (which I *think* is present). If that function *is* present, it may have also been damaged when U11 first failed(?).
You now have a PWB that you could disassemble, as required, to provide visibility and access, to trace portions of the circuit that might be of interest; and you could then use it as a soldering test-bed when you decide to attempt a repair of the 2nd PWB assembly. So all is not a total loss! :-)
Thanks for the info on that blind trace.

geoffrey · Joined: 05 Jun 2006 Posts: 3

OK - so I got 2 of these Discoveries, 1 stopped charging, so I robot sent a warranty replacement.
Now the brush on the second stopped spinning, and the brush motor is fine when I connect it to a battery (and the gears worn hardly at all) so it's a main board problem. The thing moves around though. Anyway - I now got 2 with dead PWB's :-(

Given that the PWB board looks a complete PITA to fix - I am considering using the side brush motor output off the main board to swiitch a mosfet connected directly to the battery to power the main brush - add a little circuit board of my own, and there's plenty of space for it. I don't doubt that an extra power load to the side brush output would cause that to fail as well seeing how these things are put together - however, if it is only to switch a FET, then I don't see a problem

Can anyone think of a reason why this wouldn't work (or tried it with disaster) - does the thing have current monitoring on the main brush that would cause a problem ? - or run at a different voltage to the battery output ? :!:

if all else fails, then I guess I'll have to use the one with no brush to charge the battery, then transfer the battery to the one in the cuppboard that doesn't charge.

Thanks for your thoughts.

Gordon · Posted: June 5, 2006, 10:43 pm Post subject: Re: Roomba brush problem

geoffrey wrote: ... I am considering using the side brush motor output off the main board to swiitch a mosfet connected directly to the battery to power the main brush - add a little circuit board of my own, ... I .. doubt that an extra power load to the side brush output would cause that to fail as well ..., if it is only to switch a FET, then I don't see a problem. True, since the gate biasing resistance can be relatively large.
Quote: Can anyone think of a reason why this wouldn't work...? Functionally, it sounds OK, but you would be missing at least one, possibly two levels of over-current protection. If the motor ran too hot -- a very likely scenario -- fried the insulation on its armature windings, shorting the coils, etc., there would be no sensing of the problem, and it could lead to a fire.
Before you get too carried away with that add-on switch, you should examine your PWB to see if it is one that contains an alternate set of pads, pads for a TO-220 size U11! Here are two shots of the U11 area in one such board, (U11 is located about 39-mm to the right of the left-edge of the card, and down from the upper edge by 13-mm):

Large components were removed for the pic at right; thus you can see the Drain-copper going to the middle TO-220 pad, and the Gate-trace to the lower TO-220 pad, hence, the Source is the top-pad. R14 & R15 are in // and serve as the current-shunt in the source to GND ckt. You should try to make use of that current-sensing circuit, IMHO.
BTW, this PWB came out of a 4210 with mfg-date = 041113E.

geoffrey · Joined: 05 Jun 2006 Posts: 3

Thanks a lot for posting the Pics. U11 is indeed fried, and a TO-220 pad is present, :D though it took me some time to see it. Unlike your melted U11, mine has lost a corner off the IC, about 0.5mm , and there is a tiny (less than 1mm) scorch mark on the board - so cutting out the SMT and replacing with a TO-220 is probably the simplest way to go to fix this puppy. I can't say that current limitation consisting of exploding mosfets inspires much confidence in these things - nor can I see why they even bother with current controll if the MOSFET is going to explode first.

With respect to replacing U11, what parts has anyone used :?:
If your source and drain are the right way round, it must be a p-channel mosfet - but enhancement :?: depletion seems concieveable as well, and being chineese, any level of wierdness is possible. And this is not the way would have designed this to say the least.

In any case, I am putting a fuse in series with this motor, with a much beefier U11 than was there before.

Thanks in advance for your suggestions.

Gordon · Posted: June 7, 2006, 1:07 pm Post subject:

geoffrey wrote: ...I can't say that current limitation consisting of exploding mosfets inspires much confidence in these things - nor can I see why they even bother with current controll if the MOSFET is going to explode first. Amen, brother!
Quote: With respect to replacing U11, what parts has anyone used :?:... See what was said in this thread about half way down.

geoffrey · Joined: 05 Jun 2006 Posts: 3

Thanks for your advice, finially had a little time, yanked the old U11 out with pliers, got some STP60-NE's from Newark, soldered one in and this roomba popped straight back to life. :D Only one thing concerns me.
I ran it around the floor without the lid on, and in the space of about 2 minutes, the main brush motor gets so hot that I can barely touch it :!: - I dread to thing how hot the little brass gear is getting if the case is that hot.
Has anyone got any idea if this is "normal" :?:
It would seem that the plastic would be melting in about 10 mins at this rate :!: :!:

If as I suspect, this is a very bad omen, has anyone got any ideas for a replacement - preferably non chineese motor that will fit in :?:

Gordon · Posted: July 7, 2006, 10:11 am Post subject:

geoffrey wrote: ...yanked the old U11 out..., got ...STP60-NE..., in and this roomba popped straight back to life. :D Only one thing concerns me.... the main brush motor gets so hot that I can barely touch it...
Has anyone got any idea if this is "normal" :?:... Not normal. Sorry, I have only a few minutes to devote to this post... I would suggest you measure that motor's current-draw and compare it to these data:
"factory-test 11 main-brush...
(baseline-current-ok?) mA -92 min -202 max ...
(brush-current-ok?) mA -287 min -599 max ..."
{obtained from Perquin}. Power it direct from the battery, and load the brushes with hand-pressure. I don't know what the above two ranges mean, but the latter one shows very dangerous currents, and are much greater than what I measured for a Roomba Pro-E's brush-motor, years ago (and that is nominally the same motor design).
If you must replace the motor, you will be largely constrained to use the same motor P/N. What you must think about is how to improve heat-flow out of the motor's housing, and out of that closed (mostly-closed) chamber in which it has to function. In addition, you may have to manually monitor cleaning periods, and cut them to as short as 20 to 30 minutes. I have to do that.