Adding A Rotating Side Brush to Neato XV's

[glnc222]

Possibilities for Adding a Rotating Side Brush to XV models

[edit] 2014: the 270rpm Robotzone micro gear motor is listed at ServoCity with the plastic case separate
http://www.servocity.com/html/270_rpm_m ... locks.html
but has been found also at another dealer:
http://www.karlssonrobotics.com/cart/Mi ... OgodaQQArw

Vorwerk used a custom specified motor.
A non-gear motor similar to the lidar drum motor is http://www.ebay.com.au/itm/390508885594. It may not be strong enough but not tested, and requires PWM digital control.

This hasn't been done yet, but some means appear available to explore.
For another version of adding a side brush, see thread "Neato Side Brush Mod" http://www.robotreviews.com/chat/viewto ... 20&t=16101

[6/17/12 -- working side brush design shown at end of this thread, using hex key drive shafts and Micro Gear Motor from RobotZone (servocity.com)]

The topic is suggested by comments under static side brushes, needing a separate listing
Wing Nut's

A soft rubber or silicon wheel turned by friction as the floor passes underneath could, through simple (plastic?) gearing, drive a lightweight revolving brush.

and OnTheWay's

A small brush with a flexible axis, hairs planted as "archimedes screw" and driven by friction wheel. The wheel(s) can be placed on below of bumper's front and its shaft can be as long as the bumper's itself, drives the sidebrush.

[5/15/2012 such helical brushes are made by Rubbermaid, 1/2 tube brush and over 1" frig coil brush, and many twisted wire pair brushes.]

I especially enjoy such out-of-the-box ideas. Parts suitable for such mechanical add ons are not familiar to me, so it would be interesting to see this tried.

Internal Motors
Side brush motors may be practical to add to XV models. The space used in VR's equipped with brushes is there and the motor could be connected to the main brush motor power lines, switched on and off already; this motor is 12v DC according to the label shown in a robotshop photo. For other voltages, the main motor lines can actuate a very small reed relay, if some other voltage power line can be found in the Neato (hackers have even tapped into the battery packs). 12v relay: http://www.amazon.com/12-VDC-Reed-RELay ... 047&sr=8-4 [oops -- one has no coil, must use http://www.allelectronics.com/make-a-st ... LAY/1.html ]
Robotshop has a number of motors including these:
http://www.robotshop.com/cytron-12v-130 ... motor.html
and
http://www.robotshop.com/6v-175rpm-17oz ... motor.html
(provides links to datasheets). The latter is the most likely candidate. Check Servocity's own site for useful fittings and competing prices.

External Motors
It is also possible to attach some very small motors outside the case, even with tape.
This tiny motor runs on 3V and can be powered by very small AAA cells.
http://www.amazon.com/Mini-Motor-Speed- ... pd_sim_t_4
Geared multiple ways, can go up to the 120RPM which might be good for such a brush. There seems to be a number of small gear motors among hobby supplies.

The half-inch thick robotshop motor listed above and shown below could also be an external motor, powered off the vacuum motor line.

Another way to control such motors is tapping the drive wheel motor connection, which can be reached without opening the main case -- just the bottom rear vacuum-and-wheel cover.

A lot of motors are listed at http://www.allelectronics.com/, but the 12v gear motor 140rpm is about three inches long, too big to fit internally.
There seem to be a lot of small motors even with gearing, so maybe something will show up that fits. Many seem to be too slow, and without gears too fast.

power problems
The main brush motor may be monitored for extra loads indicating stuck, so could be a problem.
The vacuum motor does not seem so monitored and is another power source; relay can trigger off other motors.
[added 5/15/21012]
USB requires an RPM spec to run the main brush, and a post mentions a VR-100 value of 1200, which makes a familiar sound on the floor with the aluminum beater. That brush at this rate uses 500ma.
(Incidentally, if "brushdisable" is entered via USB it remains so after disconnect, producing a brush-stuck error when cleaning).

space for side brush motor, approximate, 2" overhead

one seemingly fitting motor at robotshop.com 12v, 175rpm ServoCity (link in text)

bestmotor.jpg (3.61 KiB) Viewed 33660 times

This small motor shown could also be used externally being so thin.

As I don't think side brushes do much on carpets, am not pursuing further myself right now. Always interested in the possibilities.

[sotm]

Hi glnc222:

According to this post:

http://www.robotreviews.com/chat/viewto ... 05#p107605

I understand that the vacuum motor is monitored as well. Isn't it?

Best regards,
Sinner Of The Meadow

[glnc222]

the vacuum motor is monitored as well. Isn't it?

Tests under "electronic filter monitoring" http://www.robotreviews.com/chat/viewto ... 20&t=15981 show not much sensitivity of the vacuum motor, and there might not be much under the main brush either, would have to test. The "stall current" is fairly high on these motors relative to running load and it could also check for zero RPM more than current. The side brush load should be small relative to the load tapped. There's a fairly good chance it could work, and not extremely expensive to try compared to other aspects of the device. Always some risk with these things, up to personal taste.

The side brush should be powered from whichever source already has the highest load. [added 5/15/2012: the main brush motor found to use 500ma]. It's also only accessible inside the case. The vacuum motor already has a load in the right range, up to 1500ma with filter removed, and then only 90 percent speed used for cleaning -- goes up a lot more at 100 per cent.
USB has data on main brush voltage suggesting some complication in control there, as well, though can't tell what without testing.

Coupling of the side brush to the motor is best made with some slippage, supporting light torque only, allowing the motor to spin when the side brush itself gets jammed. A Roomba brush can be attached with a nylon spacer suitably adapted with set screws etc. Might be too wide. Not so hard to make a brush on a wood hub with all the brushy weatherstrips, hard and soft. Artist brushes?
Servocity has the perfect fitting for their motor:
http://www.servocity.com/html/lightweig ... b__3mm.htm
but shaft a bit short for internal. Allways the details.

It will be interesting to mount this little motor on the front of the bumper. Can't do that internally.
Having missed my Curiosity Anonymous meeting and had a relapse, have ordered a motor to be tested first with a separate 12v battery. Only if it cleans anything will it be worth wiring up the vacuum.
Decided better to test the 270 rpm MGB-100 instead of the 150 listed at robotshop -- available direct from Servocity, in a range of speeds. Even smaller, 1 inch square by 1/2 inch.

[5/16/12 -- can't mount on the bumper as drag will press it in. A large bracket needed to mount externally, as the brush must be close to the front intake, far from the chassis, similar to internal mounting. The side bumper would be partially blocked, and Neato following walls with 5/8" clearance may not leave enough room. Might squeak by at the most. A bracket can attach by right angle to the motor adding no more thickness.]

The needed brush mounting is shown in the VR-100 thread linked below. Judging from the photo the hub is 7/8" wide and the same radius beyond for the bristles. Geometry limits external mounts: when bristles short enough for the wall proximity, not long enough to reach the intake.


The hub sticks magnetically to remove when opening the brush guard. Looks like a nylon bushing or bearing mount in the case bottom, and motor mount screws.
Any kind of quick release could be a complicated fitting. A press-on fitting might be possible, but needs a D-shaft coupling. Whether a brush can attach without a specially shaped or cut away brush guard is an issue.

[glnc222]

External Sidebrush Test

A side brush is rigged externally to test performance of the new line of micro gear motors from Servocity.com remarkably small for the power. The one inch square MGB-100 rated 270rpm loadless is fixed to the Neato frame side -- not the bumper -- with a thin steel and thicker wood bar extending forward along the side bumper to the wall sensor -- steel because of a right angle useful to hold the motor. The over-size wood is just because the steel bracket was not long enough. The motor is powered from an external supply for testing, since it is not yet known whether worthwhile using.

The test brush is sliced from an inexpensive Rubbermaid bottle washing brush, a 3/4 inch plastic core with rings of ten pencil-size stiff nylon bristle bunches around the circle, angled slightly downwards, overall diameter 2.5 inches. In Ace hardware and http://www.amazon.com/Rubbermaid-RHFG6C ... 905&sr=8-1

temporary external side brush attachment

Current load with this brush 12v is 80ma on hard surfaces, 140ma on carpet, and up to 200ma crunched against a wall -- well within capacity of the vacuum motor power connection if used, which can run over 1.5 amps but is typically around 1.1 amps or less. Stall current is 1.6 amps with torque 1.5 inch-pounds; it is pretty hard to stall via this brush, hardly the final version for practical use. It looks a bit slow with the brush and I would go to the next higher 460 rpm gear box when used further. Hard to judge that factor by eye though; the VR-100 brush speed is unknown.
When connected to Neato power a fast-blow instrument fuse is recommended, given sensitive monitoring components, such as 450ma used in one digital multimeter.

A test on hard floor uses one level teaspoon of ground coffee distributed over the two inches against the wall.

level teaspoon of coffee in 2 inch wall strip

After Neato's passage with the spinning side brush, about half the coffee was vacuumed and half remained.

after pass with external side brush

coffee Neato left (out of initial level teaspoon)

coffee Neato collected (out of initial level teaspoon)

The geometry mentioned in the previous post surely plays a role, where the brush is too short to reach the front intake when mounted outside the case edge, but is limited in length by Neato following the wall only 5/8 inch away.
It is surprising any material was vacuumed at all.
The Neato also cannot follow the wall perfectly with the obstruction of this brush -- fairly stiff -- at the side, having to repeatedly dive into the wall seeking closer contact which is blocked. The entire 5/8 inch clearance is not even available to the brush because of necessary thickness of the hub to attach substantial bristle bunches and also attach the motor shaft, so fluffier brushes alone are not sufficient.

The brush was not perfectly mounted with some slant (the set screw was in one of the slanted bristle bunch holes). It is not worth perfecting because of the geometry mentioned. Might have reduced pickup some. It seemed to be on the floor in the front arc though.

Internal Mounting Factors

Just to obtain the geometry required for an effective side brush requires internal mounting to shift the brush shaft to the useful position between the side and the intake edge. It's hard to imagine any sort of belt or gear train outside to shift the shaft over underneath.

This Servocity motor has a short shaft less than 1/2 inch which cannot extend through all the plastic parts of the case. The frame itself is 1/8 inch thick around the brush guard clip, there is a 1/4 inch air gap between the case and the guard, and the guard is 3/32 inches thick. The shaft is 3mm but seems close enough to 1/8" to fit couplings that size; the only 3mm coupling found was for model boats, and 1" long (parts for Aquacraft Rio EP). The shaft fits #4 smooth spacers and bushings well, and does not fit threaded #6 spacers without some drilling out. The motor is short enough to be raised on spacers with a 1/2" coupler, shortening the shaft if needed. Threaded #6 mounting holes are on the shaft side.

When flanged nylon bushings are used, placing the flange on the outside of the case would allow removal for greasing the shaft.

The VR-100 uses a modified brush guard to fit the side brush and a quick-release hex socket fitting with magnetic adherence. The shaft extends only to the surface of the brush guard so the guard can be opened swinging on the hinges at the bottom (cannot be lifted straight up). So a large hole is needed in the brush guard for some thick fitting on the brush engaging the motor shaft. Hex head bolts go fairly small and might be found matching the socket in the VR-100 brush, in which case such a bolt could be coupled as the extension shaft, with details of threads to handle. All sorts of small magnets are common; a case for my Kindle uses coin size magnets for a closure.

[6/17/12 screw shaft couplings discussed here and following superseded by hex key shafts end of this thread]

An interesting attachment to the shaft would be a screw-on brush, with three turns or so. Because the wall sensor is on the left side (viewed from vacuum front) the shaft spins counter-clockwise, which then requires a reverse or left-hand thread screw fitting (not to be unscrewed by the shaft rotation). Unfortunate because common threaded binder post hollow-solid shaft pairs for documents, always #8 screw size, cannot be used. Reverse threads are used in rotating machinery, especially lawn mowers, for this very problem, but usually in large sizes 1/4 inch or more. But I did find this listing for "Hong Nor X3 3x20mm Left Hand Thread Screws" http://www.ebay.com/itm/Hong-Nor-X3-3x2 ... 5d3206cf6c
-- in model toy parts. This could be used as a shaft extension and the left-hand nut imbedded in the brush, or just self-thread the screw into material.

brush making

As far as fashioning brushes goes, epoxy paste in sticks makes a moldable, hard though somewhat brittle material and glue together. Brushes could be assembled in small diameter plastic pipe sections, or hollowed dowel pieces (allowing additional fitting shapes). Bristles can be obtained from cheaper door weatherstrips (some of which are expensive -- the length I suppose). Paint brush bristles, sometimes polyester, are also possible but might not be as rugged as nylon; brushing on not the same as brushing off... Paint brushes fairly expensive except for some particular throw-away ones.

Issues Remaining

A usable motor has been found but too many questions remain as to what brush and mounting to use. The best brush would be VR-100 brushes if they could be obtained as replacement parts, and reproduce the hex socket mount or adapt to something else. No VR parts are yet available, at least outside Germany; some sort of regional test marketing or whatever limits distribution at this time. But if those could be obtained it would seem to make a lot of difference, and they might show up at some point.

I prefer not cutting holes in the Neato case without knowing what's going in and whether generally useful, so I am not pursuing this further. There seem to be many users interested in side brushes more than myself, so it was interesting to see what the possibilities are. I would really like to at least have a good brush available first. I will be interested in whatever other users do on this feature. Perhaps the information collected here will allow others to pursue it more easily.

[eldadv]

Hi Glnc,

First of all, I like your work !

Second, I thought about simple and less technicle soloution for the sidebrush.
How about taking a small machanisem (mybe a toy hellicopter) that when the wheels rotating on the floor the rotter is also rotating. Put this machanisem at the buttom of the Neato (like another wheel at the font) so while the Neato is running on the floor, the small machanisem whell is running the instead to rotate hellicopter rotter it will rotate the side brush.

Hope I write it as clear as I can.


Another soloution is to use one of the following motors :

http://www.ebay.com/itm/A2204-28g-7-5A- ... 437wt_1110

http://www.ebay.com/itm/DC1V-2-5V-Motor ... 265wt_1344

http://www.alibaba.com/product-gs/43995 ... _Quad.html

the idea is to use micro motors (up to 11mm), mount them on the bottom of the Neato and connect to the end the sidebrush. the power cord can mount all the way up behide the front bumpers and there to put the battery (or to connect it somehow to the internal power supply)

What do you think ?
(I really dont know if any of my ideas are even possible )

[glnc222]

Thanks for your interest. Two other members are quoted atop the thread on a traction-driven brush -- maybe it's those old toy cars people had (or everyone is enamored of DaVinci?).

The problems I see is the Neato doesn't have much traction, and the brush has to rotate much faster than the drive wheels, requiring a torque hungry speedup gearing (try turning the shaft on a geared down motor). You can note, though, there are rollers in the brush guard already that Neato rests upon -- not that it helps. Also, any traction taken to drive something else will increase power drain on Neato's system, to drive the wheels harder, just as much as a separate motor. Drag on the traction could also be inconsistent with Neato's drive controls. Mechanical power transmission is space consuming and complex to build, the very reason electric motors revolutionized manufacture (my economist background is showing, sorry). That micro gear motor is pretty inexpensive as some of these hobby supplies go.

Personally I'm put off making anything with gears because I don't have a drill press to precisely make mountings, and I even suspect this requires some micrometer grade accuracy beyond carpentry tools anyway; all those assembly toys are punched by precision dies at a factory -- teeth have to mesh precisely. A special problem seems right angle gears which I don't see listed along with all the little spur gears available. Easy buying a thousand lot for manufacturing, but hobbies are something else. I'm not familiar with the model plane and boat craft supplies, though.

The fascinating thing with the Neato is that what I did not like about it seems to be fixable, and entertaining in the process.

Now if we could get the more equipped and mechanically expert, pedigreed engineers with Neato's to perfect this side brush add-on, the harried salespeople at Vorwerk can be left in peace.

[OntheWay]

glncc22, excellent work. I suspect brush utilized to be blamed for the remaining coffee. Also, high number of arms on the brush may cause coffee driven back to wall edge, instead of getting vacuumed. I believe this is the reason for 3-arm brushes with longer hairs are preferred on most units.

[glnc222]

Internal Fit of Micro Gear Motor

The Micro Gear Motor fits internally only with the shaft cutting into the edge of the brush clip frame.
Shaft out center of the motor side prevents reaching the corner exactly where the VR shaft is, and I would not be surprised if the VR case has slight differences around there it is so tight.

detail of space for adding motor

shaft on side edge of guard clip frame

shaft on bottom edge of guard clip frame

Looks pretty tricky. Would like to see someone else with cases to spare play with it, I'm keeping only one.
Worst case abandon the clip on that side and screw the guard closed nearby. (I'm covering my ears now -- go ahead).

Better Brush Fittings

Better prospects here. Reverse thread screws may work better in the brush instead of on the shaft, with the socket on the motor side.
The 3/8" motor shaft looks long enough to pass the 1/8" frame bottom and attach a coupling on the other side; it just won't pass out through the brush guard. The short coupling could be a spacer threaded on the inside left handed with a tap tool, for a left hand screw through the brush (or self-thread a screw?). There is 3/8 inch from the bottom of the frame to just outside the brush guard surface, you could go to 1/2"; only a shallow inside thread is needed for the brush. Add a set screw for the shaft. Aluminum spacers easily worked.

But with the internal fit problem above, the coupling would partly cover bottom of the brush clip, preventing opening the guard -- unless a bit of it could be removed, it's down to the tiniest dimensions. Could be only some press-fit solution would work. You have to really want a side brush.
I did point out elsewhere the clip could be replaced by a quick-release twist key-lock.
The second photo, horizontal position, at least puts the shaft over the back of the clip and not it's latching front part. It might still work with half the back removed.

Tapco USA lists left hand taps and dies in small sizes if the reverse thread bolts are not available. This would simplify brush-making and attachment.
Never tapped anything myself, though (except funds) -- is this really possible? The Euro does remind of a left-hand screw these days.
The same method for the VR hex head mount; hex spacers come in steel for the magnet, problem is specific size. But the socket could be filled in with epoxy and a left hand screw put through instead.

[glnc222]

3-arm brushes with longer hairs are preferred on most units.

Thankyou OnTheWay. Glad you noticed the pattern with coffee edges around a center carved out by the narrow brush.
The only way to fit a longer brush given the close wall clearance is to move the brush hub over where VR's have it, internally mounting the motor -- and even the narrow brush does not work properly against the wall.

Incidentally some coffee was thrown way ahead of the vacuum across the floor, it being so slick; insignificant volume.

The VR brush reminds me of a kind of pencil eraser which had a thin rubber disk about an inch wide, in a metal strap on which a broad brush was attached at the edge similar in size and shape. Don't know if still made.

[glnc222]

More Left Threaded Screws

The 3mm screws previously listed are for toy car shock absorbers and probably too thin to get a spacer into which they could be threaded. But another car, Losi8, uses 5-40 .125 in. (1/8) screws according to tables.

Chicago dealer: http://www.amainhobbies.com/product_inf ... d-Screw-10

UK Ebay: http://www.ebay.com/itm/Losi-8IGHT-5-40 ... 20c189c0bd

If nothing else they could self-thread into nylon spacers. A #4 spacer for the motor shaft could be drilled out 7/64 on one end and threaded with these screws put in the brush hub.
Come to think of it, the 3mm 1in. spacer from model boats might just be ready to thread, if not too wide for 5-40 screws already; has set screws installed. Cut into two fittings. Nominal "3mm" sometimes means with a clearance making actually wider.

Now will you sacrifice your brush guard clip? Dirt is piling up on the edges...

[glnc222]

simplified guard clip replacement
[edited 5/21/12 for improved presentation]
Photo shows the micro gear motor shaft position best available and the room required for a coupler to the side brush. Some of the brush guard fastening clip would have to be removed and the clip maybe unusable. Besides replacing the clip with a screw or key twist lock fastening through the guard surface, an easier method may use the side of the wheel well underneath the back of the brush guard.

A small screw can be put in the back edge of the brush guard to which a hook in the wheel well fits over. This could be wire, as from a large paper clip, fastened with a screw to the well side through a grommet providing some flex, or an additional screw near the top of the well can secure the cross piece of the hook, flexing the wire under this screw. The wire will have some flex itself.

Another method is a quarter inch bar stanchion fastened to the wheel well side with a small swiveling strip screwed into the top, to swing over the edge of the brush guard. A shallow notch for the strip can support some snap fitting into the closed position, again with maybe some rubber washer.

guard clip frame
The guard clip has a raised frame around it, in the photo with unseen part covered by the guard, to which the guard fits helping secure its position, which needs to be preserved as much as possible, though part must be removed. Then just pulling the edge down into the wheel well, holding the guard pressed onto the frame cover is all that is needed. One side of the guard clip presses on the inside of this frame and the opposite side of the guard across the clip presses on the outside of the frame. Close inspection of details to whatever is done is warranted.

sealed ball bearing
[edit] The shaft needs a bearing or bushing in the case bottom to relieve strain on the delicate small motor when the brush hits obstacles. I used a sealed ball bearing 3mm inside shaft diamete, mounted with epoxy paste on the outside, but F_Robot used a small plate, plastic or metal, on the inside on the bottom of the case.
Many small bearings available online.

The motor shaft likely protrudes through the existing guard clip hole. This may need shaping to add a 3mm sealed ball bearing (3-8-3) protecting the side brush motor and shaft from fiber gathering problems. Thickness occupies the 1/8" frame cover thickness so requires no more shaft length. The whole clip hole can be filled in with the ball bearing installed, the clip removed and replaced by the attachment above. The motor does not need a bearing but the contamination problem seems to warrant it. Rotating shafts exposed to dirt seem to act like textile spinning wheels concentrating whatever material passes by. (never use flanged or open ball bearings which get destroyed by accumulating dirt -- "sealed" bearings only, a standard specification; Amazon.com and distributors on the web)

[glnc222]

eraser brushes

A source of side brushes are pencil shaped erasers with attached brushes. Professional art Faber Castell, some expensive; this one
http://www.amazon.com/Shindigz-Eraser-B ... 874&sr=8-2
specifies length 6.5 inches, which from the pictures gives a useful brush length close to an inch.

Successor to the old disk erasers with attached brushes.

[piokrza]

glnc222, you are one true *CRAZY-ROBO-GEEK* & I like it!!

BTW I bear in the hope that one day someone (maybe one forum member?) will construct a self-cleaning/emptying station for my "rooby"

[glnc222]

I suggested that under "future neato" I think, for products. But it seems a prototype could be constructed using a regular vacuum for the heavy machinery, controlled by a relay on the extra robot mounted at the charging base.
Some sort of door would be put in the top of the dust bin, and a vacuum tube lowered by a servo onto it. Always sounds easier than the details turn out.

Am adding a finished guard clip design to the previous post on this above.

[vic7767]

I wonder if the old style side brush from the first gen Roomba would fit ?

[glnc222]

Roomba brushes are basically much wider I think, for a different layout than the narrow spread between wheels, front intake, and nearby wall forming Neato's Bermuda triangle. The bristle parts might be used, but the real problem is price around $15 a brush, when not already perfect. Thought I saw a door brush strip at Ace around $6, with 36 inches of bristles to make Neato brushes for sale on ebay...

Future plans are now to check whether brush mounting can be easily made with left hand screws and spacers, awaiting parts. If that works will go further into assembling a brush: the new word there being bottlecaps...

[glnc222]

possible better mount position

One possibility not examined was mounting the motor into the top cover able to fit over some obstructions in the photos above. I did not have the cover completely off to test this fit when taking pictures, so am not sure available. A long shaft would then have to be coupled and a bearing to hold the shaft in the case bottom. This just might get the shaft in the position Vorwerk used.

The bumber sensor circuit board is higher than the support stanchion and the motor cannot fit above that board, but will go over the support. Motor is 1" by 1/2".

I would look at this before removing any part of the brush guard clip and its frame part of the case.

possible top of case motor mounting

(To mark motor positions in the top of the case one way would be brace the motor in position at bottom and place a paper sleeve over the motor top with double stick tape facing up. When the cover is lowered onto the case the sleeve will stick to the cover as it is removed. Then the motor can be taped to the cover for checking fitting. With a drill bit or nail coupled as the shaft a mark might be put in the case bottom, and subsequently into the brush guard. A theory only.)

[zardoz]

Isn't there the risk of covering the cliff sensor by the rotating brush?

[glnc222]

Good question. This is how it fits on the VR-100 relative to the cliff sensor same as XV's.

[glnc222]

shaft hardware found

#4, 1/4" O.D. nylon spacer may serve to mount shafts. Self-threads 5-40 left thread screws (wider than 4-40's); 4-40 set screw self-threads in 3/32 side hole; drills out to snug fit 1/8" shafts. Cheaper than ready made steel couplers.
1/8" hitch pin clip wire cheaper than high strength robotics parts from robotshop.
4-40 screws and 3mm shafts fit well unaltered. Appears no need for hard to find, expensive 3mm fittings to match the motor or steel fittings for such low torque application.

#4 nylon spacer with self-threaded 5-40 screw, 4-40 set screw, 1/8" shaft

hand-made side brush

3/4" nominal (ID, 7/8" OD wider) vinyl furntiture leg tip used for hub; easily cut with knife blade. Smooth convex bottom for floors. Comes in many sizes. Inexpensive.
Brushes arranged in overlap pattern to raise ends, angle bristles downwards.

paper funnel collects and shapes paint brush bristles to glue ends

overlay pattern angles brushes downwards

thin brush with furniture tip hub

The overlay pattern makes a brush too thick for the 3/8" ground clearance of the vacuum. So a 7/8" hub used as cover over a 3/4", perfect fit. Radial brushes instead of tangential.
Base nibbled in at slots. Overhang provides down slant with top slotted just the right depth.
Screws together to compress. Flexible glue instead of hard epoxy, e.g. SealAll.
Final installation of 5-40 left hand button cap mounting screw inside the case bottom for thinness, hole in bottom for the cap's 5/64" hex wrench. Screw length adjusted for various clearances and mounting dimensions.

This is the economy model in polyester. The deluxe nylon version comes in soft, firm, and extra firm bristles (for Neato's with back problems).

[added 6/12/2012] Enclosing the brush in plastic parts wastes space, 1/16 of top and bottom; it needs to be as thin as possible. The plastic is best used only as a mold into which the case is made entirely of epoxy paste, later painted.

The key principle is how the hard edge presses down on the bristles slightly under tension to bend at an angle towards the floor. If any plastic is used it would be only a ring around the outside for this pressure, but I think epoxy alone will do that. Could be some benefit for surface strength. The 7/8" furniture tip encloses a 3/4" interior -- they all stack concentrically -- and could be used as a mold, pressing the bundles down into a shallow dish shape. Coating with grease may prevent epoxy adhesion. Slow 60 minute curing "marine" epoxy paste available besides the common 5 minute setting, to allow careful work.

shaft locking feature
A small hole or notch is needed in the side of the brush guard, horizontally, for a wire or hex wrench to lock the mounting shaft under the brush when extra leverage is needed to unscrew the brush if the screw threads freeze up, tricky since underneath the wide brush hub. Even without a set screw the mount is holed for this contingency.

side brush RPM analysis

The vacuum travels around one foot per second and Neato geometry limits the radius of the brush to little more than an inch. Four brush vanes sweep the 90 degree forward arc four times per revolution. At 180 RPM or 3 per second, 12 sweeps of the forward arc occur per second corresponding to 1 sweep per travel forward the length of the brush radius. I would like at least two such sweeps. The next size up micro gear motor is 270, then 460 RPM, a bit more than 360 rpm for two sweeps. Higher up gives lower torque.

[correction 6/9/12 on Sercovity Micro Gear Motors (Robotzone brand)]
Initial subjective impression of incorrect speeds proved wrong with later accurate measurement; the motors appear labeled correctly at the factory, and tend to be slightly higher speed than nominal rating, within spec.