Last modified:
The thrust of this page is to show the futility in thinking it may be economically feasible to replace Scooba or Roomba gears with parts from any source other than iRobot. But, having said that, it is possible to envision a small subset of robot owners that either have, or will find the resources needed to produce simulations of the needed gears from bar-stock. Those individuals will be faced with either adapting similar gears to conform to the robot's requirements, or to machining the needed gears from raw-stock. Both of those approaches involve machine-shop work, and some mechanical-design. When those efforts must be paid for out of pocket, the cost will be more than is reasonable to bear.
Finding plug-and-play replacement gears, made by a company other than iRobot, is highly unlikely. Scooba's gears, and Roomba's too, are all specially designed by, and manufactured for iRobot. Although it is not too difficult to do, plastic gear-wheels having the desired number of teeth, size of tooth, and face-width can be found in commercial stock, but, it will be a rare case in which some adaptive machine-work is not required to make use of such 'stock' gears in an iRobot floor-cleaner. If an owner cannot obtain 'free' shop-services, the paid-for services will be too much to bear. There will be no exact, or near, matches of cluster gears to be found in commercial stock; there are just too many variables to accommodate.
All of the plastic, motor-speed-reduction-gears in the form of two gears on the same axis, molded together, have some complicating feature that will drive up the cost of fabricating replacements for them. That is because the clustered gears have no axial separation between them. Molded cluster-gears can only be replicated (from bar-stock) by use of only one gear-cutting machine type; and that is a "gear-shaper". When using such a machine, the cutter travels from the outer end of a tooth, along the length of the tooth, and then needs space to kick off the removed chip, or swarf, at the end of the stroke. That is a simplified description of how a "gear-shaper" works. To create such chip-relief in one of these cluster gears, a cast-design that does not need that relief, could be problematic. Some of the cluster-gears have an overly long pinion, and for that type a chip-relief groove may be cut into the small-gear where it joins the larger gear(and where its normal mating gear-teeth do not engage). However, others have no excess face-width on either gear, so the relief-groove would have to be shared by making both gears narrower than the original -- resulting in a weakened gear-train.
Replicating internal-tooth gears, such as Scooba's Wheel-gear is also difficult, because of machining the internal teeth and the need for a specially cut relief-groove at the inner tooth-ends. A gear-shaper is required to cut those internal teeth. There could be a tooling-charge, if a special cutter must be made for the job.
Various physical measurements must be taken, from remnants of the duff gear and from mating gears, to define important dimensions that will be needed by the machinist, however, there will be indeterminate data that the 'backyard mechanic' is not capable of retrieving. Examples are "gear-material", and "tooth-clearance". You know it is a plastic gear, but what specific plastic is it? You cannot know whether running clearance (backlash) was established by an over-size cutter, or by shifting shaft centers; nor can the nominal clearance magnitude be easily determined.
These opening paragraphs emphasize the cost-drivers. At this point, it may be of value to review a cost analysis for a specific Roomba gear, followed by listing, or discussing these topics:
During Q1 of 2006 several members / guests of two Roomba boards were seeking a replacement gear for their Discovery Roombas. The failed gear was one of the cluster-gears in the Brush-Deck Motor's speed-reducer gear-head. Figure 1 shows the gear -- two of them are used in the same gear-box (to aid in appreciating their physical size, they are about half an inch long).
Figure 1. Discovery's Brush-Deck Motor Speed-Reduction Cluster-Gears
The owners were pursuing all replacement possibilities, and one was to explore the cost of having a small lot of the gears custom machined. The following information, approximate, in some instances, was submitted via an on-line request-for-quotation process, offered by Stock Drive Products, Inc.:
"This is replacement gear-cluster in a speed-reduction gear train used in a robotic floor cleaning appliance. Consider this RFQ as 'informational', and for a number of parts ranging between ten and 50.
Approximately five working days later, the following quotation was
received:
"Part – Cluster – 0.5/10T/18T {S/B "...10T/17T"}
A follow-up inquiry determined: a) that 35 parts would have to be ordered to pull the unit price down to $20, and b) the non-recurring charge would be for a custom-made 2-mm-diameter mandrel. Upon amortizing the non-recurring cost over 35-pieces, they would then cost about $25 each, FOB!
The question then becomes: What are the chances of finding and coordinating with 35, or more, Discovery-owners whom are in need of this particular gear, AND would also be interested in spending about $30 to get one?
That answer just has to be "close to zero"! But, "how might that be improved?", is a question that needy owners living outside of the USA should be asking. Perhaps the only way to do that would be to locate a gear-producer with lower rates. Once a possible machine-shop has been located, enough gear-data must be presented in your RFQ to the shop to ensure: a) the shop understands the work to be done, and b) you stand a fair chance of getting a part you can use. Let's take a look at the elements in such a list.
Here are some data you will have to submit in your RFQ for a custom-made gear. We will repeat the two-gear-cluster requirements given in the above example costing exercise, but now include more accurate, and more complete data -- to the extent possible (this form could be tailored for any other spur-gear):
NOTES:
[1] All items should be shown on the face of a drawing, except for the
Quantity. An RFQ submitted on-line, may use the list format given above, but
questions should be expected.
[2] The metric-Module, "m", was computed by measuring the outside
diameter (~outrD) of the motor-pinion (mating gears should also be subjected to
this process, so that several results may be averaged). The value is calculated
using that diameter and the number of teeth on that pinion:
That module# plus the number of teeth on a gear, tells the shop what cutter
they will need. They may also calculate the largest gear-diameter so they can
define the O.D.s of the gear blank(s), and choose the rod-stock diameter.
[3] As discussed earlier, a chip-relief groove is required to support the manner
by which the teeth of the smaller gear will be formed. The groove-width can be
defined as a "reference-dimension", i.e., one without tolerance, by deducting
the sum of the face-widths from the cluster-gear's OAL (over-all length). The
groove-diameter may be determined by deducting two times the tooth-depth from
the O.D. of the small gear. Since the original gear faces were 6-mm wide, this
groove is infringing on the desired widths. Not good, but necessary.
[4] The exact "material" used to cast any of iRobot's gears is unknown by the public. We must basically guess at likely polymer candidates, or peel back sleeves and do some engineering work. In the end, there may be few choices from which to select, to yield both a likely good performer and one that machines well. The polyamide / acetal compound, (Delrin®), is one the few plastics that machines well; however, polycarbonate should be considered too. An engineer at Stock Drive Products claimed the company can 'gear-shape' the machinable Nylon® compounds (Nylon 6/6, and 6/10 are listed in SDP's gear-design handbook). Yet, if the example gear, above, were to be a Nylon part, it should not be meshed with the motor-pinion gear, because the brass pinion may run hotter than 100ºC and Nylon is claimed to heat-distort when reaching temperatures as low as 63°C. For that first-driven-gear position, a polycarbonate or polyamide gear would seem a better choice.
Let us repeat some previously covered disassembly work, but add a little information about the gears at each step. It can be useful to name each gear, so lets use numbers for names starting with number one for the motor-shaft pinion gear, and ending at gear #8, which is the one connected to a Wheel.
Thus, it is the paired-gears in figure one that are most likely to have problems, if any, and will be most expensive for anyone to replicate.
Figure 2. Scooba's Wheel-Motor Speed-Reduction Gear-set
All of the plastic gears have some complicating feature that will drive up the cost of custom-fabricated gears. For gears #2 & #3, #4 & #5, and #6 & #7, it is because they are tightly clustered on their common 'shafts'. The wheel-gear, #8, is difficult to machine because of its internal teeth. Fortunately, one might hazard a guess that this gear, due to its slow speed of operation, and large tooth size, would be less likely to fail than any of the other (plastic) gears in this gear-train.
Not enough information was gathered from this set of gears to fully define them, nor is a "full definition" even possible! But, if any mechanically inclined Scooba owner has any of these gears fail (while lacking warranty coverage) s/he will inquire about the feasibility of finding, or making a replacement part. The above data would give a 'leg up' toward investigating the costs involved to custom manufacture such gears, but much more detail will be needed to get even a rough cost. However, Scooba is in its infancy, and there is no hint any of these gears are failing, so additional definition work is not yet needed.
Scooba's brush spins at a fairly high rate, so we find only a single use of compound-gearing in its speed-reduction gear-train. On the one hand, that means there is only one cluster-gear to give as great a replacement-headache as the one discussed earlier, but on the other hand, there are two spur-gears that stand a better chance of being replaced by an owner; an owner capable of doing quality lathe work. One may even say the one cluster-gear stands some chance of being replaced, because it may be possible to join two commercially available spur gears. There appears to be enough pinion core-space to apply some form of keying or riveting between the two gears. Look at the gears.
As can be seen the left side of Figure 2, the output-shaft-gear (name it "#5") and the idler-gear, #4, that drives it are simple spur-gears. It might be possible to procure either gear-wheel having pre-formed teeth of matching Module and tooth-count; but, a purchased gear would still need some machine-work applied to transform it into a replacement gear. As a bare minimum, its bore-size would require adjustment, and preparation of a new (metal) shaft, or bushing, would be needed for either of those two 'final' gears. If an individual had to pay for that shop labor, the job might well cost nearly as much as a replacement Scooba!
Figure 3. Scooba's Brush-Motor's Speed-Reduction Gearing
In the middle of the image, we see the lone compound-gear, #3-pinion with #2-gear. But unlike most of the other cluster-gears we have see, there is a better possibility for constructing a composite, compound-gear from two commercial spur gears. The reason is, the gear diameters are large enough to make it appear there is adequate core-space available in #2, to permit mechanically locking the two gears together. But, as just reviewed, both gear-wheels would need a specific amount of re-work, a connecting shaft and keying parts would have to be custom machined. The shop labor -- if one has to pay for it -- could run twice that for a single-gear adaptation.
Here too, we back away from further consideration of replacing gears, simply because it is too soon to be seeing which gears will fail, if any, in the Scooba gear-set.
The final set of gears to be given consideration for replacement is the set that is enclosed in the gear-case at the right (Roomba's "right") ends of Roomba's Main Brushes. There are no "cluster-gears" in this set, they are all simple, spur-gears; which tends to make replacement more tenable. Figure 4 shows the case opened (and degreased). It seems fair to say that failure of any gear in this set is less due to tooth-loss than it is to severe wear at places other than teeth, per se. Wear on the inner-hubs, or square-sockets within those hubs, of the two brush-drive gears, can make brush-drive inoperable, but it also can mean there is a worn gear that can be re-worked and put back into service; thus a new gear need not be procured!
Figure 4. Main-Brushes Drive Gearing
As before, and for talking purposes, let us use Roger's gear-numbering convention to label each of these six gears. Starting at far right, in the figure, gear#1 is driven at the reduced motor-speed rate. Gears #2 & #3 are equal size idler gears, used to shift power transmission over to where the lower, and slightly larger gear, gear#4 -- which drives the rubber-brush -- can be reached. Idler #3 also drives the largest idler-gear, gear#5, which is above and to its left; and then gear#5 drives the bristle-brush's drive gear, gear#6.
We have found that Stock Drive Products actually stocks molded Delrin® gears, having identical metric-Module (0.7), tooth-counts, and face widths (6-mm) as the brush-drive gears! The down-side is their hub-geometries and bore-sizes do not support 'plug and play'. Thus, we again find ourselves in the machine-shop, trimming the hub-widths, adjusting bore-diameters, making special shafts and hubs with square sockets, to adapt these stock gears to run in a Roomba gear box. For anyone with sufficient interest in doing that sort of work, the SDP Catalog number is (in generic format): A_1M_2MYZ070nn, (where "nn" is the number of teeth). Here are the required tooth-counts:
SDP (go to their Product List via the Quick Finder) will ship parts to the UK.
If you wish to return to the main (Scooba Technical Comments) document, click Main
Otherwise you may return by pressing the 'Backspace' key.
This page is currently maintained by G. Plews