Note, prerequisite to the operations discussed in this document, Scooba's Vac/Blower Assembly will have been dismounted from Scooba's lower-chassis, disconnected from the main_elex PWB, and the fore & aft housing 'halves' split apart so the Impeller may be worked on. A good primer for that work is provided in vic7767's ground-breaking work. Of course, you will see that 'breaking' of things, other than *ground*, had to be done!
This document relates the futility anyone faces upon setting out to remove a Scooba's Impeller from its motor-shaft, and with attendant intent of also replacing the Impeller so it may again serve its original purpose. The structure of the Impeller is such that an installation force does not harm it, but, application of a removal force necessarily stresses the Impeller's structure in a different manner, and in such a way that, as force increases, the Impeller uses that force to destroy itself beyond repair.
Before going too far into this story, it may be of use for the reader to know a little more about the Impeller than can be learned by viewing images of a fully assembled Vac/Blower Assembly. BTW, this is information that a disassembler can't be very certain about when viewing the Impeller at that level of assembly, but may only suspect. When the Blower Housing covers the Impeller, only its cylindrical inlet with nine curved vane-edges, converging on the central tapered-hub, can be seen, and much of the Impeller remains hidden. To some extent, I'm putting the end of this "destruction" story at the story's beginning, however, doing so permits me to attach names to parts that must be discussed; and, having done so, I expect readers will benefit.
There are two plastic castings which are joined together to form the finished Impeller. When viewing the assembled Impeller at the Blower's inlet, you see them both. I will refer to the "Hub-Casting" as the part that interfaces with the motor-shaft, and which forms the back-wall (relative to *Impeller-space*) of the Impeller; or the forward-portion, considered in *Scooba-space*. Then, there is a "Vanes-Casting" that forms the aft-portion (Scooba's 'aft') of the Impeller. It will soon be seen that those two castings are joined via edge-bonding along the forward edges of the vanes.
The following picture illustrates those parts, just named. At right is the dislodged Vanes-Casting, and at left the Hub-Casting, which refused to dismount. This is the nature of the destruction that must be expected to happen as one pushes against the motor-shaft's end, while tugging (reacting that force) uniformly (uniform, to the extent feasible) around the Impeller's far-side rim:
Figure 1. Hub-Casting Remains with Motor, Vanes-Casting de-Bonds
The edge-bonds between the two casting break apart!
Here's a look at the custom-tool that provided the push/pull action:
Figure 2. Special, Scooba-Impeller Puller Tooling
The seven arms (the 8th had to be removed to permit fitting the puller around the Impeller) were cut to a thickness of ~ one-millimeter over most of their length, but at the open end of the tool, they become L-shaped so a two-millimeter wide ledge can slip behind the Impeller-rim, and accommodate application of a pulling force to the back of the Impeller. To apply the tool, the pusher-screw / pin is backed out, and the seven arms are flexed outward a little bit as they pass around the periphery of the Impeller. Then the pusher screw may be run inward until it contacts the end of the motor-shaft.
When using the puller, the hose-clamp (left side in image) is secured tightly around the lower skirt of the white portion of the puller, and that peripheral constraint prevents any of the seven puller-arms from being forced away from the Impeller. However, "springing" goes on, but it is not deflection of the tool's parts. Here is what happens as the pusher-screw advances.
You may study Figure 1 to see traces of where the vanes' edges fractured and separated from the Hub-Casting. Note also that the edge-bonds were not carried onto the steeply sloped surface of the central hub, and in effect, weakening the structure. It is at that inner region where stress reached the breaking-point, then propagated outward to the rim. The bonded Impeller is a very stiff structure, and adequately strong, even though most of its plastic wall thicknesses are only one-millimeter. It is the width of the vanes, some four-millimeters and increasing toward the center, that achieves high stiffness. It is also that stiffness which transmits the orthogonal rim-force (due to the puller) inward to cause initiation of peeling the bond-line at a vane end next to the hub.
To get the answer to that question, I resorted to 'dentistry'.
I used a ball-burr in a Dremel motor to route a slot through the Hub-Casting's hub-to-back-wall region; and, kept digging through plastic until I struck metal. Additional routing of plastic revealed the entire length of the metal part, but measurement from the hub's aft extremity, back in to the motor-shaft, indicated the metal part was not the motor-shaft!
The routed-slot was then lengthened, and made radially deeper, until the motor-shaft was encountered, and its end located. So, *two* metal parts were found! One is the steel motor-shaft, and the first-found item is a brass hub-adapter that is cast / bonded into the core of the Hub-Casting.
Well! That makes it clear, as to why that Impeller is not likely to ever be dismounted by pulling on its rim while pushing on the motor-shaft: A more severe interference-fit between the brass-insert and steel motor-shaft may be obtained, than between plastic, and the motor-shaft! If only the pulling force could be reacted at the motor-side of that brass insert, it could be slipped off the shaft. Too bad, that's not economically feasible. I could tell you how an old Swiss watchmaker might go about such work, but, you may as well spend your money on a Rolex wristwatch!
Well gosh, if if its so difficult to move that brass part along the motor-shaft, how did it, along with the assembled two castings, get mounted on the motor-shaft in the first place (one might ask)? ANS: By applying a pushing force to the aft end of the hub, all plastic between that force-application zone and the brass bushing is placed in *compression*, hence, there is is no over-stressing the plastic. It is a safe thing to do. In opposition, a *rim-pulling force* puts much of the material between the rim and hub into tension (the fraction in compression is not over-stressed), and it is tensile over-stress that breaks the vane-bonds (whose joints appear to be weaker than the parent plastic casting compound).
Why was it important to remove the Impeller? To provide information, is the 'general' answer. Certainly there is no requirement to dismount the Impeller as an Impeller *cleaning* step. The main impetus for getting the Impeller off, was to gain access to the hidden motor-shaft bearing, to permit re-lubricating it.
This entire Vac/Blower disassembly (to permit cleaning, and bearing lubrication) was triggered by A THREAD at RoombaReview. One of the other concerns expressed in that thread pertains to loud noises being emitted by the blower; noises that would have to come from contact between the moving Impeller and stationary housing; or, from a dry shaft-bearing. Hence, those were the interests that caused this futile disassembly attempt.
Since I was then faced with a useless Hub-Casting, still secured to the motor-shaft, I decided to make the best of it by forging ahead--for the sake of 'science'! To that end, I continued doing a little more routing of the open slot. I tried to drive the brass-insert toward the motor, just to break free "sticktion" between shaft and insert. Something moved (evidenced by the Hub-Casting dragging on the housing part), but it was not the Hub-Casting that moved! It was probably the axial limit-stop within the motor assembly. Eventually, two flat-blade screw-drivers were used as pry-bars to break the plastic casting away from the brass-insert. Having done that, we may see, in this next figure:
Figure 3. Hub-Casting Busted Away From Brass-Insert
This is not to say that all impellers cannot be pulled off their motor shafts; but, it is statistically fair to claim that most attempts will fail. The major reason for that is the variability of degree of interference between the diameter of the hole in the brass part vs. the diameter of the motor-shaft. There will simply be more motor/impeller matings with "nominal" to "maximum" degrees of interference between the brass-bushing and steel-shaft than there are of minimum levels of interference.
One of the other concerns expressed by Scooba owners is loud noise being emitted by the blower; noise that could come from the moving Impeller's rubbing contact with the stationary Blower-Inlet Housing. Close-up views of the air-gap (labyrinth-type) seals will be shown, since these zones must have the smallest running clearances which are economically feasible.
First, let us get oriented by looking at a cross-sectioned Impeller and its surrounding Blower-Inlet Housing.
Figure 4. Section Through Blower-Inlet Housing and Impeller
In this sectional view, one may begin to appreciate how many zones on the Impeller's surface speed by adjacent surface areas of the stationary Blower Housing! To maximize blower efficiency, i.e., the ability to move air through duct-work leading to the blower's intake through the blower and out of its exhaust side, lateral, air-leakage paths that can introduce additional air into the blower's intake must be minimized.
We can identify a primary labyrinth-seal and a (possible) secondary labyrinth seal. They are knife and fork configurations in which a two-prong fork envelopes a knife-blade. The primary seal has the blue two-prong fork with black knife-blade, and the secondary-seal is inverse. Three enlarged views provide a little better view of these knife & fork zones.
Figure 5. Interior Two-Thirds of Primary Labyrinth Seal
The 'primary' seal begins at the rear-most (in Scooba-space) edge of the thin-walled (blue) cylindrical inlet to the Impeller. Vacuuming / blowing air is normally rushing through the interior of this cylinder, and we don't want leakage-air getting into the stream by squeezing past that rear edge. By design, the assembled blower has a minimum gap at that point. It can also be seen that the radial gap between the outer diameter of the thin-walled cylinder and its adjacent black-fork is held small. You may judge the gap as some number less than one millimeter by noting the impeller-wall thickness is nominally 1 mm.
Figure 6. Exterior 1/3rd of Primary Seal
This portion completes the primary seal. The 'fork-prong' is shorter, but air-gaps remain small. Bear in mind this "knife & fork" description is only of a cross-section through these parts--the "knife" is a continuous circular rib, going full-circle inside the Blower Housing. Same for the blue "fork-prong".
Figure 7. Secondary Labyrinth Seal
Its not certain whether iRobot considers this construct to be a labyrinth seal, however, I think it provides that function, but not with high efficiency. What we have here is a black 'fork' of two continuous ribs which straddle a circular row of balancing pegs. There's a little more than 100 of those pegs, which assist with the active final-balancing of the Impeller / Motor-Rotor assembly. When the balancing test reveals a heavy zone within the rotating mass, the operator knows mass needs to be removed in that region. One or more appropriately chosen pegs are then snipped off, and the test iterated.
Surely, the many remaining pegs stir up great turmoil within the black trough when the Blower Assembly does its work, and that turbulence stands to impede leakage-air movement transverse to that 'seal'.
Now it is time for more discouraging thoughts. Say a person's Scooba is emitting awful scraping sounds, and which have been identified as coming from the Blower Assembly. One might think all one must do is to remove the Blower-Inlet Housing, inspect the housing and Impeller surfaces for rub-marks, grind off the rub-marks from the stationary member to increase running clearance(s), then put it all back together. Doesn't that sound okee-doakey? ANS: Yes it does, but there can be one major gottcha, that may not be apparent! And that is: Permanent distortion of the Blower-Inlet Housing; i.e., dimensional distortion of sufficient magnitude to change the spatial relationship between it and the Impeller to such a degree that if the Blower-Inlet Housing was immediately replaced the rubbing of part would be either improved or worsened. Separating the two main housing parts is not detailed in this page, but experience has shown the operation of breaking their bonded joint to be very traumatic to those castings.
While all of that basically concludes the squeaky-clean interests, one more task is being appended to this document. Namely, providing some motor details; in which a final photo will be shown, and which also happens to give a better view of the brass-insert!
Finally, the motor mounting screws could be accessed, and removed to allow separation of the Blower's motor from its forward housing casting.
The following image reveals several motor-features, some brass-insert details, and also provides another clue as to why it will be more difficult to move that brass-insert after the blower has been in service for some months:
Figure 8. Blower-Motor Profile and Mfr's ID
Here are some things to talk about:
The bottom line is: Do not attempt this work at home! Or, anywhere else, for that matter! Seriously, anyone having intent to safely remove Scooba's Impeller from its motor-shaft must be prepared to fail.
NOTES:
1. Roomba® and Scooba® are registered iRobot product names.
2. Appreciation goes out to RR-member 'glo69', who graciously donated Scooba sub-assemblies to 'Research', a donation which then made this study possible.