Electromate Arm Subsystem Requirements

This page is intended to serve as an organizational space for the arm subsystem requirements requested by Gleeson from Electromate for motor/gearbox selection.

Gleeson's List of Requirements:

1. Torque and speed requirements for the existing system per axis
2. New backlash requirement and other updated requirements per axis
3. ‘Ideal’ gearbox style of what you saw from Harmonic Drive’s product line
4. Estimated budget per gearbox

Requirement 1 Response

Requirement 2 Response 

Option 1

Requirement 3 Response

We are open to implement any suggestions you provide, however . These are some potential options we looked at available on Electromate. To compromise on cost, we might want to try to integrate one harmonic gearbox on A1, then planetary gears on A2-A6. A1 has a low load but requires the highest amount of precision, where as A2-A6 require higher torques but have greater permissible backlash. We can look into implementing 6 harmonic drives in our arm within our budget, however we should also plan to be realistic. Here are some ideas of the types of products we have looked at as potential options.

Motor Selection Ideas

EC-flat motors offer good torque and have low cost at Maxon. These seem like a good selection choice for us, but we are also open to other suggestions/motor types. 

• Maxon EC-Flat Series

Harmonic Gearbox Ideas (to be used on A1)

These are all lightweight options which can provide an output torque required for A1. 

• CSF-2UH-LW
• CSG-2UH-LW
• CSF-2XH

Note: these may also be good options for A2/A3 as harmonic drives offer high ratios on a single stage. If we can afford these gears for A2/A3, it would be excellent

Planetary Gearboxes (A2-A6 options)

We've looked at these planetary gears as they posses low weight, low backlash and potentially have low costs. Some of these products, such as the XTRUE planetary gears will not be able to provide sufficient output torques for some of the joints (namely requirements for A2/A3). However, we are open to adding second stage reduction via belt drives to improve torque capabilities in our design.

• XTRUE gearheads from Thomson Micron (these could probably be more applicable for A5/A6)
• HPN gearheads from harmonic drive (good for A2-A6)
• Maxon Planetary Gearheads

Requirement 4 Response

How does 400 CAD per gearbox sound? For the arm, we will probably want 2-3 gearbox spares (per recommendation of Cory I think...) . If we buy 8 gearboxes, that will cost $3200 and comes to 15% of our total rover budget... seems reasonable to me I guess. I think overall the team has around $11,000 funds right now, so we don't want to eat up our actual parts budget too much as we still need to buy motor and drivetrain stuff. 

Backlash Requirement Reasoning/Justification

Creating backlash requirements for the arm is difficult. To create strong backlash requirements, we would need to know what arm configurations are most commonly used to complete comp objectives and how bad backlash actually is. Keep in mind, the goal is to have our end effector sit within a  "goal precision area" of 4mm: (this was discussed and determined in a past meeting, but is a precision based on ~half the width of a keyboard key)

In the past, we've discussed backlash as if it directly translates to the precision of the end effector. However, this really isn't the case as you need to consider the loading and repeatability of the arm positioning to determine how the backlash actually impacts our operations. For example, consider the arm configuration shown below:

Consider the backlash of A2. Let's say that the solid orange line represents the intended position of the the arm, however due to backlash it deflected to some degree and now lies on the dotted line. For examples sake, lets say it deflected 10mm. This sounds bad, but is it really? The vectors for arm and payload weight always possess some component of force that is directly perpendicular to the intended position of the arm, thus the arm's position will always stay within the "lower range" of it's backlash. Sure, the arm may be off by 10mm but if we have good enough encoders we can adjust our positioning to move the arm 10mm upwards, which should accurately position the end effector. The arm will always deflect in the direction of the arm and payload weights.

General Background Assumptions

• I'm going to assume that A2 - A6 wil utilize the same style of gearbox on each of the joint. Reusing gearbox styles will really simplify our design and make our joint assembly much easier than using a different type of gearbox at every joint. Thisreduces our BOM, and may give us lower prices when ordering as we buy higher volumes of the same part. A1 is excluded from this assumption, and this will be explained more later on. 
• I'm also going to assume that our construction materials for the arm linkages are perfectly stiff - obviously they deflect in real life but we should optimize for this in a future arm iteration

A2/A3 Requirements

A2 and A3 are always in "fixed configurations" - while they may translate, the axis of both joints will never rotate. The direction of loading on the arm will cause the backlash at each of these joints to be retained to their lower bounds, as discussed above. In other words - backlash requirements for this joint shouldn't be too critical as positioning should be very repeatable. It is very difficult to determine the exact requirements needed for these joints without having a good understanding of what our workspace and common arm operation configurations will look. We could take a "worst case scenario approach" where we see the maximum possible deflection when A2, A3 and A5 are parallel, and work backwards to see the backlash requirements at each of the three joints as shown below:

This results in a desired backlash of ~0.1º or 6 arcmin. While this is possible with high precision planetary gears or elliptical gearing, I think selecting this as a backlash requirement over constrains our design space. I highly doubt full horizontal will ever be used in our arm configuration, as the drivetrain can be used to adjust the horizontal spacing of the robot. Analyzing the arm this way will create really high backlash requirements and over design the our arm. I don't think over designing isn't really a good approach for gearbox selection, as it is one of the most expensive OTS components we buy. Buying an expensive and heavy high precision gearbox that we replace with a less expensive and less precise gearbox down the line seems like a very backwards approach to determining our permissible backlash as we operate with limited costs on this team. 

A4 Requirements

A4 is also difficult to nail down. When the arm is "straight", backlash on this joint is pretty irrelevant as it only contributes to angular displacement (which is manageable) and not rectilinear displacement.

When the arm is bent 90 degrees about A5, backlash does start to matter. However, does this really look like a configuration we will use in comp? Furthermore, if we find that we do use it, the displacement due to backlash is repeatable as it has the same loading scenario as A2/A3. 

A configuration that we might use is shown below:

Backlash is important here, as the payload mass is not along the axis of rotation for A4 - so there we can't expect any repeatability with the backlash. If we take a look at the "lever arm", we can see that to stay within our backlash requirements we would need deflection of ~1º or 60 arcmin. 

Let's keep this in mind for the next section.

Axis 5

Since axis 5 can rotate via the movement of axis 4, determining backlash requirements really depends on how our most important configuration looks. As discussed with A2/A3, we can expect high repeatability in positioning for A5 due to loading of the arm in the configuration shown below. Thus, this configuration wouldn't be a driving factor for backlash considerations.

However, as mentioned before A5 can rotate. In the configuration shown below, A5 and A1 are parallel, thus there is no "preload" to provide A5 with high positional repeatability. This is the configuration that should drive the backlash requirements.

Axis 1/5

Both axis 1 and axis 5 can be placed in orientations such that there is no preload to improve repeatability of precision. Thus, nailing the backlash requirements for these two axis' is critical. 

It is much more important to minimize backlash on A1 than to minimize backlash on A5. A1 is the furthest joint from the end effector (on average), thus will contribute to the greatest deflection due to backlash as the arm is extended. Ideally, we would have 0 backlash on A1, which is possible with elliptical or high precision planetary gearing. However, getting a true 0 backlash solution may be difficult/expensive. Electromate does off harmonic gears, however I think we should be willing to increase permissible backlash to up to 4 arcmin. Electromate stocks precision planetary gears (UltraTrue planetary) that are precise to 4 arcmin (0.067º). I think these may be less expensive than other types of of zero precision gearing, IMO we should accept this as an option for a first design. 

If we assume that the backlash of A1 is 4 arcmin, we can find that at worst case horizontal extension, the permissible backlash of A5 is ~0.6º (36 arcmin).

However, if we assume 0 backlash on A1, we can see a permissible backlash of ~0.9º but I'm gonna round that to 1º lol (60 arcmin). 

Choosing a requirement of 60 vs 36 arcmin for A5 really depends on what we can get our hands on for A1. Lets be ballsy and assume we can get a harmonic gear!

Remaining Axes

Since the backlash requirements of A2, A3, A4 aren't too critical, lets just match it to the requirement for A5 to keep things consistent. Ideally we would use the same style of gearbox across all of these joints to simplify design/motor integration. I haven't talked about A6 yet, but the requirements for A6 are so low lol. A6 the closest to the end effector and only really contributes to angular displacement from backlash, so it doesn't have any requirements stricted than A1 or A5. To improve consistency, lets say that A6 will also follow A5's requirement. 

Other Considerations

Mass

We probably need some mass restriction to make sure that the arm doesn't become ridiculously heavy. We are currently aiming for an arm mass of 10kg, and the rest of the arm will consist of mating and structural components that can be optimized for mass design through carbon fiber or whatever. Kinda freeballing it, but I'm going to give a lofty goal of maximum 0.8kg per motor/gearbox combo. There's not really any rhyme or reason for mass; I just figured that having our arm mass consist of ~45% motors seemed appropriate. I'd like input on this assumption.