As a lead-in to the upcoming Motion + Power Technology Expo, my article this month will focus on load-holding technologies. The theme of this article is the old adage, “the brakes on a car do not stop the car; the tires stop the car.” The brakes stop the tires from rotating. Perhaps a point of undue subtlety, however, we want to take a detailed look at each part of the motion control and holding technology spectrum to assess and analyze problems with and opportunities for improvement.
First, let me define two terms as they are to be used in this article:
- Motion control, wherein we are attempting to decelerate and/or accelerate a load (mass) connected to our system through an elastic connector (i.e. a rope or cable, etc.).
- Load-holding devices are those that, once a load is stationary, work to maintain the position or orientation of the load relative to a reference (i.e. at a particular position for an extended period of time).
As we look at the downstream effect of our geartrains as a device and the means by which we control them and the loads they are moving, we have to consider how we start a load that is not moving, stop a load that is moving, and hold that load at a given position. When stopping a moving mass, such as an actual mass connected to our geartrain through an elastic media, we need to be very aware of the response of the load and the support structure to this change in inertia. If we attempt to stop a large mass too quickly, the resulting load on the connector and/or support structure may exceed otherwise manageable design limits. It is generally accepted that a factor of approximately 2.0 should be used when designing systems that can make stepwise changes in the applied load (in the case of a cable, tensile load, etc.). The use of this factor is to address the ability of the cable to store and release energy induced by the load-holding device. That is only one example of the considerations that should be considered when designing any type of gearbox with load-holding capabilities. Every combination of components and industries applicability has its own values to be used.
To put all this in perspective, until now there have basically been only two ways to stop motion: friction and mechanical latching.
The friction methods rely on component interfaces that dissipate relative motion as heat through surface wear, and then when the two components stop moving, the friction holds them as long as the surface has not been damaged during the energy dissipation phase and the clamping force is sufficient. Friction-based motion control load-holding capacity is based on two factors — the local effective coefficient of friction and the realizable applied normal force. Assumptions made during the design of these systems are that the applied normal force is consistent across the entire friction face and uniform throughout the clutch pack if a multiple plate clutch pack is used. This assumption is fairly easy to achieve in practice by ensuring a rigid structure and uniform normal force application mechanism (e.g. a hydraulic piston, etc.).
The second assumption is that the local effective coefficient of friction during engagement and while in load-holding mode is constant. This is far more problematic in that it is a function almost exclusively of the friction material and the lubricant used in the clutch (this comment applies even in “dry” lubrication). The effect of the lubricant on the energy dissipation during load deceleration is to manage the interface friction and to dissipate the deceleration energy as heat. The heat, however, causes changes in viscosity, which cause changes in the effective coefficient of friction in an uncontrolled manner. Further, most mid-grade lubricants exhibit a different coefficient of friction during sliding than they do when the clutch is locked. The coefficients of effective friction in the two regimes are typically labeled dynamic, or sliding friction, and static friction.
One major drawback to this difference in effect on the system is that usually the dynamic coefficient of friction is a larger value than that of the static coefficient. This phenomenon is usually defined by the coefficient gradient. If dynamic is higher than static, the lubricant exhibits a negative coefficient gradient. Similarly, if the static is higher than the dynamic, the lubricant would be said to have a positive gradient. Why does this matter? If the dynamic coefficient is higher than the static, then as the load is slowed by managing the applied load on the friction interface (this is the only control we have on the mechanism) internal energy of the load is converted to heat in the clutch and flushed away by the lubricant. For a constant normal force (our control) into a steady coefficient of friction, at a given radius, the system develops an energy dissipation rate that in turn slows the load. The problem: When the load comes to a stop, the effective coefficient of friction as a function of the friction material interacting with the lubricant changes from the dynamic (higher) value to the static (lower) value. This is the mechanism that causes clutch judder/shudder/chatter, etc. More importantly, as the clutch goes through this very transient energy transfer mechanism, the friction material will generally degrade either by excessive wear or excess heat generation which, in turn, causes the lubricant to degrade.
Friction devices fail when the friction surface wears, as it will over time, or its frictional properties change as it gets too hot during an engagement, or dirt or oil gets on the frictional face — the list goes on.
Basically, there are a host of ways friction devices can change in an undesirable manner over their life. When the friction interface does finally get the two components to stop relative to each other, the load-holding system is required to maintain a normal force to keep the friction surfaces engaged and holding. Neither of these issues is desirable. Friction materials and systems are prone to wear-out failure and/or slippage.
Another common example of locking technology is worm and wheel gearing. It allows for relative motion in one direction of power flow but does not allow it in the opposite direction. Well, that is sort of true. A worm and wheel gear-set is said to be non-back-drivable, or not to be able to “back-drive.” The truth is that in the back-drive direction of power flow (wheel driving worm) has a very low efficiency, typically only a few percentage points. Thus, it takes a great deal of back-drive energy to, in fact, back-drive, but theoretically in can be back-driven. This is unacceptable, especially if there is a requirement for complete locking of one component to another. The mechanics of this are again based on a local effective coefficient of friction, which is attributable to the friction material, the lubricant, and the applied force to the clutch pack. Low sliding velocities (the difference in pitchline velocities at the plane of contact) are notoriously difficult to develop a full elastohydrodynamic shear layer (EHL) or a lubricant film layer.
Generally speaking, worm and wheel gearing is operating in either the boundary or mixed-film lubrication region. In these regions of lubrication, the interface between the gears is either asperity contact or, perhaps, even full interfacial contact. Either way, both scenarios rely on contact friction to work. As we know, the contact model looks at the asperity contact, actual points of friction welded together — or so the theory goes. This model is widely accepted and works well enough to use here. The contact points are in a state of stress well above the elastic stress limit and are prone to either cold flow or brittle fracture. Due to this “very near failure limit,” any additional energy added to the contact area will cause the material to exceed its yield limit and fail. In the case of a worm and wheel gear set, the contact patch is usually devoid of lubricant and thus the material is in intimate contact near the yield limit. If any additional energy were driven into the interface contact patch, say any vibration, this additional energy would cause the contact points to fail and motion to occur. This is exactly what we see if we induce a vibrational load into a static worm and wheel set. Its “non-back-drivable” function no longer applies and, in fact, the wheel can drive, albeit very inefficiently, the worm. Thus, a worm and wheel set cannot be considered fully non-back-drivable.
The other common method to hold two rotating components together is some form of positive latch, which does not rely on friction, but does rely on a component (the latch) engaging one of the two rotating components. This engagement, if not done properly, places a great deal of shock loading on the latch, which can cause it to break catastrophically or at least wear down the mating surfaces very quickly.
Latches have been used in many products for a very long time. A simple latch and eye-hook to hold a gate or door shut is a very good example. However, if there is any load on the door (such as the wind trying to blow it open) then it is difficult to get the latch to line-up with the eye-hook. As another example, we have all probably tried to shift our car into park while it is still moving slightly. The automatic transmission uses a latch (pawl) and lug system to hold the car stationary when it is in park. If we try to engage “park” while the car is still moving, we hear a number of clunks before the vehicle finally comes to rest and the latch can properly engage the lug, or parking pawl. The last clunk is generally accompanied by a lurch of the vehicle, which at the latch is a shock load.
Neither of these two technologies provided the operational characteristics we are looking for. What we really need is to not separate the load-driving mechanism (in our example, the geartrain) from the motion-control mechanism. Technologies now exist that provide a rotary path for torque to flow in the positive drive direction that also incorporated a mechanism to fully lock the applied torque load through that same gearset. The technology, and thus the mechanism, uses the geartrain that carries the torque load in drive as the brake mechanism, thus its non-back-drivable load limit is the same as the power capacity of the geartrain.
It is based on the kinematic relationship of the driving geartrain to also lock the main power-path gears, thus the two components attached to those gears, to move and lock without any friction interface to wear or change characteristic responses over time or without any latch component that can fail in shock loading. Input of motion comes from the driver of the system (i.e. a motor or other source rotational energy). This input causes an unlock, which in turn allows rotation based on the input rotational speed of the driver. As part of the design and packaging, these mechanisms can also include a ratio as a function of common gear design considerations.
This aspect of the technology presents an interesting attribute of function that none of the other locking mechanisms has. In a friction-based device, the rate at which a load is slowed to a controlled stop is a function of the friction interface, which means, over time, changes in that interface cause the response characteristics of the friction device to change. Further, it is difficult to provide the user a consistent response or to integrate a control system that compensates for the time-varying response. The latching technology affords no control over slowing down the motion of the controlled load. It can only hold the load once the system has caused it to stop moving (and the latching system is engaged).
In comparison, these new technologies use the rotational speed of the drive motor as the control input to the load. This means as the drive motor slows, the device slows the load in a controlled manner based on its ratio. When the drive motor stops, the device transmits this to the load and then holds the load stationary without any continued input from the drive motor (or in the case of a friction brake, no need for a constant normal force into the clutch to be maintained). Therefore, a controlled acceleration or deceleration of the load is affected by the same device that provides the drive ratio between the motor and load. The mechanism that provides the connectivity between the motor and the load is one and the same as the device that will eventually hold the load stationary.