Planet-9 Porsche Forum banner
1 - 1 of 1 Posts

23,161 Posts
Discussion Starter · #1 ·

by: Kenkg

Description: A Comparison of the Quaife and Guard transmissions Automatic Torque Biasing (ATB) Limited Slip Differentials for the Porsche Cayman S cars equipped with 6 Speed Manual Transmissions*Introduction: A while back, K-Man S asked if I would be interested in doing an engineering analysis of a Quaife and a Guard Automatic Torque Biasing (ATB) Limited Slip Differential. I enthusiastically agreed and started thinking of all the cool evaluations I could perform to get a definitive comparison of the two ATB LSDs, shown below in Figure 1. Unfortunately, events occurred that prevented my grandiose plans from maturing and I have had to content myself with a tear-down inspection, limited semi-precision measurements, analysis, and experience-based observations. This article describes the results of the comparison. I have presented the results in two ways: as an abstract that briefly describes the top-level results and in an evaluation results section that discusses the results in more detail. I hope you have as much fun reading and absorbing it as I did performing it.
Figure 1. R.T. Quaife Engineering and Guard Transmission Automatic Torque Biasing Limited Slip Differentials for the Cayman S Equipped with 6-Speed Manual TransmissionAs far as qualifications go, I have limited automotive experience. The major part of my direct experience was playing with differentials in a modified 1967 Oldsmobile 442 that I drove on the street and drag strip and personally maintained until it literally started falling apart in 1982. If I remember correctly, I had a total of four diffs in the car ranging from an open diff to a no-slip locking diff (which got kinda hairy a couple of times on the track, but that’s another story). My major experience is in aerospace, specifically military fighters and other mission-specific aircraft, conducting and directing design efforts and design reviews, first-article inspections, and engineering investigations on numerous aircraft components and systems over a 40-year career. I’ve seen some very interesting mechanical designs over the years, including a transmission rated at 420 continuous horsepower that weighs 22 lb and a power transmission rated at over 6,000 horsepower that wasn’t really that much bigger than my Olds transmission. Although not directly involved in the detail design, I was involved in discussions and decisions on material selection, effect of design details like helical angle on things like durability, structural considerations to react thrust and axial loads, vibration, cooling, efficiency, etc. Whether any of this experience gives my opinion any real weight in the automotive community is up to you.I need to state that I have no affiliation whatsoever with R.T. Quaife Engineering or Guard transmission. I have no business interest in either company. The opinions expressed in this article are mine and mine alone. I have received no consideration of any kind for producing the article and I have received no advice or opinions from any person affiliated with either Quaife or Guard or any person who owns or operates either unit. Obviously, I have spoken to Ken Smiley about writing the article, shipment of evaluation units and other logistic subjects, including his gentle pushing to get it done, but nothing about his Quaife differential. I have avoided using any copyrighted material. I have received exactly the same amount of technical data from each company – none, and have therefore completely avoided the technical evaluation dilemma of trying to maintain complete objectivity when one company or the other provides more detailed information and especially when they ghost the competition and imbed marketing verbiage in the technical information. Evaluation Conduct and ScopeI conducted the evaluations alone and with no influence or input from either manufacturer. This was intentional to avoid any bias from engineer contact or quality of explanation differences from the manufacturer representatives. In accordance with one manufacturer’s wishes, the units were handled only while wearing gloves and a fine coating of WD-40 was maintained throughout the tear-down, evaluation, and reassembly operations. One of the units came in an installation-ready state while the other came in an evaluation prep state with bolts lightly torqued and no preservative lube coating. The measuring devices I used consisted of a precision caliper with 0.005 mm resolution and 0.01 mm display resolution, a 15 cm metal scale (ruler). I also used a 645 nm targeting laser to produce reflection patterns to qualitatively evaluate surface finish. I want to make sure that everyone understands that the reported measurements were performed with uncalibrated equipment and the stated values are useful for comparative purposes only.Abstract:The R. T. Quaife Engineering, LTD Automatic Torque Biasing ATB) differential and the Guard Transmission ATB differential were evaluated by disassembly, inspection, and measurement with visual and mathematical analysis. Each unit was designed to replace the OEM open differential in the Porsche Cayman S car equipped with a 6 speed manual transmission. The most remarkable differences in the two units are the outer case design and the handedness of the internal gears (the Guard (a US company) unit has a right-handed helical angle while the Quaife (a UK company) has left-handed helicals. Inspection, measurement and analysis of the cases and internal mechanism indicate that there is probably no detectable difference in the functional performance of the units. Each design uses the same number of elements and the design of each element is dimensionally very similar, with the Guard unit demonstrating pinion gears that mesh over a longer length than the Quaife unit while the Quaife unit exhibits less gear lash than the Guard unit. Manufacturing quality and dimensional stability are consistent with Computer Numerical Controlled (CNC) machining practices. Both units exhibited very good fit. All parts were completely deburred and there was no trace of machining debris in either unit. Although slight differences in production operations are evident, they are not significant. It is my opinion that the Quaife and Guard ATB limited slip differentials are high quality units. The handedness of the gear train in the Guard unit results in increased oil thermal stress compared to the Quaife unit. I do not have adequate knowledge to determine if the differential is the most stressing element in the transmission.Evaluation Results:Orientation and nomenclature: In order to maintain correct mechanical orientation, I have referred to the flange end (the flange mounts to the ring gear) as the “driver side” and the side without the flange as the “passenger side” because that is the way it is mounted in our car. Gear handedness is referenced to looking down on the driver’s side and I have arbitrarily referenced the handedness of the entire unit to the handedness of the drive side sun gear. For explanation purposes, I have chosen to reference forces and vectors to the rotating reference frame of the differential case. This avoids the difficulty of attempting to think about differential axle rotation inside the rotating differential. Component nomenclature follows accepted engineering usage with the possible exception that some people would prefer to call the sun gear a “side gear” since that is the nomenclature for an open differential. A picture of the units is shown in Figure 1.Functional Operation: The stock Cayman S differential is a common “open” differential. This type of differential allows complete differentiation between the drive wheels because the pinion gears that connect the left and right axles rotate freely, allowing the left and right axles to rotate completely independently. A good visual of the open differential is available at HowStuffWorks "Open Differentials"Notice that all of the drive force is transmitted through the pinion shaft. The pinion essentially divides the force between the two axles. As long as the vehicle proceeds straight ahead and neither wheel slips, the pinion does not rotate at all and delivers the same torque to each wheel. In a turn, the pinion rotates to allow the wheel on the outer radius to rotate more quickly than the inner radius wheel while transmitting essentially the same torque. A standard open differential can be made to limit slip by increasing the friction in the pinion. In the limit case where the pinion friction is so high that it will not turn, no differentiation of the drive wheels would occur at all. The design of all limited slip differentials (LSDs) depend on friction. The difference is in the method of implementation. The simplest clutch style LSD has a clutch between the drive axles that is preloaded by a spring to provide a fixed friction value between the axles. The clutch limits the differentiation between the drive axles until the difference in torque overcomes the rotational friction in the clutch. Alternatively, the clutch provides an alternate path of transmitting the engine torque. Think of it this way; when one wheel begins to slip, the drive shaft torque spins the axle which, through the rotational friction of the clutch, drives the non-slipping axle. The Automatic Torque Biasing Differential (ATB) uses friction to provide the limitation to drive wheel differentiation, but the geometry and operation are different. The physical arrangement of the driver side with the Axle drive plate removed is shown in Figure 2. In addition, the primary drive forces are shown. *To help orient, if the car is moving forward to the right, the differential would be rotating counterclockwise.
Figure 2. Geometric Arrangement, Nomenclature, and Primary Forces In the ATB, the pinion gears float in the case. The ring gear would mount on the far side of the flange and the front of the car is to the left. The passenger side is a mirror image of the driver side and the driver side pinions each mesh with a counterpart pinion on the passenger side. The mesh of the pinions provides the normal differentiation between the drive axles. Under acceleration, the drive shaft pinion produces the ring gear drive force in the direction shown which produces a normal force on the pinion gear teeth. It is the friction of the pinion gears on the case that produces some of the differentiation limiting. Figure 3 shows the unit with the driver side sun gear and all but one driver pinions removed to show the floating pinion pockets and the mesh of the driver and passenger side pinions. Also removed is the preload spring and spacer assembly that will be discussed later.
Figure 3. Driver Side Pinion Meshing with Passenger Side PinionReferring back to Figure 2, the drive force on the top of the pinion teeth is one source of the friction force that limits differentiation. To quantify these forces, let’s estimate the forces that result in measured accelerations that are documented on the Planet Porsche website. A first gear acceleration of 0.67g results in a total thrust of approximately 2,000 lb, or 1,000 lb per wheel. Since the tire radius is about 12 inches, the axle torque is about 1,000 ft-lb. All of the torque is being transmitted by the pinion gears through the sun gears (which are splined to the axles). The radius of contact is measured at 36.7 mm, or 0.120 feet (this dimension is the same, within my ability to measure, for both units). That means that the force on the 6 pinions is about 8,300 pounds, or about 1,385 lb per pinion. If the coefficient of friction of the pinion gear tooth top is 0.1, then the force of friction is about 138 lb. Since the radius of the pinions is about 11.5 mm (average of the two units), then the torque required to overcome the friction force is about 5.25 ft-lb per pinion, for a total of 31.5 ft-lb for all 6 pinions on one side and 31.5 ft-lb for the pinions on the other side for a total axle torque difference of 63 ft-lb to overcome pinion drive force friction. Let me reiterate that this estimate is based my assumption of the coefficient of friction. The values for the coefficient of friction for steel on steel range from about 0.6 for dry surfaces to about 0.06 for lubricated surfaces. I chose the value 0.1 because it represents a “mostly lubricated” case and because it makes the math easier. The surfaces and gears in both units are newly machined and I could feel the slight tool marks in both units when I rotated a pinion against the case, so even a calibrated finger guestimate of the force of friction versus normal force is considered too inaccurate to use. I will say that I could feel no difference in the roughness of the gear motion due to machine tool marks. From observation, the tool marks are very acceptable and consistent with modern CNC equipment, although the tool marks are slightly more noticeable on the Guard pinion gears while the marks on the Quaife case are slightly more evident. Other sources of differentiation limiting friction are the friction caused by the axial and radial thrust loads as a result of the pinion to sun gear mesh. The easier to measure and discuss is the thrust loads, so I will take the easy way out and discuss thrust loads first. The thrust load is a result of the fact that both units use helical gears and the thrust load is dependent on the helical angle. Referring to Figure 4, I measured the helical angle by marking the start position of one of the teeth and then rolling the gear until the other end of the tooth contacted the paper. From there it was application of the Pythagorean Theorem. Within my ability to measure, the helical angle of each unit is the same at approximately 50 degrees.
Figure 4. Determination of Helical AngleThe effect of helical angle is twofold. Helicals are quieter than square cut or spur gears and they produce significantly less vibration. The ATB uses the helical gear to produce axial thrust and additional differentiation limiting friction because the sun gear and pinion gears develop forces on their ends because they drive each other in opposite directions. Figure 5 shows the thrust force on the sun gear. An equal and opposite force acts on the pinion gear. This action produces binding as the sun gear is driven into the case end cap and the pinion gear is driven into the machined pinion pocket (Quaife geometry). Again, we will estimate the forces using the first gear acceleration introduced earlier. Recall that the total load on the pinion gear that produces thrust is 1385 lb per pinion. That force is reacted into the 50 degree “ramp” of the sun gear by the pinion gears. In Figure 5, I drew the drive force to a clearly visible tooth face, so just imagine that the pinion producing the drive force is transparent. While the sun and pinion gear rotation with respect to the case are constrained by the counter-torque of the offside pinion, there will still be some slight rotation due to gear lash and vibration to “set” the thrust forces. With all of the drive wheel torque (1,000 ft-lb in our example) being transmitted through the sun gear, the total rotational force on the sun gear (assumed to be applied at approximately 50% of tooth height radius of approximately 25.5 mm) is approximately 11,900 lb which results in a thrust load of approximately 7,650 lb. That load is also reacted by the 6 pinions.
Figure 5. Helical Gear Thrust ForceIt is now time to discuss the most important design difference between the two units, the handedness of the helical gears. In the Quaife unit (the Quaife driver sun gear and one driver pinion are shown in Figure 5), the thrust forces drive the sun gear into the case end cap and the pinions into the bottom of the pinion pockets. In the Guard unit with the opposite handedness, the sun gear is driven toward the center of the case while the pinions are driven into the case ends. The implications of this design difference will be discussed later, but for now, back to our calculations.If the sun gear reacts 7,650 lb, the friction is 765 lb (on our assumption of a 0.1 coefficient of friction) and acts at a radius of approximately 22.5 mm, the torque required to overcome the friction is about 56 ft-lb. For each pinion, the thrust load is 1275 lb, the friction about 127 lb, the effective radius is approximately 6 mm, and the torque required to overcome friction is about 2.5 ft-lb per pinion or 15 ft-lb for all six pinions. Thus, the thrust loading produces a total frictional torque potential of approximately 71 ft-lb on each axle. However, we have to remember that the passenger side pinion is meshed into the driver pinion and balances the driver pinion rotation and thrust force. That means that the pinion friction is not present in balanced operation, so in our straight acceleration scenario, the torque friction due to thrust is about 56 ft-lb. Note that the thrust loading produces approximately twice the friction torque of the pinion tooth frictional loads due to drive force. The sum of the tooth friction and the thrust friction is about 88 ft-lb.The next source of differentiation limiting friction is radial loading between the sun gear and pinion gears. Radial loading occurs because the gear teeth engage before they reach the centerline of the gear train, thus they try to push each other apart. Without detailed knowledge of the gear tooth profile design and meshing parameters, I could not conduct a definitive analysis of the gear trains, so I did rough order of magnitude estimate based on the angle formed by varying the mesh on both sides and then halving the angle to get an estimate of the mean angle for thrust determinations. Figure 6 shows the result of restraining the pinion gear in a single orientation and tracing the angle as the pinion tooth is moved along the sun tooth. This is probably the least accurate estimating methodology of all that I have done, but it will probably produce a rough order of magnitude estimate in that the methodology essentially measures the sum of the slant of the sun gear flank (tooth face) and the pinion slant. The measure of each flank combination was only a few millimeters in length and I projected the lines by a factor of about 10 and then geometrically determining the angle. After all that, I measured the effective angle to be about 33 degrees for the Quaife and 27 degrees for the Guard, so I will use 30 degrees for this discussion. Going back to our determination of the loads on the sun gear, the pinion tooth-to-sun tooth contact transmits all of the drive force, or 11,900 lb which is distributed to the 6 pinions for a tooth-to-tooth drive load of 1,983 lb per pinion. Resolving that load through our 30 degree tooth contact assumption results in a radial thrust load of 991 lb. The arrangement of the radial thrust is shown in Figure 6.
Figure 6. Radial ThrustNow we have to modify our calculation of pinion tooth friction load by vectorally resolving the pinion drive force load and the radial thrust load. Recalling that the pinion drive force is 1, 385 lb, the resultant load vector is 1,703 lb, which means that our total pinion tooth friction load is 170 lb per pinion which results in about 6.5 ft-lb per pinion to overcome friction. And, of course, since there are 6 pinions, the pinion friction torque is 39 ft-lb per side.At last we arrive at the final result. The total torque required to overcome friction in a hard acceleration is 56 ft-lb from axial thrust loads and 39 ft-lb from pinion radial thrust loads for a total of 95 ft-lb per side or a total of 190 ft lb out of a total of 2,000 ft-lb of input. And what confidence do I have in that number? Not too much at all because it is dependent on my estimate of the coefficient of friction which is based on the kinetic coefficient of friction rather than the static coefficient of friction which is higher than kinetic. On the other hand, I think the estimates of the drive forces acting on the unit are fairly accurate because they are based on some fairly realistic measurements. Yes, I rounded some parameters to keep the math more manageable and, for the purists, I ignored the radial force and friction due to centripetal acceleration. I also ignored the minor contribution of the spacer-preload device. Here’s what we have learned so far: 1. Development of differentiation limiting friction is directly proportional to input torque. 2. Axial thrust loads provide about twice the differentiation limiting torque than radial loads. Now let’s see what happens when the wheels actually differentiate. First, let’s set a scenario: 50 mph in a 1 g right turn. These conditions result in a center of mass turn radius of 167 ft, outer tire turn radius of *169.5 ft and an inner tire turn radius of 164.5 ft. Outer tire rotation rate = 649 RPM. Inner tire rotation rate = 629 RPM. The differential case is rotating at 639 rpm. With a differential wheel speed of 20 rpm and a sun to pinion gear ratio of 3, the pinions are rotating at 60 rpm. Assuming half throttle in third gear, the rear wheel torque is about 200 ft-lb. Now that we have it set up, let’s take an intricate look at the forces on the car and what happened to transition the car from a straight path to a 25 degree/sec turn. Referring to Figure 7, the car is moving up the page and the turn is being initiated to the right. The figure shows the driver sun, one of the driver pinions, one of the passenger pinions, and the passenger sun. The geometry is offset because I couldn’t figure out how to arrange the gears in a geometrically correct arrangement while visually showing the mesh and interrelationship. The forces shown in the figure are from static conditions. Recall that both the driver side and the passenger side gear pairs are in an equilibrium condition that has them in an axial thrust and radial thrust condition that, at 200 lb of torque, produces a differentiation limiting friction torque of about 19 ft-lb (using the above results). The gears in the figure are from the Quaife unit, so the axial loads are in a direction that would driver the gears apart from each other. Also, Figure 7 represents forces with respect to the rotating case. The gears would rotate in the indicated directions if the differential was constrained, 200 ft-lb of torque on the differential with the front wheels on rollers, and several folks push on the left front fender to rotate the car to the right about the differential. With the exception of centripetal acceleration loads on the differential case, the physics are the same. With the stage set, here’s what happens from the point of view of the differential case. As soon as the car begins to turn, the right wheel starts turning slower and building an increase in torque. This increase in torque increases the axial load and thus the axial load friction on the passenger side. (Think about this: if you are on the right side of the car holding a wheel against a clockwise rotation and you want to rotate the wheel counterclockwise, you apply a torque in the counterclockwise direction). On the driver side, the wheel begins to accelerate. This decreases the torque on the driver sun-pinion pair and decreases the axial load and thus the axial friction torque. When the torque on the driver sun-pinion reaches the torque friction, the sun gear proceeds to rotate. At the point of rotation, the outside wheel torque equals the differential torque minus the friction torque and the inner wheel torque is differential torque plus friction torque. (An aside: more torque on the inside wheel is the reason limited slip differentials contribute to understeer.) Once the driver sun gear starts rotating, it engages the driver pinion which engages the passenger pinion. Here it is important to note that the driver pinion is rotating in a direction that decreases axial thrust on the passenger pinion that decreases the friction of the pinion-to-pinion-pocket interface. Once the rotation starts, it accelerates and then stabilizes when the turn rate stabilizes. In our example with input torque of 200 ft-lb, the driver side wheel will maintain a torque of about 85 ft-lb while the passenger wheel will maintain a torque of about 105 ft-lb.
Figure 7. Relative Gear Forces in a Right Turn Now let’s look at what happens when the inner begins to lose traction and slip. As the inner wheel accelerates, the relative rotation rate decreases and is zero when the wheels are rotating at the same rate. Now the roles reverse from the stabilized turn case. The driver wheel now sees the higher torque value of 105 ft-lb and the inner wheel the lower value of 85 ft-lb. If the passenger wheel continues to accelerate, the passenger pinion will not be able to counter the thrust load of the driver pinion and the torque friction value will increase another 3 ft-lb or so.As you can see, the actions and torque transfers in an Automatic Torque Biasing LSD are different from a standard open differential. The operation of the ATB results in torque being biased to the slower-rotating wheel while an open differential can only apportion the torque in a very nearly 50/50 proposition. However, nothing is free, and the action of a LSD will change the way the car handles. To what extent depends on the decisions made by the design engineers and the selection of the critical design parameters such as helical angle, center-to-center spacing of the sun and pinions, number of pinions, pinion diameter, sun diameter, and the mean diameter of the sun friction contact surface. These dimensions are so similar in the two units that I would be surprised if anyone could tell the difference in performance even in a side-by-side test. The only difference I could find at all that would affect performance. Mechanical Characteristics: The mechanical characteristics of the two units include measurements and observations. I will discuss the items and characteristics that I think are important or significantly different.Case Design and Manufacture: Both cases are machined with precision consistent with Computer Numerically Controlled (CNC) milling machinery. Both designs utilize a precision insert in the driver side to maintain axle alignment within very close tolerance. The driver side of each is shown in Figure 8. The inserts of both units are not interference fit, but they are very close to a zero-tolerance fit. Any misalignment at all during assembly results in binding. Both inserts exhibited one orientation that seemed preferential for insertion. Once inserted, both end pieces would rotate without binding but there was no detectable freeplay in either unit.
Figure 8. Quaife and Guard Driver Side End Inserts with Oil Passages shownThe passenger sides of each unit represent significantly different manufacturing approaches and are shown with the Quaife unit on the left in Figure 9. The Quaife unit utilizes a precision machined cap secured to the case by 6 bolts. The quality and precision of the end cap is consistent with the driver side insert. Guard chose to integrate the axle bushing and bearing mount into the case. In the Quaife unit, the end cap forms the outer end of the passenger pinion pockets. In the Guard unit, the passenger pinion pockets are machined through the end of the case then enclosed with inserts held in place by snap rings. Although the single piece machining of the Guard unit offers more rigidity, the pinion blocking plates retained by snap rings is a less efficient thermal transfer mechanism, and heat is always an issue in friction-operated devices, especially considering that the pinion cover plates have to absorb the loads and heat when the passenger axial load is not counteracted.
Figure 9. Passenger Side of the Quaife and Guard units with Oil Passages ShownOil Passages: Oiling is always an issue with friction-operated mechanisms or gear trains. Both units have oil passages in both ends and in the case. Both drive axle bushings on each unit have helical oiling grooves machined into the bushing surface as shown in Figure 10 which shows the driver side bushings and oil grooves. (Looking down the bushing you can also see the interior splined surface of the sun gear and the top of the spacer-preload assembly. On the Quaife unit, the hole at approximately 11 o’clock is an oil passage and you can see the end of one of the pinions. You can also see the sun gear splines and spacer-preload assembly in the Guard unit. The crescent-shaped holes at the 9:15 position and partially visible at the 7:00 position are oil passages that lead to the outside surface of the pinion pocket.) Although the Quaife oiling groove is approximately twice as wide as the Guard groove, I have no reason to believe that either is too narrow or too wide since the bushing radial loads are low (all the load is carried by the roller bearings mounted on the bushing outside surface. Sun gear loads are always balanced radially by the 6 pinions, so the only loads on the bushing are centering loads which should be low. While the oil passages in the Quaife unit look a lot larger than the Guard unit. The Quaife passages are centered on the sun-pinion mesh, so the open area is really similar to the crescent on the Guard unit.
Figure 10. Quaife and Guard Driver Side Axle Bushing Surface with Oil GrooveReferring to Figure 9, holes in the side of the cases are evident. I think the size of these holes compared to the size of the oil inlet passages reveals a difference in oiling philosophy between the two units. The case holes in both units are located on the outside of pinion pockets. The Guard unit has an elongated hole on the outside of every pinion while the Quaife unit has one hole in each passenger pinion pocket, but no holes in the driver pinion pockets. That completes the obvious physical description. Now let’s talk about operation. Both units operate as a centrifugal pump to transport oil from inside the case to outside. If the environment inside the differential containment case were benign and orderly, I would say that the Quaife unit has more efficient centrifugal pump characteristics with the entry area to exit area ratios, blah, blah, blah. But the environment in a differential containment case is anything but benign and orderly. With all the splashing, slinging, throwing and flowing going on, it is very difficult to visualize or calculate the viscous fluid dynamics. The bottom line is that I have no reason to believe that either unit is inadequately or marginally lubricated. And if you are worried about the driver side pinion pockets in the Quaife unit, it is not an issue. Oil can reside in the pocket to about an eight mm depth before it overflows into an adjacent passenger pinion pocket. Plus, the oil passages in the ends are not unidirectional. When the car is in a turn, the helical pitch of the pinions will drive oil into or out of the oil passages, depending on turn direction, and it is in a turning condition that the gears are in relative motion and friction is being generated and oil flow is needed. Also remember that the mesh of the pinions means that there is one side turning in a direction to draw oil in while the other side pushes it out (opposite handedness but turning in opposite directions). Gear Train: The gear train is really the heart of the design. As discussed previously, gear characteristics such as helical angle and mean load bearing radius have a lot to do with the overall operating characteristics of the differential. In analyzing the two units I was more surprised at the similarities than the differences. My original assumption was that one company or the other would determine that a more aggressive biasing ration would improve the handling, but it is not amazing to me that two engineering teams incorporating the ATB philosophy in a defined envelope and a compromise between street and track handling came up with nearly identical design driver parameters. Now on to the part-by-part discussion.The gear train components are shown in Figure 11. As has been the convention in this article, the Quaife parts (or “bits” in the UK) are on the left and the Guard parts are on the right, this convention being chosen to conform to the company home country’s driving preference. The figure shows, from top to bottom, driver side sun gear, spacer-preload assembly) consisting of 6 spring elements, two splined spring end caps, and an internally splined container), one (of 12) pinion gear and the passenger sun gear.
Figure 11. Quaife and Guard ATB LSD Gear Train ComponentsThe driver and passenger sun gears from both units are 18 tooth helical gears with, as closely as I can measure, the same helical angle. The tooth profile of each is slightly convex (or crowned) and the tooth depth is approximately 6mm on each. The tooth thickness is approximately equal but I could not measure it with adequate repeatability to draw any conclusions. The diameters are slightly different, with the Quaife measuring 58.81 mm and the Guard measuring 58.70 mm. The Guard driver and passenger sun gears are dimensionally identical with a total gear thickness of 24.31 mm with a 6mm bushing insert ((I have long since forgotten the technical name. I am probably also embarrassing myself on forgotten gear terminology), the only difference being the handedness of the helical angle. The Quaife driver side sun has a thickness of 28.01 mm with a 2.08 mm bushing insert and the passenger side has a thickness of 22.83 mm with a 7.11 mm bushing insert. Why the difference? I don’t know, but the entire outer diameter of the driver side sun is recessed into the end insert by about 6 mm while the passenger side is recessed a little more than 1 mm. With the outer diameter recess, the available gear thickness available for pinion mesh is approximately 21 mm on each side. The Guard unit has not recess the outer diameter into the end cap at all. A comparison of the two driver side inserts is shown in Figure 12. Dimensional stability for both units is very good. I evaluated dimensional stability by measuring the outer diameter on three equally spaced teeth at each end and in the middle. In the case of the Quaife and Guard gears, no appreciable differences were found and it is most probable that differences were due to measurement error more than manufacturing variation. The only visual difference is that the Guard sun gears appear to have a slightly finer finish than the Quaife sun gears.
Figure 12. Quaife Passenger Side End Cap, Driver Side Insert, and Guard Driver Side InsertThe pinion gears are similarly similar (forgive me Mary Barns, my high school English teacher. I once watched that woman pile drive a seven inch thick Webster dictionary into a sleeping student’s head. I, and most of the class, thought she had killed him. Lessons like that don’t leave you.) They each have six teeth. The tooth thickness is essentially the same. The tooth profile is definitely crowned and I could detect no difference visually. Examples of the pinions are shown in Figure 13
Figure 13. Quaife and Guard Pinion GearsThe difference in pinion gear length is the most prominent difference in the gear train. The Quaife design results in a driver side to passenger side overlap mesh of approximately 17 mm while the Guard pinion overlap is approximately 22 mm. As nearly as I can measure, the helical gears remain inn mesh for 16 mm, so the two units provide the same gear-to-gear and tooth-to-tooth loading. Another difference is the production differences. The Quaife pinions exhibit countersink and machine tool witness marks that indicate that the manufacturing process is a cut-to-length, then mill while the Guard process is mill then cut-to-length. Upon initial observation I was alerted to the increase in manufacturing tolerance propagation in the Guard process, but dimensional stability measures once again indicated that the most probable source of error is yours truly and his inability to measure to the same precision as the parts are produced. I did detect that the Quaife outer diameter (tooth tops) is defined by cutting tool marks and the Guard pinions are defined by grinding marks, again proving that two different manufacturing processes can result in the same precision part. Similarly to the sun gears, the Guard pinion gears have a slightly smaller outer diameter, 22.81 mm compared to 23.24 mm for the Quaife. Opposite to the sun gear results, the tooth face (or flank) finish is finer in the Quaife unit than in the Guard unit.Now what happens when we put these gears into the case? How does the case retain these gears? Well, if I measure from the outer edge of one pinion pocket to the outer edge of the opposite pinion pocket, I get 96.96 mm for the Quaife unit and 96.95 mm for the Guard unit. The looseness or freeplay in gear trains is known as lash and the lash in the Guard unit is clearly larger than the Quaife unit. Without some very small feeler gauges, I am unable to measure the lash. I estimated the motion of the unconstrained sun gear meshed all 6 pinions. That does not measure lash, but the radial motion required to eliminate the lash. The Guard sun gear motion was approximately 3 times the sun gear motion in the Quaife. I then held the Sun gear in the approximate center (simulating the restraint that would be provided by the axle) and evaluated my ability to rotate individual pinion gears between the surfaces. Again, the Guard unit displayed more play than the Quaife with the Guard pinion moving an estimated half mm and the Quaife pinion moving hardly at all. Looking at the function of the differential, the difference in lash is, in my opinion, not significant. The gear train is not subject to very rapid reversals, so the problems normally encountered due to lash are not present. In my opinion, while the Quaife unit has less lash than the Guard, there is no reason why either should not exhibit acceptable characteristics after wear-in and after a long period of use (assuming equivalent materials). Further, I do not believe I would want the gear train to be tighter than the Quaife.The last part of the unit is actually an assembly. The spacer-preload assembly floats between the driver and passenger sun gears and is located in the region of pinion gear overlap. Figure 14 shows the spacer-preload assembly in various stages of disassembly. Figure 11 shows the individual pieces.
Figure 14. Spacer-Preload assembly.The assembly consists of the spacer and the preload stack. The spacer is a 12 sided scalloped machining that is internally splined to accept the preload stack. The function of the spacer is to insure that a sun gear does not come in contact with the opposite side pinion gears and the thickness dimension of each spacer reflects the pinion overlap in each unit with the Quaife spacer at 17.61 mm and the Guard at 23.00 mm. The outer diameter is similar, with the Quaife outer diameter measuring 55.98 mm and the Guard measuring 55.62 mm.The preload assembly consists of two end caps which partially contain the springs. The caps from both units are splined with 24 teeth. The outer diameter of the Quaife cap is 39.12 mm and the Guard is 39.19 mm. The spline in the caps prevents rotation of the spring elements relative to each other. The thickness of the caps sums to the total thickness of the spacer, so that compression of the springs results in end cap-to-end cap contact and a slight increase in load bearing surface area.The springs consist of six concavo-convex circular sections with a central hole stacked in an alternating manner with the end spring mating the outer diameter with the end cap surface. The only meaningful dimensional difference is the material thickness, with the Guard material about 50% thicker than the Quaife. With the end caps installed, there is approximately 4.3 mm of available compression in the Quaife unit and 3.6 mm in the Guard unit. From measurements made during disassembly, the Guard spring pack is compressed approximately 3 mm and the Quaife unit just a little less. During reassembly I tested the loading by pushing on the case inserts prior to inserting the bolts. I was able to bottom both inserts against the spring force, with the Guard unit providing more force, consistent with the thicker spring material. The level of preload is not significant to the differentiation function of the unit and was not used in any of the differentiation examples.Thermal Considerations:Now that we have discussed the mechanical characteristics of both units, it is time to consider the implications of the design and operation to the local external environment (transmission case that enclosed the differential) and internal environment (individual parts in the differential assembly). With my lack of analytical tools, my analysis is based strictly on basic engineering and physics and my experience. I’ll talk my way through it as plainly as possible and I will try to define assumptions as precisely as possible. But even with that, these results are based on a lot of gut feel mixed with a good portion of technical skepticism.Remember the discussions we have had about the gear handedness? In the Quaife unit, forward torque causes the sun gears to react axial thrust into the end caps and the pinion gears react axial thrust into the machined pinion pockets in the case. In the Guard unit, forward torque causes the sun gears to react axial thrust from both sun gears into the spacer-preload device while the pinions react axial thrust into the end cap and plug plates on the passenger side. This basic difference leads me to have a preference for the Quaife implementation over the Guard implementation. While there is probably little difference in the total heat generated by either unit, the implementation of the Guard unit concentrates the sun gear friction heating in the center of the unit and specifically in the spacer-preload assembly.First, let’s consider the location of the spacer-preloader. It is shown in Figure 15 and, as discussed earlier, it floats between the driver and passenger sun gears and the twelve pinion gears. The location is the same in the Quaife unit, but the exposure to axial thrust loads and frictional heating is significantly different.
Figure 15. Spacer-Preloader LocationIn order to understand the magnitude of the difference, we need to quantify the conditions under which the spacer-preloader is under axial compression load. In the Quaife unit, the condition that puts the spacer-preloader under compression is deceleration torque. In order to estimate the deceleration torque, I performed two second-gear decelerations from 60 to 40 mph, one in each direction on the same stretch of the same road (to average out slope). The result was 8.2 sec and if we attribute all of the deceleration to negative torque, the differential is exposed to a negative torque of approximately 165 ft-lb. So, in the Quaife unit, the spacer-preloader is under axial compression only in a negative torque situation. Further, it is being heated by friction only when the wheels are differentiating, so there is not heating during straight line braking. The result is that the Quaife spacer-preloader experiences friction heating from low axial loads and for limited time frames.For the Guard unit, the spacer-preloader is experiencing axial loading and friction heating any time the car is turning under positive torque. From my recollection, that’s most of a lap time. From our 50 mph turn example, I calculate that that the spacer in the Guard unit heats up approximately 1.2 degrees C for every differential rotation. Recall that in the example, the axles are rotating at 10 rpm with respect to the differential case, which means that one full rotation lasts 6 seconds. For a second example, let’s consider 80 mph, full throttle in third gear, and 0.8 lateral g (accelerating out of a turn). The differential torque is 684 ft-lb per side and the temperature rise is 3.9 degrees C per rotation, which takes 12 sec. If we combine the two scenarios and construct a turn that lasts 4 sec and an exit that lasts 3 sec, we get a total temp rise of 1.8 degrees C. If a lap consists of 15 such turns, the temperature rise will be approximately 26.6 degrees C per lap (assuming no cooling, which we will discuss shortly). In a 30 minute session at 2 minutes per lap, the spacer will absorb enough heat to increase the temperature by about 399 degrees C. Please realize that this is only the heat added that would result in that temperature being achieved in a perfect thermal insulator. We will discuss cooling strategies next. And to keep things in perspective, we are talking about the heat required to heat a component that weighs less than 0.4 pounds. To put it another way, if all the heat is absorbed into the 2.8 liters of transmission oil, the oil temperature would rise about 6 degrees C.The cooling strategy and local environment around the spacer is where I run into problems. With the spacer floating, there is not a reliable surface-to-surface contact to allow conduction through adjacent parts. Recognizing that the sun gears absorb approximately half of the generated heat, they are not subject to the same temperature rise because of the integrity of the surface-to-surface contact of the sun gears to the related splined axles. Additionally, the preload device will tend to prevent good contact between the sun gears and the spacer during periods of non-rotation, so there will be limited opportunity for conductive cooling. That leaves oil flow as the major cooling mechanism. Here too, however, there are serious limitations. At 639 rpm, the radial acceleration at the outer surface of the spacer is approximately 330 ft/sec2. And, even though during a turn the pinions are rotating at 60 rpm, the net radial acceleration will significantly accelerate the oil away from the spacer. Naturally, the viscosity of the oil will allow some flow, but not much. The only other oiling path is through the center holes in the sun gears, but I cannot imagine that appreciable flow can come through the bushing oil grooves and then through the spline fit to get to the spacer. And what is the necessary oil flow to limit the temperature rise? The heat flux is approximately 1100 joule per minute. If we are willing to allow an oil temperature rise of 50 degrees C, the oil flow rate would need to be about 3.7 fluid ounces (110 cc) per minute. That’s not much oil and it could well be that the rotation of the pinion gears will provide adequate flow to limit the local oil temperature rise to a value that does not cause oil degradation. I am sure of one thing: The design of the Quaife unit causes less thermal stress on the transmission oil.So there it is. Two Automatic Torque Biasing limited slip differentials designed on opposite sides of the Atlantic and the most remarkable difference is the direction of gear rotation and the resulting heat rejection efficiency. All of the other minor differences are, in my opinion, inconsequential.I have one final thought for those folks who see the similarity of dimensions and parts counts and start thinking that somebody copied somebody. In my experience, two competent engineering companies trying to solve the same engineering problem and constrained to the same physical envelope will often generate designs that are uncannily similar.

1 - 1 of 1 Posts