A 6-speed manual transmission was redesigned and optimized using powder metal steel as the gear material. The design work was comprehensive and included deformations of the housing, shaft, bearings, bolts, and gears, as well as varying temperature conditions and loads. Paired with extensive gear fatigue data, the analysis showed that powder metal (PM) gears would survive the durability cycle despite that no densification methods were used. The safety factors against pitting and tooth root breakage as well as the confidence level of fatigue data is presented in this paper.
In the second phase of the project, a Saab 95 was equipped with a prototyped gearbox according to the PM optimized design (see Figure 1). Another three transmissions were built and tested in a test rig for durability, abuse, ultimate breaking strength, and vibrations. The transmission has also been benchmarked with the original transmission, and comparisons are presented here. Results, lessons learned, and ways on how the design can be improved in future transmissions are also discussed.
The effort to design an automotive transmission using PM has been presented in previous works [1, 2]. The ambition was to:
- Understand what kind of stress levels are present in a modern 6-speed manual transmission on each gear and how many revolutions does such a typical transmission see during a 300,000-km design life.
- Determine if PM can meet the stress and cycles, and if so, what would the manufacturing process be for the individual gears.
- Build the transmissions and put in a demonstrator car.
- Put the transmissions in test rigs under OEM-like conditions to see if the fatigue data generated in an FZG rig for spur gears could be used to predict life for helical gears in a car transmission.
- In some lesser way, validate the gearbox, but a full OEM validation program was not within the scope of the project.
- Powder metal parts can have their strength tailored through processing to a greater extent than solid steel parts. The amount of processing required affects the part price, so every gear in a transmission can have different processing in order to be as cost-efficient as possible while meeting the technical requirements.
In order to predict the strength and service life of gears, there is one important prerequisite: fatigue data generated on gears, preferably on similar gears to those that are in the transmission. Fatigue data from standardized FZG C-type gears and smaller spur gears had been generated to create 40-plus point S-n data curves as the lowest baseline level for pitting and bending fatigue. So when designing the gears, it is also possible to calculate the safety margin against failure. However, that safety margin calculation has little value if the test data is not of good quality. By performing the durability testing on the 6-speed manual transmission, it is possible to gain further understanding of how the FZG S-n data relates to the durability of the 6-speed transmission and if it can be used to predict safety margins in gear design.
A 6-speed manual transmission with a torque of 320 Nm was designed, built, and tested. It has a three-shaft layout and a transversal design for front-wheel drive (see Figure 2). It can fit in several C and D segment cars such as the Saab 95. It has seen some redesign over the years with different bearings, shift tower, and differential housing design, which was discovered during the course of the project.
In the transmission, all gears except the first and second drive gears are converted to PM. The reason for not making the first and second drive gears in PM was that they were cut on the input shaft, and it would be difficult to attach a PM gear on the shaft rather than make the gear teeth themselves out of PM. The final drive and output shaft gears were not in PM either for similar reasons.
The gears were cut using pucks and conventional gear-cutting, except the fourth drive gear that was made from a compaction tool under series, production-like conditions. The different material compositions and densities can be found in Table 1.
One of the more important lessons learned in this project was that the alloying content needs to be higher than in Astaloy 85Mo and Distaloy AQ. Adding graphite increases hardenability but leads to other problems. When making PM gears today, the Astaloy Mo and Astaloy CrA with 0.25% C is used, since that leads to good hardness curves in both root and flank. (See Figure 3.)
Adding graphite to Astaloy 85Mo increases hardness in the root but gives a through-hardened flank. Balancing hardness in the root and flank so they overlay the solid steel reference proved impossible using only graphite as the balancing element. More molybdenum, chrome, nickel, or micro-alloying elements is the preferred solution.
The heat treatment is such a critical component in determining the fatigue strength of the gears that it cannot be taken lightly. Despite this, the heat treatment specifications in gear drawings are often brief and give a lot of room for variation.
The manufacturing sequence is (except the fourth drive gear):
- Compaction + sinter of pucks
- Turning + hobbing
- Case hardening (CQT)
- Hard finishing of bore and teeth
The tolerances of the gear teeth are the same as for a ground solid steel gear. Also, for the compacted fourth drive gear where the spline in the bore is not touched after compaction, ISO 6-7 is reached on all tolerance parameters after grinding of the gear teeth. Figure 4 shows the development of radial runout (Fr) through the manufacturing process of the fourth drive gear. Figure 5 is a comparative picture of the original gear versus the PM design.
The test rig used for NVH, durability, and overload/breaking load is depicted in Figure 6.
A toothed belt is driving the input shaft, and the differential is welded, allowing output through each side shaft, since it’s connected to the brake dynamometers. Between the shafts and the dynamometers, there are torque sensors fitted to measure the shafts’ torque and speed.
The gearbox input shaft was connected through the spline interface to an adapter shaft with the torque sensor sitting on the belt transmission tooth wheel. On the adapter shaft, there was also a torque and speed sensor of the same type as for the output shafts. The oil temperature was measured at the drain plug with a PT 100 sensor. To track the wear of the transmissions during the test, a Reilhofer delta-Analyser vibration monitoring system was also used. Accelerometers were positioned on the gearbox housing and on the differential housing (see Figure 7).
The transmissions were tested for durability, overload, vibration, and breaking load. The durability cycle can be seen in Table 2. The maximum torque for the test cars equipped with a PM transmission is 230 Nm, and the manufacturing path chosen for each gear reflects the stress level and number of cycles so it matches the S-n data as much as possible.
The test torque for each gear is defined in Table 2; overload torque was 450 Nm. Only a few hundred revolutions were run at 450 Nm for each gear to simulate abuse. Torques for first, second, and reverse are reduced. This is also done for the solid steel gears by the computer of the Saab 95 by reducing turbo pressure when these gears are engaged. This is a safety precaution to avoid high abuse loads on the low gears. For surface-densified PM gears, the torque can be increased to 320 Nm, which has been tested in a separate project; 320 Nm is what the high-end model of the Saab 95 delivers. As previously mentioned, the strength of PM gears can be tailored to meet the requirements. Vibration was tested as well as overload, which is presented in the following section.
The duty cycle in Table 2 was followed. The torque levels for first, second, and reverse have been reduced just as it is reduced in the car. The stresses on these gears are quite high, so to avoid breakage, the torque is reduced also for the solid steel original gears.
The gear pairs 3-6 tested out well and survived the durability test on all three transmissions.
First gear + R (Astaloy 85Mo 0.25%C 7.25 g/cc)
The first gear was a convoloid shape gear. In the design phase, the simulations showed that lead crowning was not necessary, which led to a design without lead crowning. In testing, that proved incorrect, and edge contact due to misalignment gave pitting on one of the sides of all the flanks of the driven gear. In retrospect, it was the wrong decision to design the gear without lead crowning. Fifty percent of the durability cycle was done for the first gear, first transmission. It was then decided to stop the test to avoid catastrophic failure, which tends to damage other components in the transmission, delaying the whole test program.
However, in the second transmission, the first gear and reverse were run until runout. During inspection, pits were found on the edge of the output gear again, but never on the reverse or drive gear. Only two transmissions were tested with convoloids.
Second gear (Astaloy 85Mo 0.25%C 7.25 g/cc)
The second gear was asymmetric, and it went all the way to runout. But during inspection, it showed some pitting on the output gear — half of the teeth showed pitting on the edge. It looked like a concentricity or axial alignment problem. The same resulted for gearbox number two. Only two transmissions were tested with asymmetric gears.
Gears 3-6 (Distaloy AQ + 0.25%C and Hipaloy + 0.25%C 7.25-7.5 g/cc)
The rest of the gears passed the testing with mild wear that can be considered as perfectly normal.
The test cycle for vibration is shown in Figure 8. Both drive and coast side were tested at different torque and speeds.
Each ramp was repeated six times, and the vector sum of the vibrations were calculated and averaged. The collected data was then transformed from the time domain to the frequency domain using FFT, and the acceleration data in the frequency domain is depicted in Figure 5 for the fourth gear pair. In Table 4, there are some relative numbers calculated comparing original gears in solid steel to the PM gears, and as shown, PM gears outperform the steel gears.
However, it must be noted that the PM gears have been redesigned to increase the contact ratio with smaller modulus, higher number of teeth, and higher helix angle. Also, the root has been redesigned and improved to reduce stresses (see Table 3). The new root geometry is not possible to hob, but since the gear was made in a compaction tool, the elliptic root geometry was built into the tooling.
From Figure 9, it can be seen that the vibrations are higher on the transmission itself than on the differential housing. The solid lines are measured on a brand-new OEM transmission, and the dashed lines are from a new PM transmission. Housing, shafts, and differential were the same for both tests.
Gears 4, 5, and 6 were tested until breaking occurred. However, some of the gears did not break and the test was stopped to protect the final drive and rig.
- Fourth gear (from compaction tool): new gear stopped at 1154 Nm (no failure); old gear durability tested 986 Nm (failure)
- Fifth gear (machined from blank): old gear durability tested 1160 Nm (failure)
- Sixth gear (machined from blank): old gear durability tested 1000 Nm (no failure, stopped to protect rig and final drive)
All gears (1-6 + R) were run at 450 Nm for a minimum of 1,000 revolutions to simulate abuse, and all gears passed despite having just finished the complete durability cycle before testing at 450 Nm.
Gear designs in S-N Diagram
Figure 10 shows where the individual gears are located with respect to the fatigue limit. The S-n lines are plotted at a 99-percent confidence level and generated on an FZG back-to-back gear tester (pitting) and a Vibrophore machine (tooth root bending fatigue).
Looking at the contact fatigue graph in Figure 10, it is evident that there will be some pitting on some of the gears. It was reasoned that some pitting may be allowed at the end of service life on first and second gears. However, for the fifth gear, pitting is not acceptable. For that reason, the density was increased on the fifth gear pair. This moves the S-n curve according to the dashed line in the contact fatigue S-n graph, placing the fifth gear pair just on the S-n curve. For the same reason, the sixth gear pair was made at a higher density (see Table 1). For tooth root bending, an additional safety margin is required, since tooth root breakage is a failure mode that is unacceptable. That is why higher density is used for some of the other gears as well. This means that, for some of the gears, the S-n curve is pushed toward the upper right corner of the tooth root bending fatigue graph, giving a better safety margin than what is actually depicted. The solid line S-n curves are generated for 7.25 density, and the use of elliptical tooth root is not accounted for. Only the fourth drive gear had a modified root, since it’s impossible to hob under production-like conditions.
A 6-speed manual transmission has been designed for powder metal. Four transmissions have been built: three for testing and one implemented in a Saab 95. The transmissions have been tested for durability, vibration, overload, and breaking load with successful outcomes at 230 Nm, which is the maximum output torque from the engine in the Saab 95 1.6L Turbo. Higher torques can be sustained, but that requires densification of surfaces, which has been tested in a different project.
The breaking loads for gears 3-6 were 986 Nm or higher.
The abuse load was set to 450 Nm, and all gears passed even after having completed full durability testing.
In vibration testing, the PM gears showed lower amplitude levels than the original steel gears.
It can’t be concluded that the 40 point FZG S-n data can be used for design. More test data from the M32 is necessary to make such a claim. However, no results point toward the opposite. On the contrary, the data and calculations indicated that some gears would pass while others were borderline. In this case, a 99-percent confidence level was used, which is quite conservative. The pits found on the first and second gears were related to design or manufacturing rather than ultimate fatigue, and no tooth root failures occurred.
The design work and testing work was performed by Vicura AB on behalf of Höganäs AB.
- Flodin. A, Andersson. M. Prototyping of Automotive 6 Speed Manual Transmission in Powder Metal — How does it compare to the original transmission. Proceedings World PM Orlando 2014, paper No 272.
- Flodin. A. PM conversion of 6 speed manual in SAAB 9-5 -Lessons learned in prototyping, Proceedings EuroPM Salzburg 2014.