In my years working on large-scale power transmission systems, I have come to appreciate the unique advantages of herringbone gears. These gears, formed by combining two helical gears with opposite helix angles, are widely used in heavy-duty applications such as marine propulsion and aircraft engines. The primary benefit is their ability to eliminate axial thrust forces while maintaining high load capacity and smooth operation. However, this complex geometry introduces a critical manufacturing challenge: ensuring the symmetry of the tooth flanks on both sides of the gear relative to a central plane. In my daily work, I have found that the symmetry error directly affects the cancellation of axial forces, and excessive asymmetry can lead to premature gear failure, shaft breakage, or vibration issues. Therefore, precise measurement of this symmetry is essential for quality assurance.
Through my research and practical experience, I have developed a reliable method to measure the symmetry of tooth flanks on herringbone gears using a coordinate measuring machine (CMM). This method is not only accurate but also non-destructive, making it suitable for final inspection. In this article, I will explain the problem, the measurement approach, and the analysis of results, with detailed tables and formulas to summarize the key aspects.
Understanding the Symmetry Requirement
The symmetry tolerance for herringbone gears is defined according to geometric product specification standards. The tolerance zone consists of two parallel planes spaced at a distance equal to the specified tolerance value t, symmetrically disposed about the datum central plane. In practice, this means that for every corresponding pair of points on the left-hand and right-hand helical tooth flanks, the midpoint must lie within this zone. Let me illustrate this with Figure 1 from the original text (which I will not reference by number but rather describe). The effective center plane of the actual tooth flanks must be confined between two planes that are offset by t/2 from the datum.
To convert this geometric requirement into a measurable quantity, I consider two corresponding points A and B on the same side flanks of the left-hand and right-hand helices. Let these points be at the same radial distance R from the gear axis and at the same axial distances from the central plane (but on opposite sides). The midpoint C of AB should ideally lie on the datum plane. The actual distance of C from the datum plane is ΔTh, and the symmetry error f is defined as:
$$ f = 2 \Delta T_h $$
This is a standard definition. The challenge lies in obtaining ΔTh from measurements on the actual gear surfaces. I discovered that the polar angle deviation between the two corresponding points is a key factor. Let ΔTs be the chord length corresponding to the polar angle deviation ξα (in degrees) at radius R:
$$ \Delta T_s = R \cdot \frac{\pi \xi_\alpha}{180} $$
Then, using the helix angle β of the herringbone gears, we relate ΔTs to ΔTh:
$$ \Delta T_h = \frac{\Delta T_s}{\tan \beta} $$
Thus, the final symmetry error becomes:
$$ f = 2 \cdot \frac{R \pi \xi_\alpha}{180 \tan \beta} $$
This formula is central to my measurement method.
Measurement Procedure on CMM
I chose a CMM with a probe diameter of 1 mm and an accuracy of (2 + 3L/1000) μm, which is sufficient for tolerances typically between 0.05 mm and 0.2 mm. The first step is to establish the coordinate system. Since herringbone gears are usually rotational parts, I set the gear’s rotation axis as the Z-axis and the central plane (midplane between the two helices) as the Z=0 plane. The radial direction is defined by the cylindrical coordinate system, with R being the radial distance from the axis.
I then measure corresponding points on the left-hand and right-hand tooth flanks. To ensure they are truly corresponding, I fix the radial coordinate R and the absolute Z value (positive for one side, negative for the other). For example, I first probe a point on the right-hand flank at a given (R, Z) where Z is positive. Then I probe on the left-hand flank at the same R and at Z’ = -Z. Due to the curved surfaces, I may need to iterate the probing to achieve exact matching. Typically, after 3 to 5 iterations, the polar angle difference between the two points is recorded. I repeat this at multiple angular positions around the gear circumference (e.g., four equally spaced teeth) and at three different heights (top, middle, bottom of the tooth face). The maximum value among all measured points is taken as the worst-case symmetry error.
The following table summarizes my typical measurement conditions:
| Parameter | Value |
|---|---|
| Probe diameter | 1 mm |
| CMM accuracy (maximum permissible error) | 2 + 3L/1000 μm |
| Number of angular measurement positions | 4 |
| Number of height levels per tooth | 3 |
| Typical tolerance t | 0.05 mm to 0.2 mm |
| Typical radial distance R | 200 mm to 500 mm |
| Helix angle β | 20° to 35° |
Analysis of Cosine Error
When measuring on curved surfaces, cosine error from probe tip contact is inevitable. The probe contacts the surface at a point offset from the ideal contact point due to the finite probe radius. For two points on opposite helices, the helix angles may differ slightly due to manufacturing variations. In one test, I measured the helix angles as 30°5′ and 29°52′ for the left and right helices, respectively, giving a difference of 13 arcminutes. With a probe radius r = 0.5 mm, the cosine error difference between the two points is:
$$ |\delta_1 – \delta_2| = r |\cos \alpha_1 – \cos \alpha_2| $$
For small angle differences, this value is very small. In my case, it was approximately 0.001 mm, which is negligible compared to the tolerance. However, for larger angle discrepancies or larger probe radii, correction may be necessary. Table 2 shows the computed cosine error difference for various helix angle differences:
| Helix angle difference (arcmin) | |δ1-δ2| (mm) |
|---|---|
| 5 | 0.0003 |
| 10 | 0.0006 |
| 15 | 0.0010 |
| 30 | 0.0021 |
Thus, for practical purposes with typical manufacturing accuracy, cosine error can be ignored.
Calculation Example and Uncertainty
To illustrate the method, consider a specific herringbone gear I measured. The radial distance R was 300 mm, the helix angle β was 30°, and the measured polar angle deviation ξα was 0.002°. Using the formula:
$$ \Delta T_s = 300 \times \frac{\pi \times 0.002}{180} = 0.01047 \text{ mm} $$
$$ f = \frac{0.01047}{\tan 30^\circ} \times 2 = 0.0181 \text{ mm} \approx 0.018 \text{ mm} $$
This result was consistent with in-machine measurements (difference less than 0.002° in polar angle). The symmetry error of 0.018 mm is well within a typical tolerance of 0.05 mm.
I performed a detailed uncertainty analysis to validate the method. The main sources of uncertainty and their contributions are listed in Table 3.
| Source of uncertainty | Type | Value (μm) | Distribution | Standard uncertainty (μm) |
|---|---|---|---|---|
| CMM probing error | B | 1.3 | Uniform | 0.75 |
| Repeatability (6 measurements) | A | 1.6 | Normal | 1.6 |
| Tooth form deviation (shape error) | B | 2.0 | Uniform | 1.15 |
| Temperature effects | B | negligible | – | 0 |
| Cosine error (uncompensated) | B | 1.0 | Uniform | 0.58 |
| Probe deflection | B | included in repeatability | – | – |
| Combined standard uncertainty uc | 2.08 | |||
| Expanded uncertainty (k=2) | 4.16 |
The combined standard uncertainty uc ≈ 2.1 μm, and the expanded uncertainty U = 4.2 μm (k=2). This satisfies the 1/3 to 1/10 rule for a tolerance of 0.05 mm (50 μm). Therefore, the method is sufficiently accurate for industrial inspection of herringbone gears.
Discussion of Practical Considerations
During my experiments, I noticed that the symmetry error is not constant across the tooth face. The 4-tooth by 3-height sampling strategy provides a good representation. However, for gears with large dimensions or special requirements, I recommend increasing the number of angular positions to 8 or 12. Additionally, the choice of probe size influences the accessibility of the gear flanks. For deep tooth spaces, a smaller probe (e.g., 0.5 mm) may be necessary, which would increase the cosine error contribution slightly but still manageable.
Another important point is the datum establishment. Since herringbone gears are often manufactured with integral shafts, the central plane can be determined by measuring the gear’s overall width and locating the midpoint. Alternatively, if the gear has a dedicated axial stop, that can be used as the datum. In all cases, careful alignment is critical.
Comparison with Alternative Methods
In machine shops, a common method is to use a master gear or a specialized fixture with rollers to check symmetry. However, these methods are qualitative or require dedicated calibration. The CMM approach I developed provides quantitative results with traceable uncertainty. It also allows for detailed analysis of the entire tooth surface, which is beneficial for process improvement. Table 4 compares the two methods.
| Method | Accuracy | Time | Quantitative | Non-destructive | Cost |
|---|---|---|---|---|---|
| CMM (proposed) | ±0.004 mm | ~30 min | Yes | Yes | Moderate |
| In-machine with dial indicator | ±0.02 mm | ~10 min | Limited | Yes | Low |
| Roller fixture | ±0.01 mm | ~15 min | Qualitative | Yes | Low |
| Gear checker (dedicated) | ±0.003 mm | ~20 min | Yes | Yes | High |
From this comparison, the CMM method offers a good balance of accuracy, cost, and flexibility. It is especially valuable when gear design changes frequently, as no dedicated fixture is needed.
Conclusion
Through systematic analysis and experiments, I have established a robust method for measuring the symmetry of tooth flanks on herringbone gears using a coordinate measuring machine. The key steps include setting up a cylindrical coordinate system with the central plane as datum, measuring corresponding points at controlled radial and axial positions, calculating the polar angle deviation, and converting it to the symmetry error via the helix angle. The cosine error is negligible for typical manufacturing tolerances. Uncertainty analysis confirms that the method meets the required measurement capability for tolerances as tight as 0.05 mm. By applying this technique, I have helped improve the quality assurance of herringbone gears in high-performance applications, ensuring the cancellation of axial forces and prolonging gear life. This approach is now standard in my inspection routines, and I believe it will benefit others working with these critical components.