Bearing Shoulder Screws

Custom 303 Stainless Bearing Shoulder Screws

Standard catalog components sometimes fall short when prototyping custom mechanical assemblies, necessitating custom-machined parts to hit functional tolerances. I encountered this constraint while building a design prototype requiring corrosion resistance and an exact shank diameter and length to achieve a precise clearance fit for sliding bearings. An exhaustive search through standard McMaster-Carr shoulder screw inventories came close, but nothing matched the exact dimensional envelope required. Rather than modify the assembly geometry, I engineered and manufactured the custom screws utilizing available shop resources. This breakdown details the complete manufacturing process—from CAD modeling and manual G-code writing to tight-tolerance turning on a Haas SL-20 lathe.

Having already modeled the part in SolidWorks for the broader assembly design, I evaluated whether to import the file into Mastercam for automated toolpaths or to hand-write the G-code. Because the geometry seemed relatively straightforward and I would be running the job on a Haas SL-20 lathe, I decided to challenge myself and write the code manually. To streamline the process, I generated a 2D reference drawing in SolidWorks featuring ordinate dimensions measured from the end face of the threaded boss as the zero datum to simplify coordinate mapping and minimize any calculation errors.

For corrosion resistance paired with good machinability, I selected 303 stainless steel, as its added sulfur content significantly improves chip-breaking and machinability compared to other stainless grades. Additionally, 303 is readily available in standard hex bar stock, allowing the raw material to serve directly as the screw head without secondary milling operations. For this application, I used a 5/16-inch across-flats (WAF) hex bar, which precisely accommodates a standard 5/16-inch hex socket.

Recognizing that the part's length exceeded three times its diameter, I knew a tailstock live center would be essential to maintain the tight shank tolerances. To optimize precision, I planned the manufacturing process in two distinct operations: first, machining the threaded boss and cutting the threads without the tailstock, and second, utilizing the live center to support the part during precision turning of the bearing shank.

Selecting the right tooling for 303 stainless steel at 25 HRC meant opting for durable carbide inserts to handle the material. The job ultimately required a set of five tools: an 80-degree turning and facing tool for roughing and finishing the boss and shank, a 35-degree turning and facing tool to machine the relief groove behind the threaded boss, a 60-degree threading tool for cutting the #10-32 threads, a #1 center drill for the live center mount, and a 1/8-inch parting tool. A primary challenge involved arranging the turret layout; because the stock stick-out was relatively short for the threading operation, I had to carefully map out tool positions to prevent interference with the lathe chuck. Since the machine was actively shared across other shop projects, stripping down the entire turret wasn't an option. Through several iterations of swapping tools, redirecting coolant lines, and jogging the turret to check clearances, I locked in a functional configuration that avoided collisions.

With the tooling and procedure finalized, I turned to writing the G-code. While I typically rely on the Feed and Speed Wizard app and tune parameters on the fly based on surface finish, I opted here for slightly more conservative values recommended by NYCCNC for turning 303 stainless steel. This adjustment was particularly wise given the interrupted cuts inherent to working with hex stock. For roughing and finishing operations, I ran a constant surface speed (CSS) of 353 SFM, a feed rate of 0.005 inches per revolution, and a 0.035-inch depth of cut. I programmed the threaded boss using a G71 Type 1 roughing cycle followed by a G70 finishing cycle, supplemented by a dedicated pass for the relief groove behind the boss. To ensure a precise fit, the thread boss diameter was calculated directly from Machinery’s Handbook, sizing it right at the midpoint between the minimum and maximum limits for a 2A thread class.

Given the tight 0.05-inch width of the relief groove behind the threads, tool clearance was a major concern. Operating on a shared machine raised the stakes regarding potential crashes—particularly with tool nose radius compensation active, where a minor coordinate error could prove catastrophic. To verify the setup, I first modeled both the tool and the boss in SolidWorks. While that helped map out the coordinates, it still wasn’t enough to completely calm my nerves. To be entirely certain, I imported the model into Mastercam to generate the toolpaths for a direct comparison, ultimately updating my hand-written G-code with the Mastercam coordinates before running the part.

Next came the screw threads, programmed using a G76 threading cycle. To determine the parameters, I relied on threading formulas from a past CNC lathe class which I had converted into a custom calculation spreadsheet featuring drop-down menus and embedded dimension tables from Machinery’s Handbook. Threading presented a tight challenge due to the narrow relief groove and the risk of a crash against the shank. To verify clearances, I modeled the cutter in SolidWorks, using a Keyence microscope to precisely measure its actual angles and geometry, alongside downloading Iscar's CAD model of the threading insert for comparison. As a final safeguard, I ran the threading operation through Mastercam to cross-reference the output coordinates.

Mounting the stock turned out not to be straightforward, as the material wouldn't fit the standard setup; fortunately, a set of custom soft jaws happened to be laying around the shop that barely accommodated it. Once machining began, however, I ran into a critical hurdle during the threading operation. Initially, I ran the spindle in the low 2000 RPM range. Because spindle speed and feed rate are tightly linked during threading, the extremely narrow relief groove provided very little distance for the axis servos to decelerate and retract before hitting the shoulder. This resulted in servo lag at the end of the pass, which botched the thread profile right at the shank (seen in the picture). Diagnosing the root cause took careful testing; by systematically reducing the spindle speed down to around 1,000 RPM and below, I gave the drives adequate time to execute the pull-out move, ultimately achieving clean, consistent, and repeatable threads.

For the center-drilling operation to mount the live center, I ran at approximately 2,200 RPM with a feed rate of 0.001 inches per revolution. The shop stocked non-carbide HSS #1 center drills rather than carbide alternatives. Given the small tool diameter and the material properties of 303 stainless, my initial concerns about tool breakage proved entirely valid; one screw was ultimately scrapped when a center-drill tip snapped off in the part. However, since I only needed a small batch of four and was working within existing shop inventory, I made do with the HSS drills to successfully finish the run.

With the part securely supported by the tailstock live center, machining the bearing shank proceeded smoothly. I applied the same proven parameters—G71 roughing and G70 finishing cycles running at 353 SFM, a 0.005 IPR feed rate, and a 0.035-inch depth of cut. The final part-off operation was similarly straightforward, cleanly separating the completed component from the bar stock.

Although I could have utilized a separate tool wear offset for the finishing pass—allowing me to start further from the part and incrementally dial in the shank diameter—it proved entirely unnecessary. Careful setup and programming paid off, as I hit the critical target shank diameter spot-on during the first attempt.

To ensure repeatable multi-part production across the two operation sets, I utilized separate work coordinate systems, offsetting G55 in Z from G54 by the exact component length. Because strict length tolerances weren't required, I improvised a manual stop using the parting tool: after facing the stock at G54 Z0 and completing the first operation set, I indexed the front of the parting tool to G55 Z0 and manually pulled the bar stock out against the cutting edge. While this method was slightly finicky given the lack of a flat reference surface on the screw ends, it provided sufficient repeatability to successfully complete the batch.

Reflecting on the process, one area for optimization would be leaving a slight stock allowance on the shank diameter, or terminating the contouring pass cleanly at the threaded boss groove during the first set of operations. Because the shank was partially machined across both setups, a subtle transition mark appeared where the tool paths overlapped. However, since these components function strictly as bearing support shafts rather than high-tension fasteners subject to cyclic fatigue, the blemish remained entirely cosmetic with no functional impact on performance. The finished fasteners met all critical dimensional requirements and performed reliably in their final application.