Accuracy Iteration
Part of Machine Tools
How to progressively improve the accuracy of a machine tool through repeated measurement, correction, and refinement — the process of scraping and fitting toward precision.
Why This Matters
No machine tool emerges from the forge or foundry accurate enough to build precise work. The raw casting or fabrication has errors of millimeters; useful machine accuracy requires hundredths of a millimeter. The gap between rough construction and precision operation is bridged by a process called accuracy iteration — repeated cycles of measuring, identifying errors, making corrections, and measuring again.
This iterative approach is how every precision machine in history was originally created. Before precision equipment existed, early toolmakers used the principle of comparing surfaces to each other: two surfaces that fit together perfectly are by definition accurate to each other. By comparing three surfaces in rotation and correcting each based on the others, you can achieve near-perfect flatness without any prior reference. This is the “three-plate method” that forms the philosophical foundation of all precision metrology.
In a post-collapse environment, accuracy iteration is not a luxury — it’s the bootstrap mechanism. You start with rough tools, use them to make less-rough tools, use those to achieve moderate precision, and continue until you have machinery capable of building better machinery. There is no other path to precision without an existing precision standard to copy from.
The Three-Plate Method: Creating Flatness From Nothing
The remarkable insight: you cannot achieve perfect flatness by using one flat reference to check another (if the reference is wrong, you just reproduce its error). But you CAN achieve perfect flatness by comparing three surfaces to each other in rotation.
Why it works: Call the three plates A, B, and C. Compare A to B: any difference between them shows the combined error. Compare B to C. Compare C to A. Using a systematic scraping correction process, any repeating error (like a dome on all three) is self-correcting: a dome on A that matches B does not match C, and the discrepancy reveals itself.
Process:
- Roughly flatten three identical steel or cast iron plates by filing or grinding
- Apply blue marking compound (Prussian blue or lamp black in oil) to one plate
- Rub another plate against it with moderate pressure — high spots transfer the blue
- Scrape down the high spots with a hand scraper (a hardened steel tool with sharp triangular or curved edge)
- Re-blue and rub again. Blue now covers more area.
- Rotate through all three pairings: A-B, B-C, C-A, A-B-C…
- Continue until all three plates show uniform, consistent blue contact across the entire surface — this is true flatness
Achieving 90% bearing area (uniform contact) across a 12-inch plate requires approximately 20-40 hours of careful scraping work. The resulting plate can serve as a reference for checking all other flat surfaces in the shop.
Scraping Technique
Scraping is the fundamental hand-fitting process for achieving machine accuracy. It’s slow, requires practice, and cannot be rushed — but it can achieve surface flatness under 0.0002 inches (0.005mm) without any other precision equipment.
The scraper: A hardened tool steel blade, ground to a sharp, slightly convex cutting edge. The cutting angle is about 10-15° from the surface. Push and pull strokes both cut. The scraper must be harder than the workpiece — high-speed steel or hardened tool steel scraping annealed cast iron or medium steel.
Sharpening the scraper: Flat grind on a fine stone, then hone the cutting edge. A sharp scraper makes fine, controlled shavings. A dull scraper chatters and tears. Sharpen frequently (every 15-20 minutes of work).
Reading the blue: After rubbing the workpiece against the reference flat (with thin, even bluing):
- Small isolated high spots show as solid blue islands → scrape these down
- Uniformly distributed fine spots → nearly correct, scrape very lightly everywhere
- Large flat areas of blue with dry patches → the blue patches are the highs, dry areas are lows
- Never scrape the low areas (they’ll become lows in the next cycle)
Crossing the scraper strokes: Alternate scraper stroke direction 45° each session to avoid creating directional grooves or waves. Uniform, overlapping strokes produce consistent surface texture.
Iterating Machine Alignment
Beyond surface flatness, machine tools require that their moving axes are straight (no bow), perpendicular to each other (no squareness error), and parallel where required (no taper). Achieving these properties also uses iterative measurement and correction.
Straightness of a lathe bed: Check by running a precision cylinder (turned on the lathe itself) along the bed and measuring with a dial indicator at multiple stations. Any variation in indicator reading shows bed twist or warp. Correct by shimming the bed mounts or, for severe errors, by carefully grinding the bed ways.
Squareness of drill press column to table: Mount a test bar in the chuck. Sweep a dial indicator in a vertical circle around the test bar. If the column is perpendicular to the table, the indicator reads the same at all heights. If it reads a cycle (high on one side at the top, high on the other at the bottom), the column leans. Adjust the column angle.
The iterative measurement cycle:
- Set up the machine and measure geometric errors with available instruments
- Identify the largest error; correct it
- Re-measure — correcting one error often reveals or reduces others
- Repeat until all errors are within acceptable limits for the work you need to do
Each iteration cycle takes time but the accuracy gain per unit time is much higher early in the process (correcting large errors) than late (correcting tiny residuals). Know when to stop: the accuracy needed depends on what you’re making. For bearing bores, you need 0.001-inch accuracy. For structural components, 0.010-0.050 inch is often sufficient.
Self-Calibrating Techniques
Testing the lathe for taper: Turn a long cylinder between centers without a tool change. Measure diameter at multiple positions along the length. If diameter varies, the centers are not aligned (tailstock is offset). Adjust tailstock offset until a test cut produces parallel (constant diameter) cylinder. This uses the lathe’s own output as the accuracy reference — self-calibrating.
Squareness check with a test square: Machine a test piece (angle plate) on the surface being squared. Check with a precision square — errors in squareness show up in the machined angle. Adjust the machine setup until test pieces come out square. Again, the machine validates itself.
Pitch and lead screw accuracy: Thread a test piece and measure the thread pitch with a thread pitch gauge or precision calipers. Cumulative errors in lead screw accuracy show up over longer threads. Identify systematic errors (consistently short or long pitch) versus random errors (varying pitch) — systematic errors indicate lead screw or gear errors; random errors indicate backlash or stick-slip.
The philosophy of accuracy iteration is patience and honesty: trust the measurements, not your impression of how the work looks. A surface that looks flat can have 0.010-inch error invisible to the eye. The bluing and scraping process reveals truth that visual inspection cannot. Iterate until the instrument says it’s right, not until it looks right.