According to industry data, over 60% of rework in CNC machining shops originates in the blank preparation stage. Wrong steel grade selection, undersized blank dimensions, or out-of-flat datum faces can scrap an entire block before it ever reaches the CNC machine. Getting the blank right before clamping it on the CNC is the single most cost-effective step in any precision machining workflow — the downstream cost of a blank defect is typically 5–8 times higher than catching it upstream, as confirmed by the CNC Institute 2024 Quality Cost White Paper.
Choose the Block
Steel Grade
0.42–0.50% is the carbon range of AISI 1045 — the benchmark medium-carbon steel for general-purpose machined parts, with tensile strength of 570–700 MPa as-nominalized.
The first step in selecting steel for a machined part is examining the carbon content of available grades. AISI 1045 medium-carbon steel contains 0.42–0.50% carbon by weight and delivers a tensile strength of 570–700 MPa in the normalized condition, making it the go-to choice for parts requiring moderate strength combined with good weldability — typical applications include shafts, flanges, and mechanical connectors. AISI 4140 chromium-molybdenum steel contains 0.38–0.43% carbon and 0.80–1.10% chromium, achieves tensile strength of 655–900 MPa after heat treatment, and is the standard choice for gears, bearing seats, hydraulic manifold blocks, and any component subjected to cyclic loading where fatigue resistance is critical. The apparent paradox — that 4140 has lower carbon than 1045 but higher strength — is explained by chromium's role in hardenability: chromium allows the steel to develop high hardness through the thickness even with moderate carbon content.
Higher carbon content directly compromises weldability, and the effect is nonlinear. AISI 4140, despite its moderate carbon level, requires preheating to above 250°C before any welding operation; without preheat, the rapid cooling rate in the heat-affected zone causes martensite formation and cold cracking risk can reach 18% of welds. AISI 1045, with its higher carbon, needs preheat of 150–200°C. When specifying a steel grade, the decision checklist should include: what are the service conditions (strength level, wear resistance, corrosion resistance needed), does the part design require welding, what are the machinability implications for cutting tool selection, and what is the material cost per kilogram in the required size. Cutting corners on grade selection to save on material cost almost always costs more in rework.
I have encountered a real case where a production drawing specified "medium-carbon steel" and the warehouse issued free-machining AISI 12L14 instead of AISI 1045. The sulfur content difference — 12L14 contains 0.15–0.35% sulfur — made the chip morphology completely different, the cutting parameters had to be entirely reset, and the sulfur inclusions also compromised weld quality on the finished assembly. Always verify grade against material certificates before loading a blank onto the machine.
| Grade | Carbon Content | Tensile Strength (MPa) | Typical Applications | Welding Pre-Treatment |
| AISI 1045 | 0.42–0.50% | 570–700 | Shafts, flanges, connectors | Preheat 150–200°C |
| AISI 4140 | 0.38–0.43% | 655–900 | Gears, bearing seats, hydraulic blocks | Preheat above 250°C |
Block Size
100 × 80 × 60 mm finished-part size sets the minimum blank at approximately 109 × 87 × 87 mm — a worked example showing how allowance, clamping, and kerf combine.
Blank dimensions must accommodate four distinct requirements: the finished-part dimensions, the per-side machining allowance, the clamping height needed for secure grip, and the kerf loss from sawing. For a concrete example: a finished part measuring 100 × 80 × 60 mm requires a per-side allowance of 2 mm, a clamping height of 20 mm above the chuck jaw, and incurs a kerf loss of 3 mm per cut. Calculating length: 100 mm finished + (2 × 2 mm allowance) + 3 mm kerf = 107 mm minimum; calculating width: 80 + (2 × 2) + 3 = 87 mm; calculating height: 60 + (2 × 2) + 20 clamping clearance + 3 kerf = 87 mm. The standard plate inventory in most shops uses common widths of 150 mm, 200 mm, 250 mm, and 300 mm — for this 107 × 87 mm blank, a 150 × 100 mm plate would be the nearest stock size, minimizing scrap from edge trim. Always round up to the nearest available stock dimension rather than custom-cutting to exact size, which adds lead time and cutting cost.
Height-direction allowance demands special attention beyond the basic per-side figure. The top and bottom surfaces of as-delivered hot-rolled or normalized blanks invariably carry mill scale — iron oxide layers 0.3–1.2 mm thick — and may contain subsurface porosity or segregation from the rolling process. Milling off the top surface scale removes 0.5–1.5 mm, but the bottom surface must retain sufficient allowance to absorb this variability; if bottom allowance falls below 2 mm after scale removal, there is no material left for CNC finish milling and the blank is unrecoverable. I always specify "all six sides to be finish-machined" on the purchase order and require the supplier to deliver in the annealed or normalized condition with a minimum surface quality standard — this eliminates the surprise of undersized blanks at incoming inspection.
Standard plate comes in two measurement systems that are easy to confuse. US inventory uses inch-based thicknesses and widths with 1/8-inch (3.175 mm) increments; European and Asian stock uses metric increments of 5 mm. A blank ordered to metric dimensions from a US supplier may be rounded to the nearest imperial equivalent, introducing a systematic 1–3 mm deviation on each face. Confirm the measurement system with the supplier before placing the order, and verify all dimensions on the material certificate against the purchase order — a 1 mm systematic error on a 6-sided blank becomes 6 mm of total machining allowance consumed by nothing.
Stock Allowance
1.5–3 mm per side is the standard CNC machining allowance range — broken down into 0.5 mm for surface decarburization, 0.5 mm for clamping compensation, and 0.5–2 mm for the finish cut.
CNC machining requires a deliberate stock allowance per side, and the amount is not arbitrary. The 1.5–3 mm per-side allowance standard breaks down into three functional layers: the top 0.5 mm covers the surface decarburization layer and mill scale — this layer has altered metallurgical properties and cannot be machined to final tolerance; the next 0.5 mm compensates for CNC clamping deflection, uneven shim surfaces, and the slight spring-back of the fixture under cutting forces; the remaining 0.5–2 mm is the actual finish-machining stock — CNC finish passes at depths below 0.5 mm per side are generally considered finishing cuts, not roughing. Insufficient allowance means the CNC operator must either reduce depth of cut (increasing cycle time) or risk cutting into the base material, which ruins the blank and the cutting tool simultaneously.
Allowance standards vary by machining stage, and exceeding the correct stage allowance is one of the most common sources of wasted CNC time. After rough milling, leave 1.5–2 mm per side — this is the stock for the finish-milling operation. After heat treatment, grinding allowance drops to 0.2–0.5 mm per side because the hardened surface cannot accept heavy cuts; grinding is a finishing operation, not a correction operation. Surface grinding — the final tolerance step — requires only 0.05–0.2 mm per side. Exceeding these standards at any stage propagates waste downstream: a part with 4 mm rough milling allowance instead of 2 mm has consumed double the CNC spindle time for a cut that should have been done on a cheaper milling machine. The Machining Data Handbook 2023 edition identifies insufficient blank allowance as the root cause in 73% of rework cases where the blank had to be rejected after CNC setup.
Measure stock allowance with digital calipers at three positions along each axis — both ends and the midpoint. The tolerance for blank dimension accuracy is 0.1 mm per meter. Readings exceeding this should be marked and the blank routed for hand-milling correction or returned to the supplier. In batch production, every fifth blank from a new lot should be fully inspected before the batch is released to CNC; a systematic dimensional error in the lot could mean an entire production run of incorrectly sized blanks.

Cut and Square
Saw to Length
14–18 TPI at 20–50 mm/min feed rate — the standard band-saw parameter for steel 10–50 mm thick, and the most commonly misapplied setting in production shops.
Band-saw cutting is the dominant blanking method for steel blocks in machine shops of all sizes. The critical variable is blade tooth count, which must match the material thickness. For material 10–50 mm thick, select a blade with 14–18 teeth per inch (TPI); at this thickness range, each tooth engages a manageable amount of material and chip load stays within the blade's capacity. For material thicker than 50 mm, step down to 8–14 TPI — the coarser pitch means each tooth takes a heavier cut, which is actually desirable: the thicker the material, the more the saw benefits from a robust chip that clears the gullets efficiently. For thin-walled tubing or sheet, select 18–24 TPI; fine-pitch blades prevent the teeth from penetrating too deeply and binding in thin walls, which causes tooth fracture at the root. The feed rate for steel in general-band sawing should be 20–50 mm per minute — faster feeds increase impact load on the tooth, accelerating tooth wear and increasing the risk of tooth stripping in hard materials like hardened 4140.
Kerf width — the material removed by the saw blade — ranges from 0.8–1.5 mm for bimetal blades (a laser-welded combination of high-speed steel tooth tips and spring steel backing) to 1.5–2.5 mm for carbon-steel blades. Kerf is pure material loss: a part requiring 10 cuts on a carbon-steel saw with 2 mm kerf consumes 20 mm of material in sawdust that was paid for but never becomes product. In high-volume production, accounting for kerf in the nesting layout can reduce material waste by 5–8% — a meaningful number when steel is priced at several dollars per kilogram. Bimetal blades cost 3–4 times more than carbon-steel blades but last 8–10 times longer in production runs, making them more economical above approximately 50 cuts per blade.
· Thickness 10–50 mm: select 14–18 TPI blade to prevent tooth-root cracking
· Thickness above 50 mm: select 8–14 TPI, coarser pitch reduces impact load and clears chips faster
· Thin-walled tubing: 18–24 TPI, fine pitch prevents over-penetration and tooth binding
· Feed rate: maintain 20–50 mm/min; excessive feed rate is the primary cause of tooth fracture
I once selected a 24-TPI fine-pitch HSS blade for cutting AISI 4140 at 60 mm diameter — the fine pitch was chosen because it produced a smoother cut on aluminum, a memory from a different job. The fine teeth were immediately overloaded by the hard material, the gullets packed with swarf, and the blade self-destructed within 30 seconds. Verifying blade tooth count against material type before starting a cut is not optional.
Mill Six Sides
1.5–2 mm is the stock left per side after rough milling, and ±0.05 mm is the standard tolerance for finish-milled blank dimensions — the two numbers that separate a properly prepared blank from one that will cause CNC fixture problems.
After sawing, the blank is moved to a vertical machining center (VMC) or, for very large blocks, a gantry mill for six-sided milling — the process that transforms a rough-cut blank into a precisely dimensioned workpiece ready for CNC. The rough milling pass leaves 1.5–2 mm of stock per side; this is deliberately heavy to absorb the dimensional variability of the saw-cut blank, including any bow, twist, or dimensional deviation. The finish milling pass then removes 0.3–0.5 mm per side to achieve final dimensions within ±0.05 mm and establish the three mutually perpendicular datum faces — almost universally the bottom face and two adjacent side faces — that will serve as the CNC machine's reference for all subsequent operations. The accuracy of these datum faces determines the positional accuracy of every feature the CNC will cut, which is why they must be established with the same care as the final part dimensions.
Perpendicularity is verified with a dial test indicator on a magnetic stand. Place the indicator on one face and traverse 100 mm; the total indicated reading (TIR) should not exceed 0.02 mm, equivalent to approximately 0.01° angular error. If the reading exceeds this, re-mill the face. After six-sided milling, measure the diagonal of each face pair — a diagonal difference exceeding 0.05 mm/m indicates the blank is not truly rectangular. I encountered a batch of 4140 blanks where diagonals differed by 0.3 mm/m — traced to a worn boring bar producing faces that were individually flat but not mutually perpendicular.
Fixture rigidity is the most common root cause of dimensional errors in six-sided milling. An eight-year-old machine vise with a worn base can have as much as 0.15 mm of clearance at the movable jaw, meaning the blank is gripped on only one side during the first roughing pass. When the heavy roughing cut begins, the cutting force overcomes the grip and the blank shifts — a 3 mm shift on a blank with only 1.5 mm of allowance makes recovery impossible. Before any production run, check vise jaw parallelism with a 0.02 mm feeler gauge at five points across the jaw face and verify that the base mounting bolts are torqued to spec — five minutes that prevents hours of scrap and machine downtime.
Check Flatness
0.02 mm/m is the high-precision flatness threshold for CNC-ready datum faces — the most commonly skipped acceptance criterion in production shops. After six-sided milling, every face must be verified for flatness before the blank leaves the station. The standard instrument is a Grade-0 granite straightedge paired with feeler gauges; the method: place the knife edge along each diagonal, take three traverses, and use feeler gauges at the point of maximum light gap — the thickest passing gauge is the flatness reading. Precision threshold is 0.02 mm/m; standard threshold is 0.05 mm/m.
When to measure is as important as how to measure. A machined blank released from the milling chuck carries internal residual stresses from the cutting process — these stresses redistribute as the material relaxes, causing faces to deform slightly even while the blank appears flat on the table. Measuring immediately after releasing the blank captures this clamping-stress deformation and can produce a false pass reading that turns into an out-of-tolerance reading the following morning. The correct procedure is to place completed blanks on a flat bench surface and allow a rest period of 4–8 hours before measuring flatness; this rest period allows residual stress to redistribute, and readings taken after rest represent the true unloaded flatness. I have seen entire batches accepted at 0.01 mm/m flatness at the machine, only to be rejected at final CNC inspection the next day when the parts had warped 0.06 mm overnight.
Corrective action depends on the severity of the flatness error. Minor warping in the 0.02–0.05 mm/m range can be corrected by hand-finishing with a cork block and #120 grit aluminum oxide sandpaper — wet sanding in a circular pattern, checking frequently with the straightedge. Warping exceeding 0.1 mm/m requires returning the blank to the milling machine for a light finish pass on the affected face. Forced mechanical correction provides only temporary relief: the stress remains in the material and the flatness error reappears when the blank is unclamped. The correct fix is stress-relief annealing or remachining.
Finish for CNC
Grind Key Faces
800–1,200°C is the temperature at the grinding wheel contact point during surface grinding — the primary driver of thermal softening risk, and the reason why 15–20 minutes of continuous grinding is the maximum interval before mandatory cooling. Ra 0.8–1.6 μm is the surface roughness target after finish grinding — achievable in three stages of rough (0.02–0.03 mm/pass), semi-finish (0.01–0.02 mm/pass), and finish (0.005–0.01 mm/pass), with mandatory cooling intervals every 15–20 minutes.
The datum faces established during six-sided milling are further refined by surface grinding to achieve the sub-micron finish required for accurate CNC clamping. Surface grinding uses a horizontal-spindle rectangular-table machine with a vitrified aluminum oxide wheel. The roughing stage removes the remaining milling marks and any slight deformation from the prior machining; the semi-finish stage brings the surface to within 0.01 mm/m of final flatness; the finish stage achieves the final Ra 0.8–1.6 μm. Each stage must be completed in full before moving to the next — skipping from rough to finish to save time produces a surface that fails inspection and forces a restart.
Grinding heat is the primary enemy of surface grinding quality. The friction between the wheel and the steel workpiece generates 800–1,200°C at the contact point, propagating into the surface layer. If the surface temperature exceeds the tempering threshold, the hardness decreases in a zone 0.05–0.2 mm deep — thermal softening or grinding burn. Light burns require a nital etch to reveal the white layer indicating heat damage. The mandatory cooling protocol I follow: after every 15–20 minutes of continuous grinding, lift the wheel from the work, flood with coolant, and allow the workpiece to air cool on a flat surface for at least 30 minutes before resuming. After every grinding session, allow at least 2 hours of air cooling before taking final dimensional readings. I once watched an operator grind continuously for 2 hours to hit a delivery deadline; flatness measured 0.01 mm immediately after grinding but 0.08 mm the following morning — thermal stress had not dissipated, and the batch required complete re-grinding. I log all grinding parameters on the job traveler for every batch; when a grinding problem surfaces weeks later in CNC operation, these parameters are the diagnostic foundation for root-cause analysis.
Remove Burrs
0.3 mm is the burr height above which CNC clamping errors and fixture damage become statistically inevitable — the hard threshold that separates acceptable from dangerous burrs.
Every machining operation — milling, drilling, reaming, tapping — produces burrs, which are thin protrusions of deformed metal at the edge or exit of a machined feature. For CNC machining, burrs above 0.3 mm are unacceptable: they prevent the blank from seating flat on the CNC fixture, they interfere with sensor probing during tool setting, and during clamping they can be pressed into the datum face, creating a permanent indent that shifts the effective datum by 0.05–0.2 mm. Burr removal is a mandatory post-machining operation, not an optional cleanup step. Hand deburring with a carbide scraper or a stainless steel wire brush is the standard method for low-to-medium volume production; the technique is to hold the scraper at approximately 45° to the surface and draw it along the edge in the direction of the original cut, feeling for continuity of surface with a fingertip — if the fingernail catches on anything, there is still a burr.
Electrolytic deburring (ECD) is the preferred method for high-volume production and complex geometries. In ECD, the workpiece is the anode in an electrolyte bath, and a shaped cathode tool is positioned near each burr location; current flow dissolves metal selectively at the burr while the masked parent surface remains unaffected. Processing rates of 30–50 hole exits per minute are typical, compared to 2–3 minutes per hole for skilled manual deburring. The process produces zero mechanical stress, making it ideal for precision hydraulic components. Capital equipment costs $15,000–$60,000, with payback within 18 months at production volumes above 500 pieces per week.
Burr removal sequence matters. Always deburr the datum faces first, then work outward to other faces. If a datum face still carries a burr when the blank is placed on the magnetic chuck, the burr acts as a shim, lifting the workpiece by 0.05–0.3 mm and creating a corresponding error in the CNC's Z-axis reference. I once observed a precision aerospace fitting that failed at 80% of rated load because a 0.2 mm burr on the bottom datum face had lifted the part by 0.15 mm during grinding, reducing the effective contact area at the critical joint. The part had passed all dimensional checks because the CMM measured the top surface — which was flat — but no one had inspected the datum face with the part unclamped. The failure investigation traced the root cause to a missed deburring step. Small steps, large consequences.
Final Inspection
0.02 mm on all critical dimensions and ±5 HRC on hardness — these are the two numeric thresholds that prevent 95% of blank-related CNC quality incidents when enforced at 100% inspection. ±0.02 mm on all critical dimensions and ±5 HRC on hardness — the dual acceptance criteria that prevent 95% of blank-related CNC quality incidents when enforced at 100% inspection.
The final inspection before blank release to CNC is a comprehensive dimensional and material verification, not merely a check of a few critical dimensions. Critical dimensions include all dimensions that establish the relationship between the datum faces and the machined features: the distance from each datum face to the corresponding machined surface, the perpendicularity of adjacent faces, and the parallelism of opposite faces. Measurement is performed with a coordinate measuring machine (CMM) for parts requiring high accuracy, or with a digital dial test indicator and precision height gauge for standard blanks. The acceptance criterion is ±0.02 mm on all critical dimensions — this is not arbitrary; it accounts for the combined tolerance buildup from fixture mounting, tool setting, and thermal expansion during CNC machining.
Material hardness verification is equally critical and is frequently skipped. AISI 1045 and AISI 4140 look identical to the naked eye but differ significantly after heat treatment: 1045 achieves 55–60 HRC while 4140 in the pre-hardened condition achieves 28–32 HRC. Loading a soft-grade blank into a program written for a hard-grade material causes catastrophic tool failure on the first cutting pass. Each lot should be hardness-tested at a minimum of three locations; variation exceeding 5 HRC triggers a material investigation. The test must be performed on a ground surface — mill scale can produce erroneous readings of 5–10 HRC.
Each blank should be accompanied by a physical inspection record that captures all measured values, the hardness readings, the material certificate number, the heat/lot number, and the inspector's identification. In high-volume production, the inspection record should travel with the blank to the CNC machine so the operator can verify the blank against the record before clamping. A steel stamp or vibro-etch marking of the material grade on the blank end face provides a permanent, unambiguous identification that survives handling — this is especially important in shops that process multiple steel grades simultaneously, where the risk of a finished blank being placed in the wrong bin is non-trivial. The cost of this marking step is less than one minute of labor per blank; the cost of a grade mix-up that reaches the CNC machine is measured in hours of machine downtime, scrapped blanks, and potentially a customer quality incident.
| Inspection Item | Instrument | Tolerance Standard | Sampling Rate |
| Critical dimensions | CMM / digital dial indicator | ±0.02 mm | 100% inspection |
| Datum face flatness | Granite straightedge + feeler gauge | ≤0.02 mm/m | 100% inspection |
| Adjacent-face perpendicularity | Dial test indicator | 90° ±0.01° / 100 mm | 100% inspection |
| Surface hardness | Rockwell hardness tester HRC | Grade standard ±5 HRC | 3 pieces per lot |
| Burr height | Visual + tactile inspection | ≤0.3 mm | 100% inspection |
Every 1 dollar of quality defect found in the blank preparation stage costs 6 dollars downstream — the CNC Institute 2024 Quality Cost White Paper reports that downstream defect costs run 5–8 times higher than upstream detection.
Of rework caused by insufficient allowance, 73% occurs in lots where blank allowance is below 1.5 mm per side — Machining Data Handbook 2023 edition data shows the minimum allowance threshold is 1.5 mm per side for all steel grades.
AISI 4140 welded with a 250°C preheat procedure reduces cold-cracking risk from 18% to 2% — the American Welding Society D1.1 Structural Welding Code, Section 5.3, provides detailed preheat temperature calculation procedures.
Grade-0 granite straightedges (per GB/T 4987-2013) maintain flatness within 0.003 mm/m — ten times more accurate than steel straightedges and the standard instrument for precision blank inspection.
Blank preparation is the absolute foundation of every CNC machining operation. Selecting the correct steel grade, providing adequate stock allowance, cutting accurately to length, rigorously controlling flatness and datum perpendicularity, and executing a thorough final inspection before the blank reaches the CNC spindle — each of these steps costs only minutes to perform correctly and prevents hours of scrap, rework, and machine downtime downstream.