- Default to Ra 3.2 µm unless a surface has a clear functional reason to be tighter – blanket tight specs can increase costs by 15-60%.
- Same Ra doesn’t mean same surface – consider Rz, Rq, or Rt for sealing, wear, or fatigue-critical surfaces where extremes matter.
- Surface finish includes roughness, waviness, and lay direction – not just the roughness number on your chart.
- Measurement setup matters – sampling length, evaluation length, and filtering can significantly change results and should be part of your specification.
- Use drawing standards like ISO 1302 and ASME B46.1 to prevent miscommunication and inspection disputes.
CNC machining is a subtractive manufacturing process: precision tools remove material to create a final part. But once the dimensions are right, one “invisible” requirement often decides whether the part works, seals, wears well, looks premium, or ends up expensive and delayed: surface finish.
Surface finish is not just a cosmetic preference. It is a critical characteristic that can directly affect functional performance (friction, wear, sealing, fatigue), aesthetics for consumer-facing parts, and overall cost and manufacturing efficiency. That’s why surface finish charts exist: engineers need a practical way to translate the “numbers” (Ra, Rz, Rq, Rt, and roughness grades) into something they can specify, verify, and defend during quality control.
This guide focuses on the fundamentals behind a surface finishes chart: what “surface finish” really includes, how it is described and measured, and what the most common roughness parameters actually mean in practice.
Surface Texture Fundamentals
When people say “surface finish,” they often mean roughness. In metrology, surface finish refers to the overall texture of a surface, characterized by three components:
- Roughness: fine, closely spaced irregularities.
- Waviness: larger, more widely spaced undulations (often linked to vibration, deflection, or thermal effects).
- Lay: the direction of the predominant surface pattern (for example, turning spirals vs. grinding lines).
These distinctions matter because two surfaces can share the same average roughness value yet behave very differently in service. A surface with occasional deep valleys, sharp spikes, or a different lay direction can have the same Ra as a more uniform surface, but seal worse, wear faster, or hold lubricant differently.
From a measurement standpoint, roughness parameters come from a surface profile: deviations about a mean line calculated over defined lengths. That means the “finish number” is never purely a visual judgment; it is a computed result based on how, where, and over what distance you measure.
Surface Finish Designations and Metrology
Surface roughness parameters exist to quantify deviations in a measured surface profile. Most common parameters reference deviations from a mean line and are computed over a defined sampling length.
Two length concepts show up repeatedly in standards and inspection plans:
- Sampling length: the segment of profile used to compute many roughness statistics.
- Evaluation length (measuring length): a longer distance that may include multiple sampling lengths, often used to capture extremes and make results more representative.
Units you will see on charts
Surface roughness is typically expressed in:
- micrometers (µm)
- microinches (µin)
A useful conversion to keep on hand: 1 µm = 39.37 µin.
Why choosing only Ra can be risky
Ra is popular because it is easy to calculate and widely understood. But for many functional surfaces, Ra can be too general:
- Different topographies can yield the same Ra.
- Ra is not greatly influenced by individual spikes, so it can “hide” defects that matter for sealing or fatigue.
- Some parameters are more sensitive to spikes and outliers (Rz, Rt), while others smooth them out (Ra).
The key metrology lesson behind every surface finish chart is this:
- Same number, different surface is real.
- Selecting the wrong parameter can create engineering risk, inspection disputes, and performance failures.
Common Surface Roughness Parameters
Quick Reference
| Ra | Average absolute deviation from mean line; least sensitive to spikes |
| Rz | Average peak-to-valley extremes; high sensitivity to outliers |
| Rq (RMS) | Like Ra but weights large deviations more; medium-high spike sensitivity |
| Rt | Maximum peak-to-valley over evaluation length; very high sensitivity |
Ra (Roughness Average) — the most specified parameter
Ra is the most commonly specified surface roughness parameter globally. You will also see it called:
- Roughness Average
- Arithmetical mean roughness
- Center Line Average (CLA)
- AA
Definition: Ra is the arithmetic average of the absolute deviations from the mean line over a defined sampling length.
Why engineers like it:
- It provides a general indication of overall surface texture.
- It is relatively easy to calculate and widely understood.
Where Ra can mislead you:
- Surfaces with sharp spikes, deep pits, or different isotropy can share the same Ra.
- Ra makes no distinction between peaks and valleys and does not describe spatial structure.
- Ra is among the least sensitive parameters to peaks and valleys and is hardly affected by individual peaks or valleys.
Some sources even caution that Ra can be the least accurate value for determining surface finish quality in certain contexts.
Real example values:
- Turned steel workpiece: Ra 7.51
- After lapping with Si-C 500 grain: Ra 0.103
- After lapping with diamond 2 to 3 µm: Ra 0.009
Rz (Mean Roughness Depth / Average Maximum Height) — extremes-focused
Rz is often the second most commonly specified parameter. It focuses more on extremes than Ra.
What it measures: Rz is the average height difference between the five highest peaks and five deepest valleys within multiple equal sampling lengths.
This makes Rz:
- More detailed for functional behavior driven by peaks/valleys
- More sensitive to spikes in the profile than Ra
Why it matters in real parts:
- Rz is often critical for sealing surfaces or where peak/valley geometry impacts performance.
- In ISO 4287:1997 framing, Rz can be described via Rp (largest peak) plus Rv (largest valley) within a sampling length.
- Some practices describe Rz as an average across five consecutive trace lengths, sometimes preferred over Ra for “realism.”
Example Rz values (lapping pressure effect):
- Steel part (60 HRc), SiC 500 grit, 250 g/cm²: Rz 0.6 to 0.8
- Steel part (60 HRc), SiC 500 grit, 50 g/cm²: Rz 0.2 to 0.3
Rq (Root Mean Square Roughness / RMS) — weights large deviations more
Rq (often called RMS roughness) is the root mean square average of the profile height deviations within the evaluation length, measured from the mean line.
In practical terms:
- It is similar to Ra, but gives greater weight to larger deviations (extreme peaks and valleys).
- That weighting makes Rq more responsive than Ra when large deviations matter.
Where Rq is often used:
- Precision engineering and optical applications are frequently cited use cases.
Important note: There is real confusion in industry between Ra and RMS. They are not the same metric, and the values should not be swapped.
Selection guidance: Consider Rq when you want an “average-like” metric but cannot afford large peak/valley deviations to be washed out the way Ra can.
Rt (Total Roughness / Maximum Height of the Profile) — peak-to-valley over evaluation length
Rt is the vertical distance between the highest and lowest points of the profile within the evaluation length.
Think of Rt as the “worst-case” peak-to-valley height across the measured distance:
- Very useful in QC to ensure no extreme deviations exist
- Often the least commonly used indicator, but highly valuable as a guardrail
- Extremely sensitive to outliers (a single scratch, pit, or spike can dominate)
Mitutoyo describes Rt as the difference between the height Zp of the highest peak and depth Zv of the deepest valley within the evaluation length.
Example Rt values (same workpiece progression):
- Turned steel workpiece: Rt 31.1
- After lapping with Si-C 500 grain: Rt 1.09
- After lapping with diamond 2 to 3 µm: Rt 0.119
Specification Strategy: Ra Alone vs Multiple Parameters
When you build a surface finish callout strategy, you are choosing between simplicity and functional control.
Ra Only – Advantages
- Simple, widely understood, fast to communicate
- Often sufficient for non-critical surfaces
- Reduces inspection complexity and cost
Ra Only – Disadvantages
- Can be too general for functional surfaces
- Can hide spikes, pits, and problematic topography
- May pass inspection but fail in service
Ra Plus Additional Parameter – Advantages
- Better control of extremes that drive sealing, wear, and fatigue
- Reduces chance of “passes inspection, fails in service”
- More accurate representation of functional requirements
Ra Plus Additional Parameter – Disadvantages
- Slightly more complex inspection plan
- Requires alignment on measurement settings
- May increase inspection time and cost
Practical Selection Context: Manufacturing Reality
Even though this guide is focused on parameters and metrology, surface finish charts only help if they connect to what processes can actually achieve.
General achievable Ra ranges by finish level (commonly cited):
- As machined (standard finish): Ra 3.2 to 6.3 µm
- Smooth machined: Ra 1.6 to 3.2 µm
- Polished finish: Ra 0.4 to 1.6 µm
- Mirror polish: Ra less than 0.4 µm
And tighter specs are not “free”:
- Tighter surface specifications can increase CNC part cost by 15 to 60%
- Ra 1.6 µm can add 15 to 25% to cycle time in some cases
- Ra 0.8 µm often requires grinding or polishing and can double part cost
- Moving from Ra 6.3 µm to Ra 3.2 µm may cost almost nothing by comparison
A practical approach many shops recommend is to default to Ra 3.2 µm and tighten only where the surface has a defined job.
Frequently Asked Questions
What is the difference between surface finish and surface roughness?
Surface finish is the broader “surface texture” concept, including roughness, waviness, and lay. Surface roughness is just one component, describing fine irregularities.
Why can two parts with the same Ra perform differently?
Because Ra is an average and is relatively insensitive to individual spikes or deep valleys. Two very different topographies (spiky vs plateau-like, different lay direction) can produce the same Ra but behave differently for sealing, friction, and fatigue.
When should I specify Rz instead of Ra?
Use Rz when performance depends on peaks and valleys, especially for sealing surfaces or contact surfaces where extremes drive leakage, wear, or fit.
When is Rt useful?
Rt is valuable when you need to ensure there are no extreme defects (deep scratch, pit, burr-like spike) over the evaluation length. It is very sensitive to outliers, which is exactly why it is useful as a QC limit.
What units should I use on drawings: µm or µin?
Both are used, but most modern designs use µm. If your supply chain spans regions, include units clearly and consider adding conversion guidance (remember 1 µm = 39.37 µin).
