The MIL-PRF-13830B specification is a set of rules set out in 1997 by the Departments and Agencies of the Department of US Defence, and cover the manufacture, assembly and inspection of finished optical components such as lenses, prisms, mirrors, reticles, windows and wedges.
There are a number of defects that must be inspected for under this specification, however, there are two defect types identified as surface defects - "scratches" and "digs".
There is no definition of what separates a scratch from a dig in MIL-PRF-13830B, and so there is always room for interpretation. It is widely agreed that if a defect is considerable longer than it is wide then it is a scratch, and otherwise it is a dig. Inspection systems resolve this uncertainty by defining a threshold for separating scratches and digs. Below is an example of this.
Inspection results of a 1 inch optic. The left half showing a scratch and a dig, both clearly identified and differentiated between, the right half showing an overview of the optic after inspection.
Once a scratch has been identified, it must be graded. In section 3.5.2, MIL-PRF-13830B sets out the grades in "bands" 10, 20, 40, 60, 80, where a scratch will enter the next band based on its visibility compared against a known reference set of scratches. For example if we take the above scratch, the system has reported that it is more visible (brighter) than the 20 scratch, but less visible than the 40, meaning this scratch is graded as a 40. This is because the reference scratches represent the brightest possible scratch in their grade. A given optical component will have a target it is produced to, denoted as "max scratch grade/max dig grade" e.g. 40/60, this means that if there are any scratches with a grade of 40 or above, the part will be rejected (the second number, 60 in this example, is for digs which are covered by a seperate section of the standard).
While the concept of this definition is relatively straightforward, relying solely on visibility rather than any measurement, there are a few pitfalls that must be minimised in order to ensure optical components are being inspected accurately and consistently.
The first of these pitfalls is the reference set of scratches that components are being compared to. Producing sets of scratches that are consistent with each other is a considerable technical challenge that has caused issues in the industry for a number of years. Please click here for an overview of commercially available scratch dig comparison standards.
The second pitfall is that, as with all human based inspection, this method can be extremely subjective. Inspector's eyesight, lighting conditions and time of day can all have an impact on how a given scratch is judged. In recent years this issues has been overcome with the introduction of automated inspection machines. These systems can inspect optical components with consistent lighting and judgement, removing any subjectivity and producing unambiguous data for later reference. The below graphic shows the result of a direct comparison between human inspectors and the OptiLux SD. On the left, 10 human inspectors were asked to grade the same scratch, with the results plotted out, and on the right the same scratch was graded 10 times under the OptiLux SD. This study shows how the process inherently introduces uncertainty when carried out by human inspectors, and so the objective nature and repeatability of a machine vision system is preferable.
The MIL-PRF-13830B specification provides a standardised way of grading defects and allows for manufacturers to guarantee component quality, however the inherent subjectivity in human inspection and the challenges of producing consistent reference standards mean care must be taken to ensure reliable and repeatable results are taken. The introduction of automated inspection systems has significantly improved this, reducing human error and ensuring compliance with the specification.
RedLux Scratch-Dig Comparison Reference Set