Smart Engineering
The Phenomenal OCR
At least five operation areas can benefit from switching from manual data entry.
“Innovation is the ability to see change as an opportunity, not a threat.” – Steve Jobs
Several years ago, I embarked on an advanced engineering initiative, seeking unknown, innovative ways of processing PCB design packages for manufacturing external to the status quo, with a primary objective to drive efficiency and productivity, eliminating redundant data entry and reducing human interactions with their associated software applications. I had thrown in the towel relying on our software vendors to provide real and robust solutions. In my opinion, it was not in their interest. As their focus was selling more licenses to increase revenue, developing and marketing advanced automated solutions did not support this cause. My mantra was to take internal ownership and venture into unfamiliar territories. This path led me to OCR (optical character recognition) and, although skeptical at first, the results of our analysis were phenomenal.

Product design specifications and requirements are provided in formats such as PDF, Word, Excel, HPGL, or even as handwritten notes on scrap paper. A close associate of mine calls it “e-paper.” The variety of methodologies used to convey the same information across the PCB industry that is often ambiguous is staggering, to say the least. We print, read, interpret (hopefully correctly) and manually enter these data into our respective software application, often multiple times by multiple organizations.

OCR is not science fiction. It is real and used by some of the largest companies in the world, processing millions of documents each day. The speed, efficiency and labor savings alone are in the hundreds of millions of dollars. Since it works for invoices, purchase orders, legal and banking documents, I thought, “Why not with fabrication specifications for PCB manufacturing?” The results of our findings were staggering.

“An advanced OCR application can extract all relevant product specification attributes into XML for import into the master database, in minutes.”
A fabrication drawing outlines all pertinent information relative to the product in question. These attributes include part number and revision, surface finish, materials or stack-up, impedance requirements, drills and tolerances, dimensions, solder mask and nomenclature type and color, IPC specification classifications, and many others organizations manually enter into quoting or engineering applications.

Using an advanced OCR application such as ABBYY Flexicapture (, you simply import the fabrication drawing, and all the relevant product specification attributes are extracted into an XML file that, in turn, imports directly into the master database, quote or engineering applications, all in just a few minutes. While I certainly recommend an engineer read and review the fabrication specification, we have eliminated the time-consuming manual administrative task and error-prone entries, and replaced them with validation only.

The reality is most of the notes on a fabrication specification drawing are in excess of 20 words, and we need only specific information to complete our task. In essence, the OCR application can be taught to identify these keywords, and, in turn, that is what is extracted. Drill tables, with up to 20 or more drill callouts with tolerances, plated or non-plated, microvia or back drills, are in turn extracted. I could go on, but the fact is anything you read on the fabrication specification can be extracted into an intelligent XML file, and from there the opportunities abound with what you do with it. Operational areas that would benefit are:

  • Quote generation
  • Pre-production planning – travelers
  • Document control – AS9102 reports
  • Quality – inspection criteria or spec review
  • Finance and accounting.
Commercially available tools automatically extract the key information from fab drawings, eliminating manual entry
Figure 1. Commercially available tools automatically extract the key information from fab drawings, eliminating manual entry.
Using an advanced OCR engine with an embedded native language processor (NLP) provides the foundation to achieve speed, accuracy and a cost-efficient operation, all of which should be a major focus for any company leader in the competitive landscape of PCB manufacturing. Providing solutions to engineering personnel that take them away from menial tasks and effecting an environment for them to innovate will only enhance the operation. I looked beyond the industry standards and was pleasantly surprised what OCR technology is capable of, and the multiple benefits it could bring to our daily operations.

If you would like more information or discussions on what this can do for your company, please feel free to contact me directly. In my next column I will elaborate on how to drive more autonomous processing of the design data packages in an environment of smart engineering.

Portrait photo of Kent Balius
Kent Balius
is senior consultant at EPIC Front-End Engineering (; Listen to his PCB Chat podcast on smart engineering at