← Back to the notebook

The Fanuc Data Problem: FOCAS, MTConnect, DPRNT, and Reality

Getting data out of Fanuc machines always sounds simpler before the details show up. There are a few paths: FOCAS, MTConnect, macro output with DPRNT, maybe signals from the machine, maybe something through the OEM layer. Each one can work. None of them automatically means I have the data I actually want.

FOCAS feels closest to the source, but licensing and access matter. MTConnect can be cleaner for standard machine states, but it depends on the adapter and what the machine is actually publishing. DPRNT or macros can be useful when I need a very specific part count or program event, but then I am depending on program behavior and discipline. Hardwired signals are simple, but usually thin.

The constraint is that production data has to survive real use. Operators change programs. Maintenance changes settings. Network access is inconsistent. Some machines are old enough that the clean modern path is not available without extra hardware or licensing. Also, not every question needs the same answer. If I only need spindle running and part count, I should not design like I am building a digital twin.

What I would check first is the question. Am I trying to know if the machine is running? Count parts? Capture program number? Understand cycle time? Separate setup from production? Detect alarms? The data path should follow the question. Otherwise I end up collecting whatever is easy and then pretending it answers the real problem.

The surprise is that there may not be one plant-wide answer. One machine might be best through MTConnect. Another through FOCAS. Another through a macro. Another through a sensor. That feels messy, but it might be more honest than forcing one architecture onto machines that do not behave the same.

Notes for next time: write down what the machine can expose before promising what the dashboard will show. Keep the first collector small. Prove the signal for one machine. Check licensing before designing around FOCAS. Do not assume MTConnect means complete context. And do not look down on a simple part-count signal if it is the only thing that is actually reliable.

← Back to the notebook